Several times a month I am presented with the most generous of offers; it comes in many varieties but in the end, they are all much the same. It starts with a brilliant idea that only requires a bit of additional help/expertise/experience/financial support to make us both rich beyond our wildest dreams. In fact, it is so brilliant, "disruptive", forward thinking (insert buzzword bingo word of the moment) that I should consider myself lucky that you have even considered sharing it with me. Your greatest concern is that I may take your brilliant idea and apply the last tiny bit of whatever to help it achieve its vast potential and leave you wondering why you ever trusted me in the first place.

Trust me and please understand this comes from actually trying to help you - I have notebooks full of ideas of my own; I have tens of thousands of lines of code of my own, I can trade hundreds of contracts at a time and while I hate losing money (or maybe because of that), as is, I don't have a lot of losing days in a month.

So I am not going to steal your idea, nor are any of the other professionals you would be best served by approaching. What you have managed to do is indicate that your idea is worth more than anything I have mentioned above and when I ask why you don't just do it yourself there is always a reason and that reason almost always boils down to the same thing - I want to "accelerate" the effort and/or I lack "fill in the blank".

And if I relent and say ok, tell me your idea, it is almost always a vague variation of buy low and sell high, based on your insights into market dynamics, sentiment analysis and things you cannot remotely quantify.

I have helped/collaborated with a lot of people and I hope to do much more of that but there is a huge divide between those I share ideas with and even in a few cases code and those I don't.

To avoid this turning into a unintelligible rant, could some of the other professional traders/algo developers please add their two cents worth and could we maybe collectively explain what would in fact pique our interest, not because we think it will make us rich but because the idea is in fact interesting and the person presented it has done a lot of hard work thinking it through and isn't looking for what amounts to a handout.

1. Type 1: in most cases the computer is believed to be able to do something man can't do; here come various "machine learning" technologies. Of course google or wikipedia is not available to these guys, otherwise they would have opened it and read that machine learning inevitably requires expert training, therefore it only conveys the human opinion, although in a sci-fi form.

2. Type 2: perhaps the most typical issue with discretionary-going-quant traders is that at certain point in their strategy logic there is a tiny thing which cannot be formalised. You can shoot yourself dead after having spent hours of discussions trying to understand what the author had on his mind there, and explaining the author that you can's tell the computer to "find the latest most significant peak" without formalising "significant" first, and "most" then, but you won't be heard.

3. Type 3: in this case the trader is inspired by success stories of large institutions, especially heavily rotated in media. He dreams of replicating the magical algorithms they employ. And he doesn't understand that most of their success is in using those markets and instruments which are completely unavailable to retail traders, along with diversification strategies which are also unavailable in case you are underfunded — and from the institutional point of view all of us are.

Of course there are more, and I will try to suggest them for your consideration as it seems to me that this discussion finally touches the very core, roots reasons of failure in this part of the business.

An interesting topic of conversation for a change Scott. Thank you!

I am on the public record in these hallowed halls as stating on numerous occasions that "If you can't backtest it, it's BS", to which I occasionally add the rider "unless it's high frequency".

In this context I'll restate that as "If you can't backtest it you can't automate it", to which the corollary is obviously "If you can't automate it you can't backtest it".

How many of the "generous offers" you receive have even begun to address that conundrum? How many of the "collaborations" you engage in conclude with them asking you "can you backtest it for me?", which does of course require that they reveal their "most secret sauce" to you.

Good heavens, reading these posts makes me realize it isn't just me being "close minded", "negative" or that I "lack vision" (these are only a small sample of the comments I receive and completely leaves out the creative invectives I have had thrown at me for not wanting to participate).

As I have stated many times, my edge is primarily order handling, knowledge of exchange rules, a very clear understanding of the various probabilities involved and as Alex pointed out under category 3, a whole host of things that are simply not readily available to the average retail trader - commissions so low that they don't even count in the model, pristine data, microsecond latency, etc.

@Richard - quants in an institutional setting are often involved in incredibly mundane things. Until one works in that environment, you don't realize how tedious the vast majority of the work is. Yes, there are some ah ha moments and some vigorous discussions but it isn't like it is in the movies I can assure you.

The easiest way for someone with an idea to differentiate themselves from the vast majority of the people who approach me is to ask for guidance, not complete solutions. I love questions like - have you thought about this, have you tried something along these lines but only if prefaced by - I have worked on this for y YEARS, not d DAYS, I have traded this live but seem to be having trouble with..., etc. I can point you in a direction and give you lots to think about but then it is you that have to go do those things.

If you really believe in your work, you will carry on, sometimes for years, chipping away at the problem until you, yourself have that ah ha moment.

@Alex and Jim - you have clearly suffered through those conversations where you quickly realize the person really doesn't grasp how long some of the most basic tasks take and how difficult they can be to get right.

Some of the most frustrating things about algo development have zero to do with the strategy at hand but rather tasks like having the right database driver or getting the directory structure right on a new server (both of which I spent hours on last week). If only I could spend all my time just on creating new strategies and didn't have to actually get them to work from a technical point view.

PS. I am not going to hit the "Like" option for every comment but I do in fact, like them all. Clearly the posters understand the issues involved and have likely been through the same experiences I have.

The examples I have given are just that, examples, and are not directed at anyone that has commented. I greatly appreciate the comments and hope they may actually improve the quality of the discussions we can have in this forum.

As Scoot mention, it is perhaps not the strategy itself, but the surrounding around the strategy which takes the time. Database problems, API to broker, backtesting environment, order management etc. On top of that, the need to constantly further extend the develop & test environment with new features, maybe even rewrite major parts of the environment.

The non-developer do not realize when sending an email with a simple request, what kind of work would be behind to implement, test and run the idea.

@Soren, your points are spot on; those types of tasks take more effort than many realize, are the most prone to errors and have the least to do with the actual trades. However, these are critical to having a solid OMS/EMS.

@scott

Speaking of low transactions cost...have you seen CME memberships? They at 2008 lows yet CME had record volume in 2015.

Quite so Scott.

"The client" (invariably?) expects that "the VPS" will run itself free of charge for years on end without human intervention, whilst churning out a steady stream of profits which "the technologist" is welcome to a modest share of once they materialise.

In practice "the client" (invariably?) tends to get exceedingly excited by the first "optimization" run, whilst ignoring the "sensitivity analysis" helpfully provided free of charge by "the technologist" who has laboured long and hard to convert "the client's concept" into testable form.

Things tend to go downhill swiftly after that point!

@Manuel - I have not seen the current cost of a CME memberships but from what I understand they have come down significantly in the past few years.

@Jim - yes, yes and yes. I have only seen a few categories of trades work consistently over the years and none are new concepts. The first category is arbitrage, whether it is statistical, latency, triangular (or a higher order) in FX or some form of cross exchange price disparity. The second category is market making of some sort but specifically for me, directionally biased market making and the last category is manual trades with a long track record as is, that are then automated to provide better order handling and execution.

