Hi Mark, there are many reasons why in many cases TA can fail. Technically it is late to the party. It deals with what was and give you stats of the past. Some think that because they have stats they have probabilities. And it is not necessarily the case.

A lot of stock trading strategies I've seen (over 2,000) failed due to poor design, poor concepts, poor premises, poor assumptions, poor trading procedures and methods, poor position sizing algorithms, poor stock selection, poor code if not irrelevant procedures, poor simulation conditions, poor simulation tests, and poor long term planning.

Some time back I was analyzing someone's stock trading strategy. Right at the top of the program, he had an automated profit target of 1% with a corresponding stop loss of 5% (his program was nonetheless for sale). No need to do much math on that one, it would fail miserably whatever the rest of the code might have been.

I would say there are a couple of core reasons. Firstly, many people use an inappropriate technqiue for the specific market they are trading. No point using a 1930s paradigm used to trade industrial stocks for use in an intra-day currency trading system. 2) TA fails, because markets are efficient, so looking for cycles / periodicity etc., is simply a waste of time, 3) TA fails, because of execution methodology being employed, 4) TA fails because the methods / systems etc., are the bi-product of self-promoters and whilst a good idea looks 'good', in reality the method / system never worked very well in the first place - despite its over promotion and/or being readily adopted particularly by software and information providers alike.

There is also that TA can work.

It is the responsibility of the strategy designer to build something that can withstand the test of time. My interest is in stocks, and I do overtest strategies.

I design so they behave according to my plans and within my acceptable limits. I can understand this to be different for each person. We all have to make some compromises in an uncertain world.

I'm currently putting some cosmetic changes to my DEVX8 program and giving access from outside the program to its controlling parameters. The following 2 charts show some of those changes. Tests are with a $100k capital applied over a 20-year testing interval.

It is scaled down from what I usually present, but in line with a much smaller initial stake.

.#1

http://alphapowertrading.com/images/divers/ABT_DEVX8_P25_95_50_20_20_20_Nov23.png

.#2

http://alphapowertrading.com/images/divers/ABT_DEVX8_P25_95_50_20_20_20_0bNov23.png

...more...

Here are its trading rules. Don't buy below initial price. Buy randomly on the way up, sell on the way up. Wait for a profit. Role up the inventory as best you can. Accumulate shares over the process. That's about it.

It is purely technical, uses math, equations and almost no indicator (just a fraction of the red line at times). The buying is limited by the buying power, so, yes it will use leverage in the beginning. It will buy when there is cash in the account. As cash reserves are building over time, leverage ceases to be needed or required.

It ends up with close to 80% in cash leaving only 20% to vary with the market thereby reducing volatility and the impact of drawdowns. Two charts were presented to show the random-like behavior of entries and exits. Numbers I want to see are printed on the charts. I would describe this as a Buy & Hold variant: a Buy & weak Hold. Using only the controls, it can do more, much more.

@Alex, I can not agree with most of what you said. Especially: <...any "math" method will generate some "good" trades and some "bad", and it's interesting that on average virtually any "pure math" model generates approximately 50% of both. >

Just look at the last 2 charts provided. Good trades: 99.6%. The remaining 0.4% are still opened trades with minimal losses. In the first case 6 losing trades out of 1,488; and in the second case 5 out of 1,459 trades. Not what I would call a 50% proposition.

So you have at least one math model contradicting your “undemonstrated” claim.

Mark, I've been keen on various range-based data compression and associated techniques for quite some time about 10 years ago, but then I realised that eliminating time we throw away one of the most important component of the information available to us: the information about the market sentiment. The simplest example could be price dynamics in the fx market: simply put, if price slowly but steadily travels to a certain direction then there are many chances that this movement will persists, and if it has quickly elapsed a good amount of pips then it's more likely to revert. Of course it's overly simplified, but I think you got the point.

Another conclusion to which we may arrive if we think thoroughly about the above fact is that using price targets in any form could possibly be the worst idea ever: the market process we're trying to trade could be the same, but its amplitude could vary greatly. Therefore time becomes one of the most important parameters to the whole trade setup, and in some cases the most important.

I have thrown together a chart that is pretty sparse in explanation of why technical indicators seem to work better on block sized data. For 30 years I adapted math to be dynamic for varying market conditions. What I am doing now is adapting the data to accommodate the indicators. Which has now led to a pure pattern recognition type of trading coupled with good money management. I would encourage anyone who has not used non time based charts to try it at least. It does take some time to warm up to but when run side by side with traditional time stamped data it is more orderly. http://markbrown.com/20151124-001.png

Guy what do you consider the basic homework for building this kind of systen that you mentioned ?

glad to see that Mark is back posting in the group. I had almost given up reading the discussions as almost every discussion nowadays seems like a promotion/ad to me with little original content designed not to generate a discussion but to draw actual customers from the members..

Thank you for the suggestion, I really do not trade like that.. Because it is Mark posting I definitely will take a second and a third look and try to see better what exactly you mean and if I can use it or has anything meaningful to add..

Lol Nikolay, you took the words out of my mouth.. I just went through some posts and the whole group is one bid advertising wall.. I was starting to doubt if there were any real traders here at all.

@Adrian & Nikolay 90% is the same everywhere /all groups ..people like Guy and Mark ..are very rare ....I am sure there are many .. but you cannot see them ..

Predictability is best done without extrapolating. Therefore manipulating time and substituting it as an explanatory variable may give better accuracy, if not precision.

I rarely see people testing the limits of their trading strategies. I can understand that. Often, the strategies are inflexible, they play indicator value crossings which kind of limit one's interventions to varying indicator value and bet size.

I want to know the limits of my strategies. The following chart is pushing in that direction and has not reached its limit yet, it can go further.

.

http://alphapowertrading.com/images/divers/ABT_DEVX8_P45_96_125_60_60_60_0mNov23.png

This strategy has a chainsaw design, and in need of smoothing out some of its processes. But smoothing things out comes after an ample proof of concept thing. It is when you know that the stuff you designed does as planned that you consider bringing refinement to a methodology; enhance its positives and reduce the negative side effects. At least, I am satisfied with its controllability. When increasing parameter design settings, the strategy does respond.

...more...

Previous posts showed that you can control position sizing, number of trades and the profit spread of a trading strategy with for constraint having the capital needed to carry it out.

.

http://alphapowertrading.com/images/divers/ABT_DEVX8_P45_96_125_60_60_60_0mNov23_Stats.png

Extending the average profit spread on thousands of trades is not that easy to do. One of the methods used in the presented charts is centered around the concept of delayed gratification, delaying the first stopping time of a stochastic process. Meaning that a sell that should have been executed is randomly bumped to a future date, not taking the immediate profit and opting to wait for another day with no guarantee of a better price. In the last chart, this trade bumping to the next day occurred on 271,220 profitable and available exits.

There is a mini me version:

.

http://alphapowertrading.com/images/divers/ABT_DEVX8_P30_80_20_20_20_20_10k_0nNov25.png

One has to be courageous to trade this system...DD will huge.

@Emmanuel, you could take the mini me version... if you are that fearful of a less than 20% drawdown.

@Brian, I value your input. You always brought down to earth considerations to the table. A pity I only do stocks.

You will find a lot of stuff on my site on the evolution of the DEVX program, how it is structured, what it does and what it is aiming for. Most of it is like a Buy & weak Hold kind of thing. You are continuously turning over the accumulating inventory as price rises. Trading is technically a byproduct of the primary objectives which was to accumulate shares for the long term.

Hope you find it interesting.

@Guy, thanks for the reply. Is your strategy meant for index funds/etfs/etc? Or are you applying this to a certain type of stock or basket? I assume you attempt to eliminate as much firm specific risk as possible.

Theoretically, if you began your strategy in 2007, how would you have performed by 2009? Just curious, I see your drawdowns are low, but I'm also curious to 2 things: Will your strategy have negative accumulation (or does it stop accumulating and sell only to 0 shares), or, will you still be sitting on any accumulated shares from 2007?

Mark, if what you mean or rather your operational definition of "technical analysis" are indicators and/or indicator sets being used as anything more than simple "trigger logic" to get you into the or out of a trade, I'd honestly be shocked that you could derive any long-term consistency from the use thereof.

Without a solid defined core of Business Logic recognized before-hand, betting on the criss-crossing of lines no matter how thoroughbred your math is will ultimately land a trader at a cummulative loss.

One thing you said, Ive lived thru personally in the markets the last 16yrs., which is to say any particular market behavior your algo identifies correctly walking forward it's subject to failure once that behavior is no longer there.

One edge for me has been to play out a market tendency until its no-longer there then move on to whatever you can recognize next.

as with price defined charts one can catalog and trade patterns. toby crabel comes to mind as a master pattern trader. i discovered that timeless data also has patterns and the difference is there is no need for fuzzy logic. hard defined rules exist and either you get that pattern exactly or you don't. with time charts one has to be dynamic again i will state make the data fit the math not the math fit the data. it's like watching everyone use square wheels when round ones are now available. the idea that you can not know what a market will do is totally obsolete. in fact many time based traders i know have rule sets that clearly define scenarios before data arrives and orders placed in advance with follow up what if's. You can not arrive at this confidence level without knowing what can happen. you can not know what will happen without allot of experience, betting has nothing to do with it.

Posting another chart comparing time based daily bars with equivalent timeless range based bars. I think this has discussion has exhausted but nonetheless timeless data is here to stay. I view it as the round wheel introduced at the height of square wheel popularity. http://markbrown.com/linkedin/20151126-003.png

If you used to use charts to analyze data to execute trades then your results were inconsistent due to the variable nature of the trade volume/size data per bar. Pre-MDP3 data, this was the case regardless of whether you are using time, range or tick charts.

If you were using raw data (prior to MDP3) to analyze data to execute trades then your results were better especially if you were using trade volume/size data in your analysis. This again was the case regardless of whether you are using time, range or tick charts. At best, constant analysis is required to make sure your algorithms were optimized against the most recent data.

Currently we have a different issue with the introduction of MDP3 data. This data format now aggregates the trade data before it sends it to either the users directly or to 3rd party providers for them to distribute to their users. This aggregated data is a complete mess bundled or unbundled. First, this new formatted data, is delayed. The data is no longer provided in real time. The CME, as an example, bundles trades into groups before sending them on their way. We have documented bundles delayed as much as 3.5 seconds and 39 trades per bundle. Another issue are the registered ticks. We are seeing an average of 40% less ticks per day than before the introduction of MDP3 as well on unbundled data. This now makes tick analysis absolutely worthless.

I know some you will say that this aggregated data can be unbundled. This is true but the exchanges have not allowed the users or the 3rd party providers the specifics of the new field generators so if anyone tries to unbundle the data they end up getting different volume amounts and different tick amounts. Gone are the days of comparing data providers for the quality of data because that data is all different now. And the worst part is that the data is delayed and they aren't telling you that.

Now before some of you start throwing verbal jabs at me I want you to know that I spent well over 24 hours over the last 3 weeks with CME Support and other data engineers that have confirmed everything I just said about this crap MDP3 data.

I'm a TA guy. I love TA when it is approached in a logical manner but now it is all but destroyed. There is some fringe ways to use the current data and my group has figured that out but the data is nowhere near the accuracy and it contains none of the granularity that was there in the past.

TA will fail more now than ever. TA used to be hard but not impossible. It is still possible but far harder than it used to be.

@ William ..i am TA guy --> from “literature” and news we know that HFT and Price actions /day trading = works .. .. . today my people just finishing verification on data we receive from our brokers (4) for the last 5 years

Daily candle 0,1-1 % price/different

Hour candle 1-3% price/different

Minutes 19% different > +/-6- 8 ticks price/different

Ticks by Ticks /100 ticks candle price /time 11% price/different

Volume profile /quantification / % +/5- 25% price/different

On my opinion working with “non-time candle / fix length bar /momentum bars (no renko) ” is a real solution as long we can identify patterns ...and combine with the "real" money flow ..is a real solution.

“Are there indicators only can really make money in real accounts? ( I fallow your debate )”My opinion = NO . . . in my humble opinion the solutions is in the “mechanic of the trades” and not in signal "only" . . . .have a nice day :)

Whatever you do trading, there is math in the problem. Whatever line you put on your monitors, there is math behind it. The objective for a TA trader is to massage the incoming market data and bring it to trading decision points.

That trading be computer assisted is secondary if basically the discretionary version does not work either. If your trading methodology is based on gambling, there is no need to really test anything. But, as a consequence, don't tell me that you can control something or that your methodology can survive over long trading intervals. You will have to show me, to prove it, otherwise, for me, it is just another opinion.

The automation comes after a discretionary trading method has shown some potential and where you try to answer the question: can this work on more securities over long periods of time? That is where simulations come in. You simply want answers.

Truth is so many of these issues would be solved if regulators forced liquidity back to a single exchange and made speed equal to all. Better data, better price discovery, just better - the problem is lack of competition and therefore investment like other utilities. HFT is not exactly technology driven it is time arbitrage. But the fracturing of liquidity and the lack of a consolidated tape is causing algo traders to transact with imperfect information on both liquidity and price. That in all likelihood is where the bulk of the strategy volatility is driven from. Sadly it remains an arms race and those with the most money end up with the best data and therefore end up winning. It's not right but it's true.

@Daryl, first, you will never be able to make speed equal to all participants ever again if it ever existed. Second, regulators can not force liquidity, they never have, and most probably never will. No one can force anyone to trade, especially if it is at his/her disadvantage.

HFT is here to stay. However, I would prefer honest HFT. And at this juncture, I regrettably can not say this to be true for all HFT firms. Regulators are sleeping, if not in a coma, on this one.

@Guy .. momentum/fix size candle are not effected from the HFT . . decision is based on price and forecast as far i have seen is more easy to make .. .

My obs. --> algorithms works perfect in perfect environment... . time series analyses /forecast is not the perfect environment ( HLOC ) always in time chart algorithms has / receive "obstructions " from "events", news ,manipulations , day end ..and so .. my observations come to the conclusion that I need in the same time two different charts. ---> Ex. ..time charts 4h & 75 points candle .. . .test i made over 4 years show good performance .. algorithm i design is "simple" but very accurate (on S&P--> 65% success rate ,DD < 4% ,RMaxDD 8,89) .. this is just the beginning . . for me removing /reducing the noise from TC become addiction :)

http://dropshots.com/ispasaurel/date/2015-11-29/00:44:17 Two charts ..same story ? ..maybe yes .. but is not :)

@Aurel, interesting your aproach, as the NY session has 6.5h duration (the bulk of participation), I'm afraid 4H charts are not best suited for intra-day analysis of the SPX500 (if that was your intention). Could you do the same comparison using 2H or 1H charts for same day? Thank you Sir.

To show strategies don't fail all the time, I've added the notion of a retirement plan to the DEVX8 program just to show off. Also added minor cosmetic modifications.

The program starts with a $30k initial capital to which is added monthly contributions of $300 for a number of years (15), then starts to retrieve $150k per year (5) from the cash reserves.

This is based on the mini me version presented earlier (some were very fearful of drawdowns). You can push higher with more capital as was also presented in earlier posts. I opted to show two views: one on how it starts. The other, how it ended. The program prints on the chart the stuff I want to see.

.#1

http://alphapowertrading.com/images/divers/ABT_DEVX8_P30_80_20_20_20_20_Nov29_wRP_1.png

.#2

http://alphapowertrading.com/images/divers/ABT_DEVX8_P30_80_20_20_20_20_Nov29_wRP_2.png

ABT had a 11.55% CAGR over those 20 years, and would be valued at $267k on a $30k position.

For those few that know a Buy & Hold variant has to go through the ups and downs of the market for the simple reason that you are holding on to stocks most of the time, here is a more appealing scenario. Naturally, one has to have the initial capital to carry it out. But it does compare nicely to the mini me scenario presented earlier.

.

http://alphapowertrading.com/images/divers/ABT_DEVX8_P40_95_95_30_30_30_Nov29_RP.png

This way, I would think that all bases would be covered, those that know there will be drawdowns and that they are inevitable and will grow with the portfolio size. And those that play so small they keep their bet size in check in order to reduce their drawdowns forgetting the need to increase the number of trades in the process.

The retirement plan was designed as an option to be carried out or not. It is a matter of choice, but I know some would like to take some off the table at times. Such things need to be planned out as much as anything else.

@Vitor this is for the 4h strategy .. and has to be 100% TA strategy ..don't take in account any session .. if you want to trade by hand .. go lower .. to 1H .. and 50points ..

@Guy ..maybe yes or maybe not.. i am waiting his answer .. i think i know what is using . .

I don't think I am being understood here, and this subject is just too much to explain here now, what I am saying is that I have built hundreds of systems, I do have 3 that work, not all are 50/50, which is chance, you could have a system that for one year it has 60% wins and 40% losses, the next year it could have 40% wins and 60% losses, so if the win / loss size ratio is 1 to 1, you would have a 50/50 system, but if you tested or ran it for a year you would think it was working because you had 60% wins.

Also one of the worst things you can do is mess around with the trade size, if you are winning then you are happy and push the size way up, that's when it has more chance of loosing so you end up winning small and loosing big, then you bring it down and start winning small again, it is a serious mistake, the win/loss size ratio, and win/loss percentage ratio are equally important, you have to increase in a small controlled way, or if you want to push the size up you have to leave it up win or loose, and make sure you have enough funds to do that, greed is normally the biggest killer in trading,

@Robert .. this group is about TA.. 90%-100% where is the grid in case you have setup your system ..with a particular win/loss ratio ? .. and you don't touch as long stay in the back test conditions/ parameters ?--> losing big. . . .come from your setup conditions ...!!? ! . .i believe ...

I just looked back at what I wrote, I thought maybe it wasn't clear, but I think it is very clear, at no point did I say it is just a 50/50 or chance game, I was pointing out that it is difficult to make a system that is not 50/50, and also that it is equally difficult to test and make sure it is not a 50/50 system, I never said it can't be done

@Mark .. how many orders you have in average /day ..

I try on FX this system .. problem come from slippage and spreed /change (no news ) in the time of the trade ..witch reach my SL instantly ..in no time ..back test performance are good with Bid/Ask Chart ..when it come to real/live ...=fail ..

My trades are a calm 3 trades per week to 2 trades per month.

@thanks .. 10 points in EURAUD /Multicharts = 1 pip .. this is scalping .. now I understand 10 points in S&P is different story :)

I built software to check the brokers, the results were amazing, I had a meeting with my broker, informed them that I would be making money, and that they should manage my, and the people that trade with me's, risk profile accordingly, now I don't have problems, but the manipulation is real and it is out there, I am not saying all brokers are bad, and I would love to list all the bad one's, I have been instructed to close accounts when I have addressed these issues, remember the smaller the trade the easier it is to manipulate, also the larger the cost becomes, if your spread is 3 pips and you move 6 pips 50% is going to the broker, if you take 1000 pips it is only about 0.3% of what you make, so the smaller you trade the bigger the cost, and the more susceptible you are to manipulation

@Robert AT become very complicate- - > from arbitrage I cannot get my money from broker (45K) forex com close the account of my friend (because he made 5% /day !!! .. and they just send email --(:) . . scalping / was possible now spreed is so high in the day(no news ) that make impossible trading. < 5pips ..my advice is try to use your signal on point charts > 6P-10P minimum .. this is what we need to do

@Guy - I agree that Mark's range bars are more sensitive to directional changes, especially now knowing he is using then to trigger trades from 3 trades a week to 2 trades per month. Now that is comfortable trading. Nice!