The various categories require skills few possess and costly infrastructure, often tens of thousands of lines of code to get right, in depth market structure knowledge or a long track record. None of these are likely to be just happened upon. A brilliant new idea is a possibility, but it isn't something I want to bet my time on - which is what I am normally asked to do.

If programmer made sure that

the trader's tracking record is geneuine AND achieves "in ratio" EQUAL or MORE than programmer's tracking record,

THEN, if he doesn't accept, there is a deep lack of management from one or both parties.

@Abdullah, The main thing I am trying to accomplish with this thread is to give insight into how senior algo developers think and what would interest them enough to actually take on an outside project.

No one I know would stop trading or quit working on a 100% known quantity (i.e. their personal systems) for a percentage of an unknown quantity (i.e. the track record of the trader with an idea). So suggesting you can compare a system idea to what an algo developer trades on a regular basis as is, does surely point to a deep lack of management as you suggest.

So what are the things that might interest an algo developer. I can start by telling you what won't be of interest and hopefully others can fill in what would interest them. Anything that is at the idea stage or at a standard back test stage has virtually no chance of being of interest with the possible exception of something that has undergone rigorous walk forward testing.

I will let others take it from here.

"Tracking records" are not "ideas".

@Abdulla that is very true and along those same lines, back tests, however impressive are not track records. In fact, maybe I am out of touch but when I see huge returns and it isn't a market making algo, I become very concerned.

Remember, I am not telling anyone what to do or what can work, I am only discussing what will help or hurt your chances of finding someone who will collaborate with you.

Anybody, and I do say anybody that can program can program a worthwhile trading strategy. Geez, how hard can it be? You buy some stuff, resell it later and make a profit or not. What is there not to understand?

You want a sure thing, forget it. But if it was the case, and you had one, you would not need a trading strategy in the first place, you would simply execute.

You develop a trading strategy because you don't have a sure thing and you know it. You are ready to play averages where hopefully you would be winning more than you lose. And that is why you do those backtests, to gain confidence in your trading methodology. Now if you don't do your job properly, cut corners, curve fit, peek (cheat) don't expect me to follow in your footsteps either. Should you put out performance results that are statistically insignificant, or cherry picked examples, again forget about it.

...more

I can not reject or deny all the studies and academic papers that have been written over the past century. On the contrary, I rely on them to explore trading avenues that have been considered in the past and test them with state of the art programs and equipment. I will also modify them to suit my current views of the problem as if always seeking better ways. But before accepting any new programs, they will have to perform better than what I already have, otherwise again why bother, they simply end up in my code snippet library as nice try but no candy.

In the end, you will find that you have only one person to convince and that is you, nobody else, just you, since you will be the one to monitor your trading programs, it better do what you programmed them to do. It is with that conviction that you will be able to interest others in your trading methods, if at all.

@Abdullah, Richard's suggest is excellent and probably the way to go if you have the means. It solves virtually all the issues at once.

To directly answer your question, there is very little a trader can do to tempt a programmer to program a system for free (a cut of the profits technically but this rarely results in much if any income). Traders who are not programmers often do not have an appreciation of how precise the trading rules need to be for a programmer to be able to replicate them in code.

Even when I am the trader and the programmer, I have difficulty communicating my ideas (and this is me talking to me). I can tell you about an API for market data that sends order book update timestamps as yyyy-MM-dd but sends trade timestamps as MM/dd/yyyy (can you spot the difference immediately). Resolving that type of thing can take a lot of time.

I say, take Richard's advice.

@Guy - very well put.

https://www.fxcm.com/services/forex-programming/

above is a link to FXCM programming for 'retail'

- also you can get EA programs for MT4 trading platforms which can be linked to NY4 and LTD5 IBX servers for low latency trading programs - ICmarkets.com provide these EA portals and they are ECN so have zero spreads in FX [even negative spreads]

good for spot FX and FX cross stat arb or scalping - as have 50 liquidity providers which some include Virtue of HFT firms

Richard - this is now over, strategies are not programmed any more. Check this link soon to be public.

http://www.ecom.onl

Let me add just my 2 cent. While I 100% agree with everything said by Scott he seems to have missed one possible reason why a programmer doesn't want to be paid from "future returns": when the programmer clearly sees that the suggested strategy is informal, cannot be formalised and cannot be coded at all. From my (relatively brief) experience coding strategies for others, approximately 99% of strategies suggested to programmers are exactly of this sort.

100% of those systems are various visual systems, including drawing various lines on charts. Traders simply don't understand how a difference in the 4th digit to the right of the decimal point in the next bar turns into several percent difference in a distance. Needless to say that most of them draw lines the way they want them to look, not through somehow defined points.

I have had quite some experience working with traders who were not only profitable, but were almost trading superstars some decades ago. Having spent several months of work we inevitably came to a certain point where the trader suggests to do some guesswork. He can call it "anticipation", "prediction", "looking like", "nearly", "almost" or whatever you like, but the fact is that his strategy has never been formal, and all his success came from something he did unconsciously.

So, at certain point I stopped helping traders transform their strategies into formal algorithms. In general it seems to me that in this business (coding strategies for traders) either you are honest and then don't work, or you are dishonest.

However an aspiring trader won't ever stop trying to find "the right programmer". As the most recent example, I've had a talk with a guy who draws lines through "most prominent" peaks and valleys and then does something with them. I even didn't listen to it any longer when I understood that he simply doesn't have a definition of "most prominent". I explained that no one will ever code this for him, or, in case someone will claim to have coded it, most likely it will be something different from the author's expectations.

You would further not have a single strategy running, but a portfolio of different instruments - under abnormal market conditions, they will tend to compensate. A robust portfolio, not a single strategy. Besides, a strategy would contain information from the rest of the surrounding market, which will stabelize the strategy, and the portfolio. But to do that, you need the development environment which provide this (with the risk of Alex sending a private message).

Alex, yes I've thought a lot about this. In a dynamic environment such as the market, all of this technology, data, machine learning, and analysis techniques are only serving as amplifiers or multipliers, where in the end a human must make a decision. I've thought a lot about whether a small group of people could be put together to rapidly process this information from the tools, cut off dying strategies, identify new opportunities, and rapidly develop the new strategy in ever quickening cycles. This would be an example of an "agile trading" group.

My intuition tells me something like this exists already. But maybe I'm wrong. This industry is not very forthcoming on how they operate.

And finally - you have your automated trading running for example from 6AM to 12AM CET. And then go home. That will cut off the majority of the market news.