But to clarify, Constant Volume Bars are activity neutral, big volume neutral and large block neutral if used correctly. The advantage of using them is to see price action in its purest form by neutering the artificial influence of trying to figure out and interpret what large traders are trying to achieve. I've seen large blocks completely cancel out each other in the market. If you are concentrating on their actions you miss what their interaction is creating . . . pure price flow.

Each method has its own attributes. It's like a American made race car is better at the Indy races and the Nissan 200SX is a drifter's choice. Crappy analogy but I think you get my point.

@Aurel - FOREX spread trading is fixed (corrupt) . . . period!

@William, sorry to disagree. But constant volume bars are not activity neutral, on the contrary, they depend on trading activity.

With stocks as example. Using a constant 20k shares bar delimiter. We record the open of the first trade and close of the last one in the series no matter the time interval or the number of trades required to get there.

This would correspond to your constant volume bar definition.

I would need 200 trades of 100 shares, down to one trade of 20,000 shares to make one bar. In either case the price might not move. If a 100k trade goes by, I'll get 5 horizontal bars, more likely a dashed line, since the price would be the same for all 5 bars. Two successive 100k blocks would just extend the line further. It won't tell you if the pressure was effectively on the buy or sell side. It will be when you will see a staircase price movement that it should become interesting. But, notice that is what Mark's method is already doing, detecting price movement.

i look at volume the same as a natural spring on my property. the water runs all year everyday there is 125,000 gallons of drinkable water flowing somewhere. i notice sometimes it's flowing a bit more and sometimes a bit less but the volume never crosses my mind as along as i have confidence that i can get water when i want it. same goes for whatever trade size i want to get off at anytime by going to market. doesn't matter what the volume is as long as it's massive.

what matters most to me is where that price is going and how it gets there, thus the pattern recognition. range bars provide more stable patterns than time based bars. i have cataloged all the patterns but have found one that is exactly as i used to trade on big sp's in the pit. so like home cooking i am sticking with my comfort food's and just improving the ingredients, rather than chase the unicorn. but make no mistake i know the unicorn exist and have the wounds to prove it.

@Mark - I know you understand this but for those confused. Volume can be used in multiple ways. As you have described it, as having enough Volume at any given period of time to confidently execute your trade at the moment you choose. Another way is how I use Volume which is a specific trading increment like; time, ticks or range.

I really like how you use range bars. I find them a bit irritating during periods of consolidation but you have seemingly mastered knowing the point increment for the symbols you trade to minimize that or at least make it more manageable.

@William, lol, you must have invented the constant volume bars in the 70's if not earlier like in the 80's, the 1880's that is.

@Mark, you, Aurel, or Philip, use generic data smoothing techniques. We can aggregate price data by variable and constant ticks (trades), variable and constant time intervals, variable and constant volume bars, variable and constant price movements. They are all acceptable methods, kind of variants to each other. We can even use combinations of these to paint our multiple monitors.

With over 200 years of US market history, none of those smoothing methods has proven to be superior to the others. They each have advantages and disadvantages, strengths and weaknesses. If anyone of them really had shown long term longevity, it would by now have been adopted by all trades, wouldn't it? Note, they all suffer from market lag!

One should simply use what he feels most comfortable with, and design his trading logic around it.

@Guy . .

In few words, if you use the (x) ticks view in combination with the classic intraday time-based view, you can enrich your chart analysis with the following advantages, providing not only different, but also more accurate information:

1.Clearer analysis

2.Not time-based

3.Confirmation of trend-line breakouts

4.Clearer signs when to exit the market

5.Correlation between the volume and the price development

..my opinion :)

There are as many nuanced methods of trading as there are traders. There is no right or wrong way to trade. There is only a profit-making trade or a loss-making trade.

http://dropshots.com/ispasaurel/date/2015-12-04/23:43:33

T/A has become a bit of a religion with the "faithful" preaching to the ignorant or the "unfaithful". My approach is to be guided by T/A and fundamentals but be a slave to neither.

@William, you say: .

Then, you now have a new invention to make: a market adaptable (meaning variable) constant volume bar.

This way, a 1,000,000 share trade might show something else than 50 dashes in a straight line across the screen based on a 20k constant volume bar.

Constant volume bars might be the only indicator out there that will most effectively distort everything else one might want to put on the screen. Any indicators, time, range or grid based, would just decay and flatten out over those 50 dashes lookback period, either converging to zero or to a moving average of some kind. Providing nothing additional to help the decision process to trigger a trade, up or down. This sorts of limit one's trading decisions to only one indicator: the cvb. This is too restrictive.

So, sorry, if I am not an adherent to such a thing, never have, and most probably, never will.

@Guy - Please provide a single example of a variable constant. Variable - a variable is quantity that may change within the context of a mathematical problem or experiment. Constant - a constant is a number on its own, or sometimes a letter to stand for a fixed number.

What you are asking isn't a realistic problem. If you are trading a stock that will occasionally have a trade of 1 million shares traded against it, you would never, ever, trade using a 20,000 Constant Volume Bar Chart. That chart speed is far too fast. I pointed this specific point out earlier. This goes to "knowing the environment you are trading inside". In comparison, that would be like saying you would find trading a single 500 E-Mini ES contract lot using a 1 second chart acceptable.

You are purposely arguing about something you are completely unfamiliar with. Why? You are condemning something you know very little about? Why?

@Guy - You say your example limits one's trading decisions to only one indicator and that is too restrictive. Allow me you give you a simple mathematics lesson.

If you are using two indicators on a chart and each indicator has an accuracy rating of 80% then the combined accuracy of the two indicators, used together will have an accuracy rating of 64%.

.80 x .80 = .64

If you add a third indicator on that chart and it has an accuracy rating of 80% then the combined accuracy of the three indicators, used together will have an accuracy rating of 51.2%.

.64 x .80 = .512

If you add a fourth indicator on that chart and it has an accuracy rating of 80% then the combined accuracy of the four indicators, used together will have an accuracy rating of 40.96%.

.512 x .80 = .4096

Do you see a pattern? Some traders think that the more indicators they use the better. They are simply wrong!

just a reminder to keep our debates civil we are all here to share and discover.

@William, you are confusing statistics and probabilities.

If you could find 4 indicators with an 80% rating, you could change the nature of the game, and to boot, apply a Kelly criterion to the betting system in order to optimize performance further. Even if you did that, you might still blow up.

Because the statistic you would have gathered is simply that: a statistic of what was observed, and not necessarily the probability of what will be.

Even the premise on which you want to set your observation is “technically” flawed. Most indicators are not totally independent. A requirement that would be needed for your simplistic example.

Most short term traders use multiple indicators for confirmation, a consensus on what they see, thereby, weighing in the sum of “technical” evidence before them. If a trader sees the SMA, MACD, ADX, RSI, and STOC rising, he/she might be more inclined to take a long position which they might not take if only looking at a flat line.

@William, still the aggressive verbal attitude, not conducive to sharing, is it?

When someone uses an impossible example to make their point, whatever logic they might want to derive becomes irrelevant since based on impossible premises. I object to this kind of flawed “logic”, whatever its source.

There are lots of examples, in trading, where you can use constant, momentarily fix or totally variable parameters. It is the developer's choice to use what he/she might find useful.

Maybe you haven't encountered these since operating on only a single fixed volume indicator rendering de facto all other possible indicators technically irrelevant.

We had this discussion over 2 years ago which led to the same conclusion. I opted to simply ignore what you said. You were trading visually, not automated, and in an automated trading forum, it almost rendered your arguments close to irrelevant.

Things have not changed. So, I will try to go silent on you, again, if I can.

Technical analysis is a simplified version of statistical model fitting applied to time series data. Often times, people simply use the wrong model (ie, assumptions are violated) or they use a good model on bad data. Here is what I mean...

People may have a "working" trading model (it makes a profit for a while). But then when the statistical characteristics of the underlying data (i.e., the stock prices) changes, the model is no longer a good fit. It looses money. This happens when traders don't know the source of their profits. In this case their trading model worked when the technical indicators were actually predictive of the stock they were trading. This changes over time. Alot. And the model the trader was using simply didn't change as the underlying data (stock price data) changed.

You wouldn't use a trend following strategy with a mean-reverting stock. But do you KNOW whether the stock price series is trending or mean reverting? Have you VERIFIED your model assumptions?

The vast majority of traders have no clue whether their trading model is appropriate for what they're trading. A trading strategy may work for a while (because its a good fit, statistically). Then mysteriously stop working (because the price data changed properties). Most traders don't even know whether the technical indicators they're using help-hurt-or have no predictive value for what they're trading. This is the source of the vast majority of confusion and disappointment.

@Dwayne, I like very much your opinion.

@Dwayne .. I have friend .. is trading for the last 30y .. no indicators .. charts & price only :) .. but the charts are 200 ticks / 4h / 1D ..

I think true discretionary trading skill is rare. Most people who think they have it do not. Most people would be better off trading with a system.

Gentlemen. As someone that has a fully automated day trading strategy system that can trade all of the Index Futures (ES, NQ, YM, TF, EMD, VX, FDAX, NK),

I have some general observations and questions. Some of the observations may be in conflict with some of the previous statements in this discussion.

Please consider my observations just as a point of view of what works for me.