Jacob, yes it does exist, funny how it sounds, such a process can even be more or less formalised. At least this is what I've been doing for the last decade, and I'd even dare say it's a "Russian old school of trading" :)The point is that you're unable to do all you mentioned looking at the performance of your strategies alone, and I doubt if the decision to increase exposure or stop trading a particular strategy could be made based purely on analysis of price time series (OK, +volume, +OI maybe). However as soon as you look rather at the market processes which provide tradable opportunities, then the overall picture becomes much clearer.

Jacob - yes you could make a group, but the tech race is running fast now, the future looks very different from what the old boys did the past years, and before you have gathered such group, it will be disrupted by technology.

Millions of VC money are injected into robotic trading - JP Morgan Chase has robotic trading and big data as focus for 2016, like others in the market - so we can expect to be "run over" by high tech systems in the future.

Well said Soren Haagensen

Richard - you are right there is a demand for getting ideas programmed - and there is a supply. The point is, this is not operational.

Statements I've read in this forum.

Statement #1: All automated trading strategies eventually fail. Great, then problem solved!

Absolutely no need to design or develop any kind of automated trading strategies whatsoever since they will all fail.

For a trading strategy to fail, it will first have to lose your money. So, imagine all the savings you will have by not automating. As a bonus, there is no need to learn how to program either. And as an extra, you can go around from forum to forum telling them all: I told you so.

Why would someone want to learn to program to automate losing money? Do I need to explain this one, provide some logical detail that might have been missed? Here again: program will fail, program no good, program will lose money, program bad, bad.

Can't traders lose money by themselves, on their own, any way they want, any time they want?

...more

Another point would be that a forum like this one is no longer needed or required since all aspects of automation will only guide you on how to fail, on how to lose money. To say that you might have paid for courses, software, and data subscriptions just to give yourself the ability to lose more. Isn't this wonderful.

So, for sure, for those that think that all automated trading strategies fail, you should not listen to any of it and stay on your discretionary apprenticeship. Which would mean indirectly, that you should continue to make peanuts plays for peanut rewards, and even this, only if you are good at peanut discretionary trading.

If you do not do any programming because it is hazardous to your financial health, then how do you know that all automated trading strategies fail?

...more

Don't you think I would find your opinion on this irrelevant and expressed by someone not really knowing what he/she is talking about. All, is a big word. Maybe we should leave the door open for a maybe it can be done.

I contend that anyone can program worthwhile systems, anyone at all. But they do have to do the work.

Multiple failing systems do not give you a good portfolio of trading strategies. You will just fail differently. Also, running multiple systems will require that they also be funded. Then another question will rise: why would anyone want to run a good and a bad trading strategy? Shouldn't all the effort be put on the good trading strategy, or was it also a failing one?

For those expressing the first statement above, maybe it should be worthwhile to consider the implications of such a statement. Making such an assertion might require that you bring some proof to your argumentation. Otherwise, you are just blowing some wind.

@Jacob - you are exactly right regarding the usefulness of Agile practices in the design and development of trading algos. Just having the equivalent of a Kanban board and thinking in terms of being a product manager with daily "standups" is an amazing boost to productivity, especially if you have a team distributed over many time zones.

@Guy - your statement "but they have to to the work" may be the key to what I am really trying to convey. One has to do the work; in my opinion, there are no real shortcuts to the process.

@Scott,

I'm really good with development and technical stuff, but I am still new to markets and trading. I'm glad to hear that you are successfully implementing some agile practices in your trading. I'm actually writing about this at the moment. I'd love to ask you some questions. Does this "standup" also include evaluation of your trading strategies? Is there a research component to these meetings or are these done separately? How much influence does the lowly developer have on the strategies being used?

Blimey. I pop out for a few hours and look what happens in my absence!

@Scott - Re your "Yes, yes, yes". Am I correct in interpreting that as:

1) You have never come across a "non programmer discretionary trader" that comprehends all the nuances of backtesting and (over?) optimization?

2) Possibly as a consequence of #1, you have never yet had a (financially?) successful partnership with a "non programmer discretionary trader"?

@Alex - Re your "advice to newbies". Am I correct in interpreting that as:

3) Unless a "non programmer discretionary trader" learns at least the rudiments of computer programming there is no way around 1 & 2?

Re your "this business":

4) Even if you successfully negotiate steps 1, 2 & 3 you have never yet come across "a client" willing and able to accept "the truth"?

@Soren - Re your "personal message from Alex":

5) Can I introduce you to Sid? He's one of my benchmarks for "visual" algo development platforms.

@Jacob - Am I correct in interpreting your resume as:

"Non trader programmer"?

If so, do you have any experience with "Complex Event Processing"?

If so, what's your "favourite programming language"!

I manage my time as I would any portfolio. I have some steady returns for a steady amount of work (trading), I have some safe haven investments (incremental feature additions), I pay taxes (maintenance of my code base/reducing technical debt, going through a learning curve on something new, etc.) but then I have an entire category of risky investments which may or may not pay off.

This category includes strategy development (very risky), etc. but it also includes an often overlooked resource - contacts. I have some very senior people I interact with regularly but I also interact with very junior to mid level traders and in the technology area, even people still in or just out of school.

Advice to the senior people in this thread - someone you help today, in 10 years may be running a major commodities firm or be an expert in a field that doesn't even exist today and believe me, they remember!

Statements I've read in this forum.

Statement #2: I only take trade setups with a 3:1 or better risk/reward ratio. Geez, that's just an unsubstantiated number thrown in the arena. Why not make it a “guaranteed” 3:1 bet, since all that is required is just to declare it so. And, why not go the the 10:1 or the 100:1 gambit, why limit yourself?

After all, you don't need to prove it, you just need to state it, and surprisingly, as crazy as it may seem, some will believe you! After all, it was written in the forum.

The individual saying this can't even provide any statistical evidence to support such a claim. But he will throw at you for answer: it's proprietary IP. How convenient! So, the novice has to determine, without the benefit of experience, what is crap and what is not.

Going forward, everyone is at the right edge of the chart. Everyone is faced with the same uncertainty, except those that are cheating.

Rarely is there anything new under the sun. All trading models are some form of momentum or mean reversion. The "secret sauce" is usually the implementation of these trading models. Every once in a blue moon, there is the discovery of some non-proprietary readily available information that is underutilized to profit from trading. This is probably the only type of information which would ever capture my attention. I have often found that many traders who thought they had an edge with their strategy really didn't have an edge at all. They luckily benefited from a bias that occurred during their sample period.

@Jonathan, what follows is only about what you have posted, nothing else, and it is a different point of view.

There has been over 7B “new” people on this planet over the past century alone. Over the same time period, these people have built maybe over 80% of everything you see: cities, buildings, houses, roads, cars, planes, boats, etc... Last year the US Patent Office issued patent number 9 million, that is a lot of new stuff. So, it is not: , it is more like close to everything is new. Also, if there is nothing new implies that there is no change, and that is not what you see. On the contrary.

You add: <...All trading models are some form of momentum or mean reversion>. I find “all” to be a big word. Could you substantiate your claim. Have you surveyed “all” trading models? Do you have statistics? Anything, studies, papers, anything to corroborate your statement.

...more

You then say: <...Every once in a blue moon,... >. This is a phenomenon that appears about every two or three years. So, one is to presume that your trading method is not HFT. But one trade every 2 to 3 years based on <...the discovery of some non-proprietary readily available information >. This sounds rather vague. You added: <...This is probably the only type of information which would ever capture my attention>. Have you automated your trading methods or is this still discretionary? You know, people don't discover , that just read about it late, after it has been “readily available” for some time, which, in trading, implies you are late to the party.

@Guy, All I do is automated trading, with what can be considered HFT timeframes, and some of it is also low-latency dependent. I said "the discovery of..." such readily available information happens infrequently. That doesn't mean that such information is currently being harvested for trading. I have many times come across such information and used it on a very regular basis. I am sure I could dig up plenty of white papers to back my statement about momentum and mean-reversion, but you too have access to a search engine. My experience as an algorithmic trader spanning 25 years is sufficient. I realize that many prospective traders' egos will not allow them to accept the fact that their method is just some other method that has been rehashed and embellished. There are only so many ways you can manipulate and analyze the data. Most major advancements have relatively minor impact on core system performance.

@Jonathan, see, I would tend to agree with your last post. Like I know I build my trading strategies on developers that preceded me. As a matter of fact, I usually start with publicly available code which I then transform to suit my needs and trading philosophy.

However, I would not agree with: <...I realize that many prospective traders' egos will not allow them to accept the fact that their method is just some other method that has been rehashed and embellished. > I understand you make that as a general statement, but it implies that somehow someone knows that his strategy is a rehashed or embellished version of something else when we have very little means to ascertain that argumentation. I know that when I've transformed a trading strategy, none of the words you used (rehashed or embellished) would suit to describe the strategies afterwards. At times, there is almost if not nothing left of the original version.

@Guy there is one sure way to work on the hard right edge of the chart - determine what your trade would be if price moves from the current price on the hard right edge to some other point and it takes a specific minimum time to get to this other point and if you want to get more sophisticated you could add in volume.

At the right edge of a chart, there is no sure thing. So, even if you set a future price point, up or down, you can't be sure which one will be hit, if hit at all.

@Guy, that is exactly my point. I don't predict, I react with a plan for each of the major probability paths. If there were a sure thing I would quite imagine someone would have found it by now and be living the life of luxury somewhere. Or maybe they did and were smart enough not to share it with the world. However, I am going to put my money on, there is no sure thing.

I have to say that I find an interesting aspect in this discussion concerning the relation between the trader and the programmer. The general consensus seems to be that the trader, as an uber-intelligent pan-dimensional being, comes with wonderful ideas to the "mere" programmer (most of these ideas half baked TA from last millennium when, in the largest bull-market in history everyone thought they were good at TA for that matter) who does the dirty work and gets some "golden crumbs" falling of the plate... ??? And what the other way around? A genious in development comes to a trader with really good code thinks he can apply it to markets ... they find a way to make it work / adapt to a market paradigm ...? I think this last combo has got the larger innovation potential ... is it rare because it is hard on the ego of the pan-dimensional beings?

@guy marcelis. It might be that you somehow misinterpreted some statements. In previous threads, we have discussed at lenght the fact that so many quants/number crunchers, ... are no longer in touch with the markets and the underlying reality because they have never spent time at trading desks, etc.. . So, if you do not have that background, it is not obvious to come up with a trading strategy. However, from the discussion above, I think it has become very clear that both (trader + developer) are a team where the trader in fact is the "dumber" part when it comes to the core of this group, i.e. Automated trading. Without my IT, I do not exist !!

@Guy M - Thanks for bringing this discussion back on track. After returning from the wilderness to a (for once!) potentially interesting thread here on LinkedIn I was beginning to get bored again by all the "there's nothing new under the sun" and "I did it my way" stuff.

@Søren - I note that you have yet to answer my question about how easy it would be to program/backtest/verify superstitious stuff like "phases of the moon" using a visual algo programming platform.

@All - Why do I suddenly start receiving a series of PMs when I start commenting on a thread about how sending canonical LinkedIn PMs is not an awfully cunning plan?

@Jim H: don't mention it ... liked your emo-bot but could not find a reference to the yields ;)

Jim @ if the development environment is "chart independent", then you can use any kind of data you want, also from the dark side of the moon ;-)

@Jim - Sorry, I thought I was staying on track

@Guy M - I can only speak for myself, but I started out as a trader and then decided my best path to implementation was to write the code myself. I do have a partner who focuses entirely on the trading aspect, while I mash out code, but I do feel that a strong grasp of both domains is optimal.

@Guy F - I was more referring to the fact that the resulting strategy will still be some form of mean reversion or momentum, no matter what transformation applied.

At some level I think Jonathan is describing a third category - the trader who can code, who also understands order management, exchange rules and execution management.

Well not in terms of the threads headline ;-) he would then be talking with him self, saying no to the other "him" ;-)

Which perhaps touch a sensible issue:

"Is the coder at all the right person to be in full doing every thing ?"

From my experience the answeris , no. The coder will typically get no where, go wild in huge projects, as the tech and the innovator takes over, and leaves the economy behind as secondary.

@Soren, most of the guys I know who are successful in this business control their IP end to end. Some were able to pay for it to be written for them, some were traders who learned to code (very rare unless they came to trading from some form of technology background) and some were coders who learned to trade.

I am highly suspicious of any part of the trading process that I don't know as well as the person who did it. If you are the trader with the idea, can code it yourself and understands the order/execution handling well enough to trust your code in live trading then you understand the entire process because you understand every small part of the process. I use generic packages for logging or low level communications but I wouldn't outsource a single line of critical code.

Anything that does something for you, can do things to you.

Whenever I request a new trading script, my software instantly answers with the following script:

var Bar: integer;

for Bar := 20 to BarCount - 1 do // from first bar to last bar

begin

.if not LastPositionActive then

..{ Entry Rules }

..begin

...{ Insert Conditional Entry Code }

...BuyAtMarket(bar+1,' ');

..end

..else

...{ Exit Rules }

...begin

….{ Insert Exit Condition Code }

….SellAtMarket(bar+1,' ');

...end;

end;

That's it. The whole program framework, even customizable to whatever I want. So, when I hear: it is so hard to program, I don't see the difficulty. You certainly do not need to be a PhD to do this stuff.

You can take a doctor and make him a taxis driver in not time at all, but you can't take a taxis driver and give him a scalpel. It is a lot easier to make a programmer out of a trader than to make a trader out of a programmer. What's more difficult is setting the trading conditions, the environment, and designing the desired procedures.