Observations on automated strategy development:

Time has been mentioned as a factor of the strategies performance into the future. It was mentioned about different style charts being used such as seconds, ticks, hourly,

daily , what ever. Time does have a factor in the evaluation of a strategies performance, but the improper use of time can start a process of trade signal quality decay of scale

into the future. Often when a developer begins the process he/she decides on the instrument to trade , then chooses a style of chart and then the back test history period.

This is the single most important time using simple math and logic to consider what the probabilities may be of the strategy working well past the development period.

Case in point.

Day trading the E mini (ES)

Using a 15 minute day session chart

1. year back test history.

With the above setup the strategy will be trained and evaluated on only 68,000 bars. Even if we used the 24 hour continuous 15 minute bar chart

we only have 233,000 bars for training and evaluation.

My question is simple. What is the starting point for the number of bars? Is 500,000 or several million?

correction to the above:

1. year back test history should have been 10 year back test history

I think Mark actually answers the problem in the very first paragraph at "Only to get back in sync". Technical models become more and less efficient over time as different fundamental cycles occur. This is something I observed in my ML models in the early 2k's and actually began to monitor. When a model became extremely efficient, when it was doing very well I would actually begin to reduce leverage. When the efficiency of the model began to taper off I would either pull something different off the digital shelf or create something new. You don't here much about tracking Model Efficiency but I think it's something algo's should track. We waste a lot of time "creating" when we likely already have tools that work. - JP

Stay with a demo account old boy

Expansion and contraction coupled with the use of percentages has historically been useful to me in taming changing market data. As has been mentioned percentages level out the data and provide have consistent analysis.

Hi Aurel,

I agree.. My trading method allows about 2-3 set ups in CL/day and 3-4 set ups in ES/week.. not 50-200/day.. I suppose you can take the same concepts down to a smaller time frame but I've never tested that.. my method is very difficult to automate and I have tried.. I think if I can automate my method, the execution can be even better.. Once the context is determined, I think a computer will be able to execute much better than a live person.. I think it can also factor in order flow and etc with more efficiency..

Context, levels, PA & Order flow at the levels is all I use and in that order.. My method uses no indicators at all which is also not good for automation..

There's also a time element coupled with the order flow but context is the key determinant..

@Jack .. "around you " are "people" with high skills in algorithms making and designing and of course trading .. trust me :) .. anything in trading can be coded .. the only issue is that the rules to be clarify .. clear .. and not based on felling .. or you know . . . I give you one advice ..piece of paper ..and before entry "cut paste "the charts pictures" one by one ... rules one by one on paper .. and this you have to do for the next 10 trades you made .. and you will be surprise .. will not be the same . . than I am sure you know same one witch can fallow/reverse your trade .. simple :) .. today i done the same .. first by hand second by PC .. have a nice evening "you all" . .

Now look at the same situation for CL at the European opening.. price consolidated underneath a level.. The assumption is price will go lower but it never did at the opening.. this indicate indecision.. so price is waiting for a catalyst.. it has not tested higher or lower.. the assumption is to go lower and you will just take that trade and wait.. the probability is lower... as time increase with consolidation underneath, that probability is increased so you can add to the position.. You scale some when price goes your way to remove some risk and hold the runners.. If you get stopped out, its fine because that set up was good..

Aurel,

I'm sure there are many skilled programmers and experienced traders here, no question.. everyone's method is different. My method has been successfully tested for probably 5 years now.. I've done all the charting and journaling.. It can be 100% defined with full rationale.. its not black and white like most methods but rather a way to read the market to determine probabilities.. more contextual but very learnable and simple once learned.. I've actually explained it to many top programmers including some well respected institutional programmers and they did not get far though they understood..

The reason they did not get far was not because they did not try.. We worked closely together and the products they showed me did not duplicate even the levels accurately.. and determining the levels is key... the decision zones.. Once the decision zones are determined, then the rules go into operation.. My method reads how market decisions are determined real time.. probabilities of range, stop outs and etc... Then finally when it applies, order flow and how this order flow is being received... Its the most comprehensive method I know.. it captures all the relevant information and ignores the less relevant information..

This is not necessary for successful coding but when the market is waiting on news, that is also factored in... a big part of the method is determining what price should do and what its failing to do... Its a live read...

valuation is a good eexample. How much valuation is a company worth? Multiples of earnings or sales is determined by future PROJECTIONS and those projections are subject to change.

Fundamentals is like what a woman tells you and what is supposed to happen. The technicals is how she is behaving.. both should be considered but the short term trade has to be off the technicals.. the fundamentals only provide context though the context should considered.

ultimately, train context has more to do with bigger time frames than fundamentals in the short term trade . The bigger time frames are reference points wit how someone has behaved in the past te last time the were in the same situation... the market will either balance or seek balance from imbalance and as traders, we aim to recognize thos opportunities.

price also test levels. This is because nobody really knows where the market is headed including the market itself. It tests up and down... price will typically move from level to level inside a balance and from balance to balance until a new balance is established. The fundamentals might hel you recognize the driving force but ultimately the technicals must confirm.

indecision times, the market wil respect smaller time frame levels and framework. .. decision zones where bigger decisions are made, the smaller framework are violated and becomes noise.. but the principles all remains the same..

probabilities can be extracted from technicals but just llike good Quant models, it must be multifaceted. It's easier with Quant models than technicals wit coding..

The context basically tells you which side is preferred for greater probabilities and range and hence better R/R.... Clearly there are trades on both sides at all times but your trade selections matter... Then its a matter of position sizing to enhance the math.. If you can determine how you will size your positions optimally, your results will grow exponentially.. many traders also neglect the math of position sizing unless they're big outfits... The trade selections and position sizing is where the math is most powerful..

Check email !

If code can determine context, the rest of it is easily programmed.. and the results can be incredible.. A computer can probably assess probabilities much more precisely than a human trader.. A human trader just ball parks position sizing and timing.... Look at CL right now.. Its consolidating but the greater probability is to the short side... but where do you place that stop? The stop probabilities does not offer any statistical edge.. hence a no trade or very small position.. not the way to trade if no positions has been taken.. There are lots of various rules to add into various contexts.. but the context must be determined first..

The stop probability and the range must be determined for R/R... CL right now is a short but the stop range does not offer a good trade...

What might be a good idea is have someone program the execution aspect to be used together with a live trader.. That's another option..

Does anyone recall what fundamentalists said about the tech sector in March 2000 and what they were saying about oil and energy in 2006? All based on fundamental projections... Fundamental analysis.... It appears those fundamentals were "revised" a few times since.... The technicals clues is in long before that..... quant models warned us long before that but couldn't get the timing right.... Its the T/A that clued others on... The T/A along with the fundamentals in the background as context... clued others in... then the T/A further clued others in with how to trade what's happening... T/A has many aspects... some only look at MA or S/R or patterns like H&S and etc.. there's much more to it than just that... CONTEXT IS KING...

With the shorts trade on with CL, there's no point of even watching it... you just leave your stop and targets in place and might as well turn off the computer... Let the math do the work.... lots of people are trying to trade to the long side here with small positions and getting stopped out over and over again.... 35.40 has some slight support but likely a little speed bump along the way... Time to leave the house.....

That last move down were sitting orders getting hit... you can see this in the order flow... often when you get sitting orders getting hit, that's how you get these spikes...

Sitting orders are typically stop losses or sometimes trade signals.. in this case some BO signal or stops... nobody knows for sure but this was a decision zone for someone substantial...

It appears to be some kind of absorption... perhaps absorption of stops to accumulate.. These things can randomly happen.. no way to predict it but one can recognize it.. Sometimes it can go in your favor and sometimes against you... its all factored into the math.... the thing with the market is nobody has any control over it.. you just play the math and stay objective... focus only on the math... no need to try to predict every aspect of it or try to make sense of every aspect of it... unpredictability is all part of it... Sometimes these surprises work in your direction and other times, they take out your trade... no biggie... When you see it, you can either close out your position or hold for your thesis... I typically just hold for my thesis... who knows, there could be more surprises around the corner against that position/order flow... The market is a bunch of entities trading against each other... playing water games in the ocean... gotta go...

Aurel,

My strategy is already fully developed and has not changed for years. It does not require fundamentals at all.. I don't use it in my trading other than high impact news items.. my trads aim for 30-50 ticks on the average runner wit some runners aiming for 100-200 at certain locations if the set up calls for it.. it all varies.. it's 100% based on TA with the execution that has some Quant qualities to it.. it's very reliable for me...

Many find its hard to believe. I don't mean you but I have had some get very arrogant with me when I told them my method has not changed because it reads the market and is self adjusting.. some high level Quant people in the business finds that hard to believe often from their own frustrations. This method works and its 100% T/A.. I gave a small example real time calling the market actions ahead of time.. I have done it's for others in the past many times.. with TA, the market can be read... the bigger time frames for assumption on the shorter time frames and te shorter time frames confirm the bigger moves.. very doable.

contrary to what many believe, one market method does not always translate to other markets because of differing variables. This method I demonstrated will not work wit currencies but almost all energies. Gold will also have some other variables. My method with ES has some similarities but order flow means nothing there.. ES is typically a better reversion back to means vehicle unless in times of volatility. The key variables have to be considered. A good method works on all markets is a myth....

I can demonstrate again next week . Have a good weekend.

the best markets are wher order flow reveals an edge. CL is one of the most tradable. You want the volume to have meaning but not so light to whip around like silver... currencies and indexes has lots of players where reading price is almost impossible. You can trade it's characteristics but not reading price.. the best instruments are where you CAN.. that's a big tip..

a 30 to 200 tick move in CL has nothing to do with fundamentals. Often jus market rotation. Even market noise in the bigger context.. BUT there is a reason it rotates that way.. just the market going about its business and fundamentals is not the reason for those ranges most of the time.. just technicals.