Look at EA on MT4 trading platforms using MQL4 language .......even python is an option and easy to code = you can create your own trading program and have it operational to go either live on demo accounts for back testing or live on the trading platform - if get in trouble, someone can help - an ecosystem for the dumb and dumber retail trader - not to push an agenda or repeat myself ad nausium : the risk is with the trader - and a programmer is optional

If a programmer is interested in financial markets the world is open to them - if a trader who knows no code but can trade a discretionary system - and wants to automate, it has never been easier = it is their journey

I am at this point scoping [pilot build] UI for discretionary trader to create an automated trading program but retain 'their' discretionary element = not boundified by strict rules which can leave a lot on the table

- hard to explain : maybe mad, but I think the journey is important and who knows

@Guy F..you state, "It is a lot easier to make a programmer out of a trader than to make a trader out of a programmer".....First great thing that I have heard from you in years...Just kidding...Domain knowledge is paramount. Look at so many of the Kaggle competitions for examples.

@Jonathan - My profuse apologies if I have inadvertently offended you in some way. My critical remark was directed at much of what preceded my "Blimey!" comment. I was however particularly taken by Guy's "ego of the pan-dimensional beings". ROFL!

@Guy M - Sid's yields live trading the FTSE were not terribly impressive. However I'm waiting for Søren to impress me with how easy it is to port him from MQL4 to a "visual" environment and then backtest him against the Hang Seng instead.

@Richard - Sid is an "educational EA on the MT4 trading platform". Please feel free to examine his source code and let me know what you think. I take it that you are aware that the MT4 "Strategy Tester" leaves much to be desired?

@Scott - Quite so. However what should one do if one is a trader but not a born programmer? Alex & Jonathan's "Teach yourself to code"?

Maybe sometimes it's best simply to stay a trader and trade? Sometimes I think that if I were born a trader, I would have never spent so much time on market analysis, finding ideas and coding.

Alex, very true.

@Jim - Offended? LOL. I wasn't offended. It would take a lot more than that to offend me, if that's even possible. Thick skin is something most traders have developed over time.

I find it most laughable that I've already been approached by several groups selling their systems/algos after I posted in this thread.

Jim @ There are thousands of EA´s and people trying to sell their backtest results out there. "Sid" is just one more EA in this ocean. I personally simply do not have the time to get involved in experimental projects.

I had the other day a contact to a guy, who has used the last 4 years trying to find an EA which perform - I think that explains all. Get real, no one sell a working signal for 20$ - they sell the signals which do not perform, or are too risky to use.

@Jonathan - It seems that we both have thick skins and we're both laughing, which is exactly as things should be!

@Søren - I wasn't suggesting that you should "get involved in experimental projects". I had in mind more some sort of explanation/demonstration of how a "visual" algo development system might cater for an "outside the box" trading methodology, together with the "confidence limits" that I mentioned previously.

@Guy F...I have got to hear this one. Tell me what you think that I mean.

@John, they are not the same. If you play momentum, even temporarily, you are following whatever short term trend that is there. While if you play mean reversal, you are betting that this momentum, in whatever direction it was, will reverse.

@Guy F. I invite you to contemplate more deeply your statement.

@John B. At the risk of hijacking my own thread, I'll bite. I am with Guy on this one but am open to a different point of view unless this is all about semantics.

Hey Scott....I am off to bed. I will explain more tomorrow. Hey, I told you to take it easy on those big words with me. LOL...No, not about semantics.

Søren - OK, thanks. I await your suggestions with interest. Please bear in mind Scott's OP though!

1) Is "coding" Sid and/or his CEP contemporaries in VS instead of MQL4 easier and more intuitive for a "non born programmer"?

2) If so would that actually help to make a "partnership/collaboration/group effort/joint venture, etc." between traders and programmers more productive than has historically been the case?

3) Could it even cut Guy M's "genius in development" out of the loop entirely?

I'm kind of late to the party, but I can say from experience that as a non-programmer, I survived because I had an ability to explain my concepts logically to the programmer that is still with me since APAMI was first released. So I have the benefit of developing a 'language' of the different components of my trading system when discussing ideas with my programmer. This is similar to how Jackie Chan uses different sounds to indicate whether he will kick high, waist, knee level, etc when performing his martial art sequences. That takes time to develop that kind of rapport with your co-workers. 100,000+ man hours paid off.

What does logical mean? That it can be explained in if/then statements, screenshots to demonstrate the sequence so programmer has an idea of how the end result should look. Consider . Programmer can perform a 'proof of concept' where he is also checking on his end mentally if it is feasible....then when programming can ask questions if bugs come up...

Roll out a prototype and then test to see if it conforms to what you requested. Remember the programmer is responsible for what you REQUESTED. Not whether what you requested ends up being profitable. And a decent programmer would do some sort of 'proof of concept' to see if it is even feasible to implement your idea. They are likely to get paid by the hour which may seem expensive but it is actually cheap in the long run as it encourages spiral development (incremental development vs fixed project where you supposedly state all of your wants up front; insta-fail when your wants don't meet your expectations).

Similar to a car mechanic, they will often warranty their work for xx months in case that with further testing it turns out that a certain function does not work under certain conditions (zero divide, platform upgrade breaking functions....[metatrader??]) So there is a LOT of tick data backtesting to triple-check functions work and SIGNIFICANTLY avoid costly results...

As your system becomes more complex, then you will need to ensure that new functionality you wish to implement does not break the old functions. So making sure that each new input or function is justified or earns its way into the algo is critical to spending your programming dollars wisely. I use the question "what new problem is this solving?" The larger question "How does this new function help reduce risk AND serve my customers [that I am making the market for] is for a new topic.

Most traders will NOT have the patience to iron out their observations and requirements to their programmer (or themselves) to see whether or not their idea can be programmed. Books like "The Four Agreements" and "5 levels of attachment" by Don Miguel Ruiz and family will assist the aspiring trader to lay down their ego and be more honest with what they want + the steps they are taking to get there....

@Scott Boulette, your observations are spot on as usual. Especially the mention that essentially there is only arbitrage, market making (value traders), or news trading as the primary speculative methods of profit. That piece of wisdom can save the aspiring trader many years of spinning their wheels. Trading and Exchanges: Market Microstructure, Ch 8 Why People Trade goes into more depth. Wouldn't you do this research of knowing the ins/outs of the business if you were opening a restaurant or retail store?

@Søren Lanng, the problem with visual traders is that it is based on templated If/Then statements. As soon as you run into a scenario where there is no predefined function that covers what you want, you must be able to dig into the code and add it manually. What if I don't want to trade on Thursdays from 17:00-22:00, but only if I am beginning a new sequence of trades. If I am in an existing sequence, continue allowing algo trades in that time until sequence is finished.