I've had a guy with an impressive resume with electronic trading tell me its impossible to have a method that works in all market conditions because every method only has a shelf life and some very short shelf lives... This guy tried to pull rank with his resume and insist its impossible.. This is a dead giveaway that this guy simply does not know how to trade..

Most methods that people come up with is dependent on market conditions.. meaning, its either a method that is designed to work for trend conditions or reversions back to the means and etc... Trend methods only work on trends and reversion methods only work on rotational markets.. Most methods does cannot estimate in advance the type of market conditions and that's the key.. My method is not a method at all in reality.. its a way to read the market and its up to you to create the trade set up..

Meaning you can read market probabilities real time and assess the probabilities of your stops and range.. and its up to you to decide if you can create a sensible trade selection with good R/R.... My method is in reality not a method at all... its a way to read the market... and then its your own trade selection within that read that is "the method..." Within each day and week, the market will present you with many opportunities.. its up to you to make good choices to capitalize on those opportunities.. that means you must have good judgement and the discipline to wait for the right R/R profile.. Everyday, there is typically 1-3 set ups on CL but they are not all equal in their R/R profile.. a big part of the success is not only learning to read the market but having the discipline and judgement to wait for the right R/R..

Some offer lots of potential range with high probabilities of stop outs.. some offer low probabilities of stop outs prior to 1'st scale but limited range potential.. and everything in between.. The trader must make those types of choices.. A computer can also make those types of decisions but the variables will become exponential in potential choices and evaluations if you know what I mean... What is easy for the human mind to process can be very complicated to code because everything has to be defined and then the variables considered.. This will not just be a programmer but will require a team of programmers to create..

What if you have a potential trade that can offer you about 100 ticks on your runner but since you missed or get stopped out on your initial trade and the market ran and your stop range if you still want this trade has to be placed almost 40-50 ticks away, will you still take that trade? Most will say no but you also have to consider the probabilities of your first scale.. suppose the probabilities of your first and 2'nd scale is very high but your stop is 50 ticks away with a potential range of 100 ticks, is that still a good trade? This is a decision traders face from time to time especially when the market makes major moves.. big news releases at key locations.. NOW, try to code that decision.. IN ADDITION to market evaluations.. Something even as simple as this can be very difficult to code because the variables quickly become exponential.. all interconnected in the code.. what the human mind can segregate, the code must correlate..

if this is done right, the math is statistically impenetrable. It will literally require a statistical edge to be an outlier over very large sample sizes.. meaning if a dice has a weight in it to roll 6 six out of 10 times, ove a sample size of 100 roles, 6 only appeared 40 times.. what are he odds? Almost impossible.

suppose you optimize the math to be 1000 rolls, the math only gets stronger.. even 10 rolls will wil allow the statistics to manifest.. NOW, suppose you will cherry pick 3 out of the 10 rolls to guess which one will be your 6... has you math just been enhanced or neutralized some?

clearly your edge has just been diluted some from cherry picking. Rather you take them all... you just position size accordingly. . THAT'S THE SOLUTION

I hope that makes sense to some... to math and quants, it should make perfect sense.. it should also make sense to experienced traders..

smart traders learn to be self proficient - beyond being able to trade they can build their own computers, align satellite data dishes, prepared for electrical failure/spikes, prepared to trade manually in case of automation failure, have a plan for catastrophic failures/nature but of all most important program and test their ideas independently from outside help to prove concepts and ideas actually have an edge.

it takes a long time to learn to program your own ideas but any trader who does not start that path will likely fail in the competitive markets of today.

Context and multiple time frames.. the code must factor and correlate those elements just as a start.. those are the probabilities... the challenge is not because the method is whims and emotions and lack definition. If that was the case my method will fail daily and not test teal time successfully for years..

the market sucks in your own biases.. this is because the market always offers trades on both sides of the market at all times.. if want to see a long,the market will give plenty of reasons and the same for the short.. it's like one of those pictures of the old and young lady.. you must have a real objective rationale or you won't be able to trade. Why does ego hinder trading ability? Because it takes away objectivity. .. if there is no objective rationale, I would not test successfully real time in any market conditions for years.. there is no reason the back testing will show different results... especially if it's within the last several years while it's been testing real time...

over 5 years of testing. . The method has only been refined and enhanced since then but the market read has worked for at least 5 years..

the math has been optimized.. I can redesign it for a smaller account but in doing so the most powerful aspect of the math.. the position sizing wil be removed.. even then, the account size will need to be able to take every set up and not cherry pick when the dice will manifest it's statistical probabilities. .. I can do that but I would be doing something foolish with this method and with both eye open.. I believe it can be done but not motivating given what it can do if done the right way...

Jack = anyone who has traded support resistance, value areas and point of control understand the dynamics of flexible order entry with the knowledge of high percentage market pattern recognition.

The challenge for you now is either labor away and enjoy manually trading your setup. Or start to understand that you will never go to the next level unless you are capable of facing the reality of Clearly Defining your trading without involving mystical mind reading.

You have never run across the level of programmers who are here on this forum. All of them love the challenge of what they do and it is not about the money with most on this forum. This forum is about discovery and opportunity to build something large than yourself alone are capable of.

I learn something here everyday myself, what I really know is to be honest with myself and my logic will prove worthy via testing or not worthy at all. Computers are brutally honest when it comes to testing ideas.

Meaning on each set up with anywhere between 3-9 entries/attempts, evey single entrentry got stopped without 1 scale... and this happened 3 times in a row or with 9 to 27 trades on set ups with your best R/R set ups.. that's a black swan.. the account size assumes this wit a max loss of 15%...

my method has lots of set ups but it's the contecontext that defines it.. the market is always operating on multiple time frames..

CL this morning after inventory report consolidated and came out to the long side.. that was a long set up but it's not likely to go far.. so you have a choice.. you can go long or wait for another short above.. I typically don't take those longs but it's a set up.. however, if I was holding positions to the short side looking for more, I might consider taking some off & etc.. those types of scenarios.

Mark,

Sorry, I need to make a correction.. I have not discussed this with others for some time.. Its not 3 best R/R scenarios that will bring the 15% max DD but rather the 3 best R/R scenario attempted 3 times in 1 day that compose the 5% daily max DD rather than the .5% max DD.. and all the trades that attempted those 3 trade scenarios all got stopped.. Then, this happens 3 days in a row or within close proximity.. Its this scenario that is the statistical anomaly and the black swan..

What is likely to happen is that best R/R scenarios do not show up 3 times in one day... but typically 1 time in a month on average.. and they do not show up 3 days in a row.. BUT this is hypothetically possible in times of high volatility.. my method is enhanced by volatility because it can capitalize on both balance and imbalance... while most methods depend on one or the other... Meaning it has many buffers for black swan scenarios from the execution to how position sizing is built.. basically for this to happen, it would basically mean I have no edge.. If my edge is real, that should never happen according to the statistics.. If your edge was .55, it doesn't sound like much.. but you can build on that .55 math to enhance your math to be much better than .55 if you use that .55 to buy a better R/R... the probabilities of the first scale is typically around 70%.. will explain more in evening.. picking up daughter from BB practice..

Aurel,

I'm not going to spell everything out but the concepts I shared should allow many to build their own system with some study and research.. My own methodology was built from concepts... the types of concepts I shared on this thread.. it was not built on someone giving me a step by step manual.. and even if they did, I may not trade the same way they do... my mind may not filter information the way they do.. Its the bigger concepts that traders need to focus on... and TRUST the math.... T/A works.. manual trading even from professionals use TA..

a

@@@ Jack .. please dont spend more any time

I don’t want to say but ..maybe your way of trading is profitable .. definitely has nothing to do with what is algorithmic trading .. Algorithmic trading full automated 100% or partial 90% , also called algo trading and blackbox trading, encompasses trading systems that are heavily reliant on complex mathematical formulas and not with clouds or MA or RSI .all are inside the strategy ..,

Just stop .. you just embarrassed yourself in front of professional

This is my LAST .. enough .. sorry

Aurel,

I just embarrassed myself by discussing the merits of extracting math from TA in this discussion title? How many times and posts did I say I don't use indicators? did I mention MA or RSI which are indicators? Who is embarrassing themselves? Indicators are tools of the program trader that aims to incorporate TA... Blackbox is a fancy word for algo trading..... Isn't the discussion about failures of TA among algo traders? Just ignore my posts..

If this is a discussion about how to program algo trading, then I have embarrassed myself imagining I have anything of value to share... but this is a discussion about the merits of TA... I don't think I'm being outclassed here in that subject at all....

Needless to say, the market at the opening favored the shorts today given the combination of the context and what it failed to do... It doesn't mean a trader will capitalize knowing this along considering the possibility of getting stopped out before 1'st or 2'nd scale... BUT that info will put you on the right side of the trade... you wouldn't be looking for longs at the opening... The market is looking for balance at the opening and probabilities favor the short side and today offered confirmation with what price failed to do at the opening..

You recognize which side of the market you should be trading on and then recognize probabilities of 1'st scale and where price should not go for your thesis to play out and its game over.. You have all the edge you need.. nothing else is necessary in terms of market edge... the rest is trade management..

If you want to take your trading to the next level, the answer is not to automate it or you will never succeed but to leverage POSITION SIZING.. That's how you take your trading math to the next level.... Computer trading can add value only if it can enhance on the math.... and theoretically it should be able to if the metrics can be bottled...

Leaving the house now...

@Aurel, many posts back, yesterday, you said: <... i just finish the random trading on excel. >

I would like to hear more about it. Can you elaborate?