The 2) Is quite central - developing a strategy involves a high count of "tries" and adjustments - this is what kills a project (and the traders account if he pay a coder). Optimal is the trader do the "development", the testing of the high count of variants and adjustments - and leave the coder to add custom specialized inputs of data and criteria, which the trader use in his "development".

I see so many brillient developers, having used many years on developing sophisticated ideas and technologies - which could be brought into live life, by injecting these ideas and technologies into a framework.

@Scott...The mean reversion characterstic is what makes technical analysis and automation worthwhile. Overbought and oversold are descriptiops of excursions from a mean. The instrument can be modeled as having a drift and a stochastic component. Look at the similarities and differences between Orstein Uhlenbeck and Geometric Brownian Motion models.

Multiple cyclic drivers are governing a market. A point to recall is that EMF waves follow the principle of Superposition. Each driver has a drift (trend) and stochastic component. The instrument evolves around the drift. What is overbought for one cycle can be oversold in another cycle. For example, going short on an overbought level on a lower timeframe can lead to losses when the higher timeframe is at a oversold level. Specifically, being short for multiple days in the S&P 500 since February 2016 could lead to losses.

@Scott The weekly (drift, trend, mean) changed direction in January 2016. Succintly, the cycle is the drift, and an instrument travels to both sides of the drift, hence overbought and oversold and mean inversion.

Aside: The Ornstein Uhlenbeck model supposes a grand mean. This grand mean is the drift component of a higher timeframe.

When MULTIPLE drivers are at relative lows in their cycles, the move is amplified. Finding these lows before the fact is where the currently known Cronus Polarity keys come in to play. The known Polarity keys allow me to state the future (drift, trend, mean) a decade into the future. I found the first 2 polarity keys on March 5, 2015. I have been testing their efficacy since then. I found another polarity key on August 24, 2015, hence the long Spx call on August 24, 2015 at my twitter account. During January 2016, multiple drivers were at lows, which is one reason for my long call.

@Scott Aside: a smaller time cycle driver can dominate a higher timeframe for a short timeframe, e.g. the flash crashes and extremal events. These characteristic time periods can be anticipated beforehand and confirmed real-time.

For the philosophers here, the drift is the Fate portion and the stochastic component is the Fortune. Let's helicopter over the forest. An instrument is governed by multiple cyclic drivers. Each of the drivers has a drift and random component. The mean reversion is the total effect of all of the drivers.

You can trash all the theories about mean reversal, multi-timeframe cycles, quant-only development - dosnt work - if these math hypothesis were true, we would all be billionaires.

... but keep on using those theories, keep send money to the 10% who know how the markets work.

@John, One only has to look to how N dimensional spaces are formed and navigated to understand what you are saying is mathematically correct and I agree your analysis holds true for physical phenomena (think in terms of astrophysics in general).

However personally, I don't see the markets as following a predetermined course and therefore, I don't agree with the concept of a driver in the way I perceive you are using it. There are definitely influences at work but the term driver seems to imply there is more to the drift component and less to the stochastic.

FX models often use a variation of what you describe and a multi strategy internalizer will hand off and take positions to/from other components (often other time frames) in the same manner, so I would be hard pressed to say there is no precedence for what you are describing.

On the other hand at another level, you are describing my style of trading - directionally biased market making. Use the drift component on whatever time frame you choose to operate on as the directional bias (while accounting for the drifts within drifts within drifts to use your terminology) and let the stochastic component take you out of the trade. Of course you will have times when the random walk works against you, it is unlikely but not impossible to flip 20 tails in a row, but overall the stochastic component will work in favor of the type of trades I employ.

Another way of saying it is, enter on signal, exit in noise.

Hey guys. I appreciate the continued discussion. You've given me a lot of jumping off points for further research. I don't purport to know anything about trading at this point. Only software development, team process workflow, and agility.

Scott earlier asked me to start a discussion on agile methods, so I think I will do that soon. I've been busy lately so I haven't had a chance to start :)

While I can understand the physical reasons that take Scott in and out of a trade, I can see nothing but just statistics to support the MR idea in the form suggested by John. And purely statistical models are dangerous to trade because ten times you make a profit, then for the eleventh time you lose your whole account, but ON AVERAGE you're fine. A bit exaggerated, but hope my point is clear.

@Jon - As a representative member of the "client" class could I ask you to scroll back towards the top of the thread for a bit more context and then let me know how much of your "most secret sauce" you're prepared to divulge to your trusted programmer?

Everything but the "optimal" parameter set?

@Alex As you mention, trading on pure statistics without a plan for the "fat tail" will eventually take your entire account or at least put your time to recovery so far into the future that you are effectively out of the game. To be clear, the fact I understand John's point of view doesn't mean I agree with it. We have all seen 20 tails flipped in a row; one difference between those of us still in the game and those who are not is how we handled that situation.

@Jim - "everything but the optimal parameter set" is the least of it for certain algorithms. I will tell anyone how I trade, it is not remotely a secret. Knowing how I trade and being able to produce the volumes of code involved (correctly) are beyond the patience of all but the most dedicated. And in all likelihood I wouldn't be competing against my own idea for the several years it would take to implement. With a bit of work on my part, by that time I will be well into the next phase.

But that brings me back around to the original premise of this thread. The idea is often the easiest part of the entire process (not to diminish anyone's ideas) and frequently the "secret sauce" is well known or worse yet, a curve fit set of parameters that won't last a week in live trading.

Do any of these ideas that have a "secret sauce" component think about what to do with a position that is the result of an on the fly execution because a cancel crossed with the fill? How many of those situations will your programmer have to implement and how much time will that entail?

Hint - begin an exit process (often a sub algorithm in its own right) immediately; if your signal dictates cancelling the order, by implication, you don't want the position.

Jim wrote: "As a representative member of the "client" class could I ask you to scroll back"

This is a good point, and a major problem. If you as a trader have a strategy used for years discretionary with success, would you give it all to a programmer ?

I think not - from my experience, these traders start learning to program, and implement the strategy them self. They could split the strategy up in pieces of criteria, and use several coders.

There is a major difference from having an idea, just and idea - and a strategy used discretionary for a long time.

@Scott...Only part of an instrument's course is predetermined, the drift component. Consider that many seeds are sown in the Spring. The course of an individual seed is hard to determine, but the overall harvest of many plants is the expectation, where plants of diverse sizes exist. We call this seasonal trading in many commodity markets. The driver is what causes the seasons.

The influence of the driver is our reason for being trading. A drift component of zero gives us straight up Brownian motion. The size of the drift component is the magnitude of the time parameters in the system features/input variables. A relatively cool summer day can occur, but seeing 8 consecutive (drift, trend) daily high summer temps of 60 degrees in Florida is relatively rare. The stochastic component is what will make the instrument evolve around the drift, trend line. Think Orange Juice markets. One of the first MAJOR tasks is to determine the size of the time parameters in the algos.