36.25 would be the first scale and target should be top of that congestion... 36.40.... A trade for 2 cars... with a bigger account maybe 4 or 8 where half comes off at 15 ticks... Let's see if I can get 4 for 4. on this thread..

All in advance.. Now that price is making an attempt up, where should price not go for my micro thesis to play out within my macro context? I have many places I can place that stop.. 36.03 gives me a 7 tick stop range.. with 2 contracts, I'm risking 14 ticks to attempt a 45 tick R/R potential... If I get stopped out, so what? Future trades will cover it...

Since my stop i 7 ticks, I can potentially remove all risk by taking my first scale at 7 ticks..... In this scenario, if I would have just kept the 1'st scale at 15 ticks and just got stopped... but in other scenarios, where the stop is narrower, I would take off the 1'st scale at the stop range to remove risk...... PRice is likely to go lower to test that lower level... not going to short into this.... possibly a long lower down..

There's another long about the same quality as the one I described... this one is better because it failed to break lower... another turtle soup scenario... A tight stop.... and potential range because volume failed to take out a level... You first scale probabilities in relation to the stop range is extremely high... and range potential good.... If I took 2 cars or 3 cars on this trade, I would take 1'st scale off around stop range.. if price is making another attempt to test lower... On my screen, there is also a micro chart with order flow... these micro decisions are made on the micro charts but the ideas all remain the same..

I have just given a small sample of trading with context and reading price action real time in a real scenario along with some simple PA concepts.. and putting it together.. These are all simple and time tested concepts I've shared... should all be easily understood from traders that have been active in the market for some time...

For anyone that has studied the market and PA at all, they should now be able to see how the math is extracted and how trading decisions are made within this method... I gave a detailed though by no means comprehensive description... but it is an overview.... A generous description... I'm sure some of these principles will help many read the market better.. at least not lose as often...

Time is an element, and if you have unlimited capital, you can ignore it as a factor. More importantly, and to the point: TA is a strategy looking for patterns. If a pattern becomes overly apparent, it's going to be gamed. That is arbitrage theory at its best.

For example, if one finds a perfectly sinusoidal pattern, someone else will eventually discover it and execute a competitive trading strategy to "flatten" the pattern.

One more important point about TA is that too many, and too few parameters will provide false results if the model doesn't fit the current regime. In other words, over-fitting and/or using the wrong parameters to a forecast will fail. The first one to come up with a 100% correct forecast for the next tick will be awarded a Nobel Prize. The one who comes up with a 100% correct forecast for the next 10 ticks will be richer than Bill Gates, Warren Buffet and the next 10 billionaires in the world. I'll fade that bet (my TA on that one shows 99.9999% probability :) )

vidis,

There's no such thig as 100% anything in the market but greater than 50% is doable.

My method don't use patterns at all.. unless you consider balance and imbalance a pattern. Suppose you have 50% chance of getting the next 10 ticks and 30% chance of the next 30 ticks and 20% chance of 45 ticks, do you think those stats are achievable? If so, the math has just been extracted.. in combination, even mor powerful.

Sorry for typos, on phone and not home.

Vidis you can come up with a 98% correct forecast up to 48 hours in advance, the problem is that it takes a lot of computational work to get there. Furthermore, once you make your trade and make your money you are back to square 0 and have to start the computational process again. This is not the kind of life I want to live, so why try to do it this way?

Should add that trying to do this process, as I described above, at tick level is moot because you simply do not have the time to do the calculations needed.

Oscar,

Lol... there's no 100% anything in the market and that includes 98% even in the next 10 minutes.. market probabilities are constantly changing... the only 98% certainty you will have in the next 48 hours is if the market is closed in the next 48 hours..

Oscar just curious no other reason are you talking about Counting Elliot Wave from the tick level? or something else, I once did a project for a large trader who did just that and it was computationally intense.

often, people's trading methodologies are like looking for patterns in the stars... they see what they want to see.. we can all form pictures from the stars in the sky just as we can all validate any pattern from the wiggles on our screens. Elliott waves are all over te market.. people just get the length of the legs wrong.. or pick the wrong legs all together...... but the market has lots of Elliott Waves all over the place...

if some get 98% accuracy in the next 48 seconds, I will be their slave for life.. shoot if they 98% accuracy in the next 4.8 seconds, I will do that.. 70% is probably the best anyone can do. . My scalp ability is as good as any and 98% even in the next 2 seconds is preposterous.

I also don't believe in complex mat in the market.. it's BS... the market is all statistical math.. simple statistics but assessed on multiple time frames... combination statistics. Adding the statistics of correlated instruments is about as complex as it should get.. if anyone shows me a formula that will launch a rocket to the moon a a Blackbox secret formula, it's a loosing formula.

I used to play competitive tennis and people used to cheat by calling your serves out when it's on the line or clearly inside.. their favorite line is "let me show you the mark." The tennis court has 1000 marks on it... just pointing out a mark means nothing..

Mark, you correctly found the culprit ! TA needs to take time into account and you need your TA formulas to be time deformed. You can use topology or relativism to do that, but it will greatly enhance your results by adding this layer. In Physics, there is an increasing contestation coming from academics (read Carlo Rovelli and Lee Smolin) stating that physics is in a dead end : indeed, most formulas they use are "stationary" because they do not take into account the "arrow of time". You end up having theories that can give results only locally and that fails when conditions change. You end up having unstable results. This is wht you experienced. So, either you deform time and you get better results (but the math to do so if extremely difficult : we did it, and it took 8 years of research) OR you do a heuristic on your own algo so that you grade your signals : you know when not to trust your algo when it is in an environment where it will likely fail (much simpler to implement !)

@Jack, Black Boxes are not all the same. Yes, some are developed to trade a signal instrument many times a day for small ticks. Then there are some developed to trade

a portfolio of say 10 to as many as 50 or more instruments looking for several points instead of ticks and still goes home flat each day to prevent the gap draw down.

A Portfolio Black Box does require large capital, but, it usually gives a better return per share/contract on the individual instrument .

TA is not bad. It does not necessarily fail. But when looking at the very short term, read seconds, minutes or hours, it is about all you can rely on. You try to read the tape.

Fundamentals won't help, they are buried in the market noise and represent only a fraction of what you see.

To trade short term, you need a reason, an underlying excuse, some kind of motivation to initiate a trade. It might not matter much what it is. In that sense, everyone's method described here could do the job which starts by taking the trade under whatever biases.

Does winning 10 trades confirms your methodology? I would say no, not even close. There is so much noise in short term prices that it is akin to reading a randomly generated price series as it develops, which makes it come close to an expected profitable outcome of zero as you view it from a much longer time horizon.

Your trading methods still have to survive randomness, and it appears that everyone has their own methods to do this.

Mark,

My method is simple statistics but if a programmer tries to convert it to code, I think it can be called complex math.. Nobody knows what these blackbox formulas look like other than there is math involved.. Pattern recognition which this mathematician leverage is simpler than my method but I don't consider my method complex math.. though I can see how it can be considered "complex" in code language..

James,

That depends on what is considered small.. My method can be traded with a $500k account but it will just have a greater potential max DD.. 15% max DD is an outlier.. a more realistic max DD is more like 10% on a $1mil acct but the expectation is closer to 5%.. A $500k acct would double all that.. Also what is your definition of small? $1mil might be considered miniscule in that world.. On a $1mil acct, min size is 2 cars and max size is about 20 cars.. 2 car trades pretty much daily and 20 car trades less often and everything in between.. For the max DD to get hit, it would be a long string of stop outs w/o scales on max size.. statistically not as likely if the method has a real edge but always possible with potential outliers.. Typically the trades with the most range potential can have higher stop probabilities due to volatility.. Typically less volatile days have better stop stats and better R/R set ups have higher stop probabilities..

James,

Program trading for points and not ticks are typically either reversion back to means or trend methods.. imo they all work similar... meaning when there's a trend, all trend methods do well... in rotational methods, all reversion methods do well... The key is to know when to use trend and when to use reversion... by the time that is recognized, the opportunity has passed... my method can forecast these probabilities ahead of time with greater than 50% probabilities and that's all I need.. As I've shown in previous threads, I just need 50% probabilities for 15 ticks, 30% probabilities for 30 ticks and 20/10% probabilities for 45+ ticks to extract math from the market.. but its multifaceted how I extract those probabilities..

Mark,

Its always easier to manage risk and become a billionaire if you have a big account.. meaning my method is 1 instrument with a 15% max DD.. If I have $2mil, then I can add a non-correlated asset and bring my max DD further and etc... with a $100mil account, I may have 100 instruments each with a 15% max DD potential but its not all going to happen at the same time.. now my max DD might even be 2-3% if they all have the same risk profile as my CL method.. The R/R of the portfolio only improves... If I can achieve a max DD of 10% and consistently deliver 10%-15% return for my investors, I would be a billionaire too...lol not from my performance but from my fees and take off my net off a few $bil under management...

@Jack, you say: <...with a $100mil account, I may have 100 instruments >. Have you noticed how unrealistic such a statement is for someone trading discretionarily!

@Mark, take a look at Brian Miller's second chart in his article, if you haven't already. I think, if you look closely at the details, you will find something that might improve on your trading strategy design.

Saw your two articles, nice charts. Appreciate your sharing.

Guy - in fact I did read his article and did immediately gleam something I am definitely going to try which is varying position sizes depending on position direction. I typically am all in both directions, but I am going to write an algo to optimize position sizing based solely on direction. Could be something there, Thanks as always I respect your input, Mark

@Mark, I use the same kind of technique. It gives you multiple positive outcomes on multiple trades. Instead of trading a “position”, you are trading a cluster of positions. Notice that all the trades were positive.