@Alex...Regarding the Mean Reversion (MR) idea, the indicator/input feature is the representation of the instrument's realization. An feature/indicator's formula smooths/transforms the path. Consider the characteristics of a Simple Moving Average, MACD, and Stochastic. I am not talking about trading strategies. Maybe you could elucidate more clearly what I mean? I am being serious. NOT baiting. Thank you

We have alot of smart, knowledgeable people following this trail. You have just blown up my profile views number. Feel free to correct my incorrect statements. I take constructive criticism VERY well.

I am VERY slow. Please, bring explanations to the 5 year old level, except for the math. Thank you in advance

@Scott...I am not talking about a pure random walk. The stochastic component is only part of the puzzle. I am talking about the combination of a deterministic and random component. In the aforementioned models, the deterministic part is the drift, and the stochastic component is what encompasses the random variable. An example would be easier.

The flash crash of 2010 was a result of a very short term driver (3 day driver) experiencing an extremal value. What we had was a very short term burst of energy. This value was smoothed out by the longer term Cronus measures but was clearly evident in the (3 day cycle period). Did I see the 2010 flash crash beforehand? NO This is where I learned this lesson. Moral: A short term driver can dominate a longer term driver for a very short duration.

I formulated the 3 day Cronus measure as as as consequence of NOT seeing the 2010 flash crash beforehand. The 3 day Cronus measure had an extremal amplitude value at the flash crash, while the 5 day Cronus measure's amplitude was only slightly extremal. The 5 day smoothed out the extremal energy value, which the 3 day's amplitude shows. The 5 day Cronus measure is named the IntraDayJewel in my publications. Note, my publications are NON-COMMERCIAL, and Cronus is a STRICTLY deterministic, astrophysical measure. I discovered the Polarity Key for the 5 day Cronus measure in the early a.m. on 8/24/2015, which is why I published in the early a.m. and called for a Spx long real-time on Twitter near the end of the 8/24/2015 trading day.

Hey Scott...Thank you for letting me hijack your thread temporarily.

@Soren L...You state:"An optimization is simply an average - the longer back data used, the more average, and the less is the probability the strategy will work today."

Simply not true. Charles Isbell at GT taught a Machine Learning course that elaborates on optimization uses.

CS 7641 & 4641

Machine Learning

http://www.cc.gatech.edu/~isbell/classes/2009/cs7641_spring/

LOTS of lectures with notes to help you.

Stochastic Optimization with 2 of my profs at GT, Shapiro and Ahmed

http://www2.isye.gatech.edu/so/resources.htm

http://www2.isye.gatech.edu/so/resources.htm

http://www2.isye.gatech.edu/people/faculty/Alex_Shapiro/SPbook.pdf

This is a book on Stochastic Optimization.

You will see many different optimization algos in the Machine Learning class above.

Good Luck

@ John, the comment is based on my personal experience during 10+ years of development, not on someones theories. Technology is not static, much has changed since 2009, you link to something from 2009.

@John - Brute force development, mean reversal, curve fitting, and single chart development is outdated. Today is about BIG DATA across markets and instruments - you are going to be run over by technology.

@John B. Do you by any chance have any real money "forward test" results based on your "Cronus measure" available for inspection?

If so who did the coding, who did the backtesting, and what platform(s) were employed?

@John - check this link: http://www.ravenpack.com/

Which gives an indication where we are moving, there are many similar companies having been preparing the past years. This is now a tech race, where the prop trader and one man show will be run over.

@John, I understand the math pretty well (maybe not as well as you but enough for the practical application of it). When two opposing drifts collide, if they at very disparate levels, one will overwhelm the other and that will effectively obviate the weaker signal.

Fortunately the math works much in the favor of a very short term trader (me). Those larger collisions are relatively rare and can even be somewhat predicted. This leaves me with the equivalent of a Hurst exponent - I don't know where price will go but I do know where it shouldn't go unless there is a larger force at work.

Use stops that will protect you if there is a collision between your time frame and any of the larger time frames above yours and the Brownian motion (stochastics) will result in the profit targets getting hit with a significant enough percentage that making money is just short of inevitable. Will you have a bad day once or twice a year, sure, but you will have lots of pretty good days in between.

@John - two other points. First building on your seeds analogy - I trade at the lowest time frames possible so I don't starve waiting for the crop to come in and know very quickly that the crop has failed.

The second and most important thing every algo developer needs to keep front and center when designing a strategy - Don't do stupid stuff and if you fail at that, at least Don't do Incredibly stupid "account decimating" stuff.

@Soren L...See those military drones flying over your head? You are welcome. I can give you a personal introduction to the GT profs to help you help your country.

Regarding points 2 and 3, I am limited in the classes that I could point you, since GT profs are notorious at hiding their lecture notes. I wanted to give you the most recent of Dr. Alexander Gray, a prof at GT until recently in the College of Computing. Gray no longer teachers at GT. He is now head at SkyTree. http://www.skytree.net/

Yes, I understand that GT is ranked only 6th in artificial intelligence in the U.S., but they are engaged in continuous improvement. Although the lectures are 7 years old, the ideas are still relevant, just as is the linear ideas used in the machine learning optimization algos used at SkyTree. You give RavenPack as Big Data example, but you discount mean reversion and other ideas. I invite you to check out their white papers before you continue educating us.

@Soren L

http://www.ravenpack.com/insights/white-papers/tag/global-equities/?page=1

Aside:Statistics and Machine Learning have been my hobbies for 20 years. I am still learning both.

Regarding 3, I see that you are not a trader. Others have also stated that traders will be useless with innovations in technology. I recall what people said about the demise of traders with the introduction of the ticker tape. The need for traders is DOMAIN KNOWLEDGE. Think GIGO-Garbage In-Garbage Out Principle. I was taught to trade using only a barchart. My teacher told me that I would be weak until I could trade with only a barchart. This algo stuff is just gravy to me. Consider why a RSI chart will continue UP with prices going DOWN with both on the same timeframe.

@Jim H....You ask: "Do you by any chance have any real money "forward test" results based on your "Cronus measure" available for inspection?"

I got my first professional job managing money in 2002 for my GT teacher when he became my client. My family thinks that I am free labor. I presented Cronus at GT in 2002. I am a bit confused by your question. If you are asking if I have put boots on the ground ( USD) trading using Cronus, the answer is yes.

You ask: "If so who did the coding, who did the backtesting, and what platform(s) were employed?"

I do not understand what you mean. Cronus is SOLELY my Intellectual Property. You do understand that I have the past and future values of Cronus in perpetuity? Are you offering your services?

@Jim H

My friend, Philip York with Alt224, and I are working together, developing automated trading processes with Cronus. He is patient and understanding that I am slow.