In my strategies, I don't sell all the shares in a cluster since I also want to accumulate some for the long term.

Jack - just wondering are you looking at Value Areas to determine regime changes from rotational to trending? If so do you trade trends or just set aside and trade only rotational? I ask because if it were up to me I would only trade rotational days. I am not a fan of trends really, I wish they would just go away - lol.

Market Profile with traditional Value Area and Point of Control with value area lines set to contain 70% of the days market activity. http://markbrown.com/linkedin/20151219-009.png

@Mark, another technique I use is delayed gratification using the same trade clustering as described before.

The principle is this. You have multiple entries as in Brian's 2nd chart, and you are in, or approaching a sell zone, in the sense that your program has instructions to accept a sell on the next bar. To reach the sell zone, prices had to go up, so most trades would be profitable. A corollary to having reached a sell zone is that price were going up, but I still don't know if I've reached the top of a price swing. Being in an up move, I opt to randomly bump the exit to another day. Since you have a trade cluster out there, all the positions will more likely benefit from the delay than not: prices were going up.

You have a more detailed description in the following articles:

.

http://alphapowertrading.com/index.php/papers/187-delayed-gratification

.

http://alphapowertrading.com/index.php/papers/188-delayed-gratification-ii

Hope you like it.

Mark,

I trade both trend and rotational.. Regime change is not determined by value areas but rejection zones.. Biggest size trades are on trend trades, then reversion trades at the edge of balance at rejection zones.. I don't use value areas to trade off but I do watch where price opens in relation to value areas and POC at the opening. and greater balance.. My method is actually very simple but has many potential variables..

Mark,

I also have some confirmation stuff in the PA to confirm.. meaning I don't just have a sitting order waiting... I look for some confluence in entry locations and wait for price to test at or near location and trade my bigger context assumptions.. The most powerful aspect of my method that everyone can also achieve is that I can usually remove a lot of my risk with 1'st scale with pretty high probability and then have a pretty high odds I called the type of day right so my bigger thesis has a chance.. Its incredibly simple but most don't trade this way though some I'm sure do....

Mark,

When price enters the edge of balance, those are typically battle grounds or decision zones... If its an indecision zone, volume and order flow will be light... when decision comes in, volume and order flow is high and you just watch to see who is winning in the price itself.. and you just trade with the winning side and remove your risk ASAP.. I can typically get into a move early and by the time the rest gets in, all my risk is removed... I don't need to capture the whole move, get 30 ticks and let the bigger thesis play out and it does more than 20% of the time... MUCH MORE... As price keeps moving, that' when you trail the stop to where price should not go... sometimes price comes back down to hit your stop but you have to give your thesis a chance..even at 20% hit rate of final targets, you're still further off ahead than pealing it too early.. its important to trust the math and not just do what feels good... delayed gratification as Guy puts it.. sacrifice...

Mark,

The biggest of my methodology is not determining where to get in but how to get in.. The early entries have higher stop probabilities but often offer tighter stops. .. so smaller entries are taken... as price confirms, more is added but the stop range is bigger though less likely... The goal is to choose entries where first 2 scales have high odds especially the 1'st scale... the math is such that if I get stopped out 3 times after 1'st scale, it take 1 trade with 3'rd scale hit even at 40 ticks to recover.... basically 3 failed attempts are allowed for each successful attempt... basically a very bad day still allows me break even... hence I assess the probabilities of that first scale to stop range very closely... Once I've determined type of day, my trade selections are only concerned with that stop range to 1'st scale probability.... basically I scalp to remove risk in order to buy my runners... but its the runners that deliver my math...

im sorry of this comes accross as direct but i dont have much time. t/a doesnt fail the mistake is leveraging up on your strategy. you should study your technical strategy from the perapective of its historical distribution of returns. this should also allow you to guage whether your strategy even has an edge from the skew. then you should only ever compound up positions based on profit not leverage up based . unless you have specifically studied a separate leverage strategy overlayed onto you t/a strat. if this strategy then does not perform consistently enough for you then you should study it to see what types of market behaviour it is not good at dealing with and add in a different strategy that copes well inthose conditions. generally trend following strategies are very linear they smooth price data and throw away alot of information. you can add in a non linear strategy to mix things up. trend following strategies do not cope well when decent trends dont emerge, some people call these sideways or rangey markets throw in a third strategy that is not dependent on trends like vol break out or moentum and then you will have formulated a mqre complete description of market behaviour that will perform more to your likeing i think.

@Jack, if you can articulate it you can automate it.

Many of our algos have never been fully automated. Although I'm often surprised at how good the alpha-signal and parameter choices are for these previously manual systems have been, I have ALWAYS been able to improve these strategies enormously once automated.

@Sanjeev, no doubt position sizing is crucially important, but I feel you may not have studied fixed fractional betting. You may want to look at Vince, The Handbook of Portfolio Mathematics.

In our environment we use variable fractional staking, but I am always aware that increasing position size into profit and decreasing into losses, although the former compounds return and the latter reduces risk of ruin, BOTH increase standard deviation of trades, which in itself can ruin you.

Conversely, not decreasing position size into losses can lead to ruin...

The answer is testing and being conservative and varies depending on account size because of the issue of minimum lot size.

To elaborate, when trading a small account size then you really need to be reticent to scale up or down because of the clunkiness (large relative size) of the minimum lot sizes, therefore you should trade as if it is a larger notional account.

For example, instead of trading 1% per position on a US$100k account, trade it as a US$1,000,000 account with 0.1% per position and maybe don't even scale back into losses (as this increases volatility). That is, a US$10k profit is considered a 1% increase in account and hence position size as opposed to a 10% increase in position size.

As you're account gets larger you have the flexibility to fine tune your volatility or risk of ruin by either trading more aggressively on a portion or still more conservatively on an even larger Notional Account. Typically a fund manager will lean to the latter to improve "Sharpe" and trade very conservatively to minimize ruin.

What this leads to is a crucial understanding of the importance of "Notional Equity" versus "Account Equity" and how they are intrinsically linked to leverage and aggression. ... Most traders just think of their "Account equity" and for a retail trader that can be a killer.

Same as equity traders who typically think of risk as how much $$ they place in trade, as opposed to how much volatility (risk) there is in the instrument they are trading. All these issues are critical if you want an optimal portfolio.

remove the time element and all we get would be point & figure chart.

@Muhammad PF chart is not the same as "Equal size candle chart " .. and ..algorithms are working on PF dont work on ESCC.. .. and vice-versa.. trust me :)

@ .. last nigh i have the opportunity to test on real account HFT (institutional broker).. my algo/strategy ... it cannot be compare with retail brokers .. now i understand with what we are fighting .. :) .. you can double account ..size .. entry in a trade size . . .where before was impossible ... cost is 1/3 I am paying now .. but leverage is 1/4 .. I need HELP .. to find a institutional access to trade on 1-200.000 USD/Account

This are the know / usual types of charts ..where algorithms are different and most case don't work from one charts type to others (maybe.. MA /RSI /MACD..) ..also are .. same very . ."special charts " custom made .. same charts are not available in MT4 ..because cannot load tick data .. maybe now the answer is complete :)

1. Line Chart

2. Bar Chart

3. Candlestick Chart

4. Volume Chart

5. Tick Chart

6. Range Bar Chart

7. Point & Figure Chart

8. Renko Chart

9. Kagi Chart

10. Three-Line Break Chart

@Sanjeev, I totally agree with this approach of strategy development and deployment in a portfolio environment. If the strategy/algo’s are indeed robust,

then it should be able to start with very small share/contract trade size and a small allocated capital per instrument and build trade size using profits.

If the strategy-instrument combination requires more capital, then stop trading it and do a complete evaluation of the combo.

A portfolio is only as strong as it weakest strategy-instrument combination.

Not a precise read of the market but a statistical read... and those statistics are in a consconstant state of change but it can be read real time.. the market is constantly giving you information by its behaviors... constantly....

the last scale delivers all the math. When I trade ES, my first scale has highe hit % than CL but my CL results far outperform ES mostly from the runners. The runners deliver the bulk of the math. ES is a great instrument during times of volatility but most of the time, not a great choice to trade. There are better better choice instruments out there..

scalping wit size is like selling nake puts.. you may win even 7 to 8 out of 10 but that 1 or 2 loss typically clears away all your wins.. it leave a lot of openings for outliers is a given.

The options market is a good reflection of trading.. is it better to sell or buy options? Again it depends on the market probabilities and it's again determined by context...

selling options is like scalping... buying options, you can take many hits and recover with a handful of winners.. but sometimes the premiums are too rich and you should sit it out.. all those decisions are made from context..

Trading in the market is like this profile picture...

Do you see a long set up because the vase showed up for you or do you see a short set up because a face showed up for you? There are always trades on both sides of the market when you scalp but they don't consider directional bias... I see this most often in traders that scalp the ES... in fact, I've done it myself in back in the days... Yes, I think we've all done that...lol.. stop fronting by pretending you never have.. Typically its something like aiming for about 4 ticks target with about an 8 tick stop... That means if you're hitting 70%, you break even and most don't even get that.. At 80%, there's a small profit.. just do the math... Basically spinning your wheels.. That kind of a model typically does not work... That's why you have these new power computers at the exchange rapid firing for a few ticks.. The typical algo trader looking for a scalp model is going to have a hard time...

Interesting post Mark - thanks for sharing. In my opinion, two of the main culprits for TA fails are a) insufficient analysis and b) unrealistic expectations. a) Too many quants focus / rely upon creating one 'perfect' model instead of many 'good' models. The markets are too complex and historical data sets too small and full of noise to be decoded using a single model imo. b) Even with robust analysis, outside influences are always in play and cause even the best models to not work for periods longer than most traders are willing to endure. So the solution for me has been to create many robust models (ensembles) and apply them across many instruments (and give them time to work).

Jack - I have a weird way of trading. I always practice reversion to the mean as the primary market action. As such when I want to short I do so and about half the time that position will flip long. My long positions are only just long, I don't flip them short. I didn't come to this conclusion out of thin air, it took billions of man hours. So I don't trade trends I only trade counter trend's with reversals which then try and go with the trend if it is present.

Scott - I hear you and your method is a very viable one indeed. Many successful money managers trade just that way, no doubt.

Matt - the only thing about trading many markets is, now you have a bed of fire ants to manage with all their problems rather than just paying attention to one fire ant. I would rather focus on one liquid market and be great at it than to chase 20 markets and be fair at it. It really not a bed of roses either way.

Mark,

That's not a weird way of trading. Most methods and particularly aldo methods are reversion methods. In balancing markets which is where the market spend the majority of its time, reversion methods are fine.. trends are when the market is in a state of imbalance and those are more tricky set ups..

If you know what you are doing one trading asset is enough, if you don't, one is to much.

When you trade your own account, 1 instrument can do fine.. when you trade opm, you have to bring the max dd to an acceptable level.. and uncorrelated asset classes can do that.. also the bigger the trading capital, the more vehicles necessary.

It's always best to trade your own money..

Trading one stock or instrument is not enough. It might, at best, appear to fit the constraints of the really small player with a limited trading account.

One has to have some kind of long term perspective not only to manage but also build up a portfolio. Otherwise, don't even think long term, and don't think you will still be a player then either, long term, I mean.

Some more advice.

.

http://www.marketwatch.com/story/john-bogle-never-play-a-losers-game-2015-12-24?dist=lcountdown

@Jack, thank you for your incite and thoughtful comments. I am a old timer at this stuff. I am very experienced in strategy development

across all asset classes. I have many thousands for hours of personal R&D in doing this. When the ES first started trading I developed

my own charting system/method to trade it. I will be the first to say that the ES may not be the best instrument to trade with small size.

There are as many strategies/methods to trade the markets are there are traders. Not a single one is the correct or wrong way to do it. If the strategy/method

meets your over all objectives, then that is the only correct method for you to use.

Over the years my conclusion is no matter the methods used it is a game of “ hit and miss”.

How you analyze the results of the hits and misses is the building blocks of a robust strategy and not really the algo’s

within the strategy. You can build a strategy out of a “broom stick” and get some interesting results.

Happy Holidays

James,

Yes.. the bottom line is all of it is nothing but a statistical guess and hopefully we get the statistics right.. There are many ways to extract some math out of the market but the goal is to extract maximum math.. Yes, I also agree the ES should be traded with a larger account for maximum edge.. For my way of trading the reason is not so I can put on my size to get more out of the scalp but give me more versatility for my entry and trade management which increase edge..

Happy Holidays

What are you referring to in "taking the time out" ?

Charts that don't use time such as Point&Figure , Renko charts ?

Cheers,Manuel

Mark - I do appreciate your comments and can respect the "mean reversion" game. It pays to know what's going on with the other side of your trades. Having said that, the mean reversion game is best practiced by (the once but gone) floor traders. It's interesting to look at that mindset on the Lewis Borsellino story: SOLD!

https://www.tastytrade.com/tt/shows/sold-the-lewis-borsellino-story/episodes/sold-the-lewis-borsellino-story-12-11-2015?locale=en-US

Recently he launched a new mentorship room called ManOverMarkets: http://manovermarket.com/indicators/

But what I found interesting was his mindset of using VWAP standard deviations to fade moves... I mean, that's his personality... it's ingrained in his brain that once a market move goes into a certain standard deviation he just takes the other side. The one thing that is missing is bet size and trade management, but I guess we can take that advice of S&P 500 pit trader Greg Riba https://www.youtube.com/watch?v=8xyUaSY4ZMg

Matt - mean reversion to a floor trader is quit different than the perception an off floor trader has of it. It would not be fair to assume that you're on the other side of my trades just because I trade reversions. Nor would I discount anyone's methods as being extinct. I did state that my reversions will flip to the long side only as that is just the way I trade. Price action has to force me to not revert. I would never underestimate anyone's methods, there are zillions of ways to trade these markets.

I appreciate everyone's opinions here and listen carefully to what is presented. Sometimes traders will drop clues and tips that will provide a missing part of the trading puzzle. The best fundamental stock tip I ever received was from a security guard who's wife was a nurse. Her company was being bought out by a larger company for years after that stock was phenomenal.

Univariate, quantile analysis vs. an asset's performance space "removes" time in a way—by collapsing into a pseudo-scalar window (2 weeks in this case): http://stocktwits.com/message/46349934

I agree with Bogle as long as objective is investing. Even if it means investing in yourself for a startup you should not have short term expectations of making it big. However, if we are employing T/A for purpose of generating income, keep the lights on, pay rent, etc, You have to manage risk through short term, disciplined, and highly structured strategies. Not everyone has access to billions of investor funds that help offset management costs for long term prospects.

Look at crude oil today.... How would you trade it? Today order flow was very visible and tipped the hand some but it would have been a hard trade to take..

Only shorts should have been considered today for what should be obvious reasons.. The short locations were 35-35.10 area which was an ok set up and the appropriate size was about 3 cars.. and then another set up came in around 34.30-34.40 but again very so so and a 2 car set up... If you were trading a smaller account, both set ups should not be taken because the market offers better set ups.. The good trade was in the Euro session just below 35.50 which was worth probably 9 cars... on a small account 3 cars... If you missed it, then you are left with lesser set ups... On a small account, the right move was no trade..

I have the macro direction all pointing to short.. and the micro set up in the US session was a long.. In those types of scenarios, the right move is to only take shorts trades... If you take long trades, they must be small in size and range.. so no more than 2 car trades.. Today, there was no good long set up but the micro structure says long... and then balance... in those scenarios, the shorts are very difficult to take though they are set ups in small size only for a larger account.. in smaller accounts, the trade selections must be more disciplined for the math to work...

The assumption for crude is it will continue lower... with that statistical assumption, the question is where and how do you capture the math? Suppose I took 3 cars... I would have scaled out of that trade around 20/40 and 60-70 ticks on 3'rd... If I took 9 cars, I would have dropped 3 at 20 to mitigate risk, 3 at 40 to lock in my profit and with the house money I would have took 60, 90 and 120 on the last 3..... now do the math how much more I would have captured on that move that was supposed to give range? With a bigger account, the statistics improves geometrically because you are able to capitalize on the opportunity much more powerfully... your initial risk is always the same or similar... its how much of the math you capture that changes with your ability to capture the math.....

When the market offers you range trades, you have to have enough runners to peel off.... Suppose I have 1 runner and not 3... Would I take off my 1 runner at 60 ticks or aim for 120 ticks? 60 of course... the reason is because I can easily get to 80 ticks and then reverse and stop out... I'm not going to let my lone runner aim for 120 ticks watching it whip up and down like that... you want to lock in your gains.... The bigger the range expectation, the more runners you need to build into your trades... so you have more to peel off and the odds of your runners hitting their target improves and your risk is further reduced... while the whole time, the initial risk remains similar.... The risk remains similar if you trade 3 or 9 cars, but your reward is geometrically improved... now imagine 20 cars on a 300 tick move... the market offers those opportunities several times a year if it allows you in.. If it does, you need to be ready to take 21 cars and not just 3

With 21 cars, you have bot 7 runners to peel off with your expected range... you then aim for somewhere short of that range expectation and divide it by 7 and peel off at each interval.... Suppose its 300 ticks... Are you going to peel off at 15/30 on 2 cars and then 300 ticks on your lone runner over a week time? Hell no.... Crude also has an average daily range.. Statistically when that range is reached on a day, more than 50% of the time its done... and that range is not 300 ticks... so you might carry some overnight... but by that time, all your risk and a bulk of your runners should have been peeled off... You're hoping to wake up to a nice surprise...

Jack ..when you will finish your saga ..let us know ..

I think Jack is not near end yet.

The saga ends here... take care... I might open my own thread for CL somewhere down the line...

I cannot agree more fully with your final assertion. What many technicians fail to realize is that the most critical part of establishing a profitable T/A algorithm is not only the underlying function but the method of plotting price activity. To clarify, this does not only refer to the use of standard modified charts such as Renko but the ability to customize the plotting based on specifications of your underlying function.

Actually Mark, you could do the opposite and add time to the equation. Let's think about this, you make a trade with the aim of getting out of the trade very soon (you fill the time period desired). Judging by what you posted, if you take the time element away from the calculations all is well. As far as I see it this can be done in the extremes, if the time interval desired is so short that it can be eliminated from the formulation, or if the time interval is so large that it also adds nothing to the formulation. I believe people like Buffet play with long time intervals and that is fine for me. I also believe that there is another successful school of thought that deals with extremely short time intervals multiplied by a multitude of trades in an actionable (by humans) period of time where the time factor is also negligible. This other school of thought is also successful. So, we may have two flavors available for trading in which time is not a major concern, both successful.

I was thinking about time and how it relates to trading. Since I am trading price levels and that alone is what determines profitability, then why should I focus on the element of time. Of what use is time in trading other than to monitor an imaginary race of some sort? Is trading a track and field event where timing is what determines the winner? So why waste any resources on an element which is not a factor in trading profitability.

We spent several years applying the scientific method to classical technical analysis only to find it doesn't consistently work. As a result we developed our own proprietary metrics and logic that handily proved themselves in a very difficult 2015.