Philip got to hear me call the Spx low for January beforehand, realtime in early January 2016. I told Philip another low was coming in February but not the exact date.

This is what the Cronus Polarity keys allow me to do. Dependent on the time parameter of the Cronus measure, a Polarity Key tells me the highs and lows of the Spx up to a decade into the future.

4/13/2016

Aside: Folks, be careful shorting the Spx now. What looks like a Change In Trend point is actually an acceleration point. An acceleration point is where the 2nd derivative changes sign. A flag such an example. During the first movement and continuation of the flag, the 1st derivative is positive, although the 2nd derivative changes to negative at the end of the flag pole and changes to positive at the end of the pennant. Lest I forget, a flag is due to the influence of a HIGHER timeframe's first derivative changing sign. Pay attention to the higher timeframe, unless you want to give me your lunch money. Quantitatively this means pay attention to the double cycle and higher.

John B - If you have read this thread from the start then I'm afraid I fail to understand why you fail to comprehend my questions. In an attempt to clarify:

1) No - I am not "offering my services".

2) I will however attempt to sell you a tee shirt. Slogan © Philip York:

http://trading-gurus.spreadshirt.co.uk/ray-the-random-robot-a-new-dawn-A16337185/customize/color/1

3) Is your "boots on the ground" equity curve "available for inspection"? It sounds as though that equity curve, such as it is, is not fully or even semi-automated as yet?

4) Am I correct in concluding from your comments that Alt224 are currently doing some coding for you? If so which algo trading platform have they recommended, how is the backtesting coming along, and how is your "relationship" progressing?

5) Does Philip know all your "most secret sauce"? If not, where's the dividing line?

6) At the risk of drifting off topic - How large are your SPX profit targets? How tight are your stops?

I’ve never seen/witness any off-the-shelf neural algorithm that learns a specific past period and continue successfully trade with the “learned” parameters in forward tests or live trades. Coders always force them to re-train the near future data to handle the new market conditions. I don’t see much difference in the “approach” of a continuous optimization of conventional algorithm and neural algorithm. Don’t mix up with neural engineering applications since the data set is extremely different and usually learnable.

Great T-shirts JIm, would that be for Ray The Random Robot I or II ?

Thanks Søren - The tees depict Ray Sr. - Ray Jr. utilises non-random entries. He is slightly less stupid than his pater.

@Ersin I have your same opinion regarding not seeing much difference in the two approaches; actually I agree with your entire post.

@John, can I infer from your reference to 1st and 2nd derivatives that you have a function to work with (that is a serious question).

Alex Krishtop Scott Boulette The key to using statistical models correctly is to a) model them after real world scenarios (walking forward with tick data preferably) and b) apply them to real-world situations. As a market maker of sorts, you need provisions to handle unexpected volatility in real time. I think Scott realizes this, especially if doing HFT. You cant trade statically (one grid fits all) to cover all situations. This is where real time, coincident indicators like APAMI helps. You make real-time adjustments that would reduce your risk.

Not much different from driving a car. If heavy rain or sleet suddenly appears, do you continue to drive the exact same speed as with clear skies and dry road? Unfortunately, a template-based algo builder is unlikely to handle this kind of sophistication.

BTW Jim Hunt, that is the primary reason why Ray the Robot failed; No provisions or poor provisions to handle real-world volatility.....

Jim Hunt To answer your inquiry about "secret sauce" or trade secrets....well all companies have to deal with this on some level, especially the military. 2 things to overcome: 1) understanding that what you teach is not 100% proprietary. Your style is yours, but the underlying concepts are not. There is an element of trust and the actual undertaking that makes a professional trader....not solely technical knowledge of trading. You need direct market access, pool of client funds large enough to make it profitable, WILLINGNESS to take risk/losses (responsibility), etc.

2) Prevent a "Zuckerburg" or facebook-like usurp. The key to this would be to have regular source code updates. When you pay a programmer by the hour, you should get this automatically. Even when you pay by the project, if you are paying in installments, you should also get some source code. Part of the supervisory role of an employer is determining the level of trust granted to workers.

Jon - Did you read Ray's tee shirt? He was cunningly constructed to be as artificially stupid as possible. He purposefully lacked all prudent real world provisions.

Nevertheless I was contacted by someone who had tested every MT4 EA they could either buy or download free of charge on a live micro lot account. I got the call because on an extended test Ray was one of the best performers of the bunch.

Q.E.D?

Both Ray I and II are actually performing quite well - most important is not to lose the account, which it seems Ray the Robots are not doing.

Jim Hunt I would expect ray to perform reasonably well for a time. It is using mean reversion / cost averaging which is a natural part of any derivative market. It's too bad you you kept ray stupid, saw that even in his stupidity he worked well for a while, but made insufficient attempts to see WHERE his stupidity failed him.....how to make him avoid the same errors.

A big part of the engineering process is looking at [potential] points of failure and then asking "How can I make this product/service more resilient?" Fortunately, tick data backtests speed up this process 1000% with very realistic market simulations and much less risk on real capital. You don't have to crash real (or demo) accounts 100 times. In fact, similar to commercial pilots, the stakes are too high with $1MM+ equipment and livelihoods at stake. You obtain a certain confidence level at the simulator level FIRST, then step into the live environment.

Not to get too far off topic; hope this helps someone.

@Jon - you and I agree on the idea of when you can use probability models and we also likely agree on when you should not. I know many are fans of pure data mining, I just don't happen to be in that group. As you mention, there aren't that many purely new ideas around and those that occur to one person may well have occurred to others that are not publicly discussing them.

Søren - Ray Sr. did eventually blow his account. Long after most of his competition though!

http://trading-gurus.com/ray-the-random-robot-commits-suicide/

There is however (almost) no chance Ray Jr. could do the same. Non-martingale money management!

Jon - Your advice may help someone else, but I fear it's of no use to Ray Sr. and I. He had two purposes in "life" during his all too brief existence.

1) To give non-programmers a bit of a clue about what's involved in programming an algo.

2) To demonstrate to anyone who might be interested that the vast majority of MT4 EA's are in actual fact even dumber than Ray, hard though that may be to imagine!

@Jim H

6) At the risk of drifting off topic - How large are your SPX profit targets? How tight are your stops?

Depends, you are looking at 200 Spx points now from a Trade initiated for January 2016.

https://twitter.com/CronusMaster

I first saw this trade on March 5, 2015. When I discovered the Polarity Keys for 2 Cronus time cycles. Folks, when I say cycles, I do not necessarily meaning trignometric cycles.

@Ersin D...you state, "I’ve never seen/witness any off-the-shelf neural algorithm that learns a specific past period and continue successfully trade with the “learned” parameters in forward tests or live trades."

This is simple. The Garbage In-Garbage Out Principle is at play. When people, who cannot trade their way out of a paper bag, are designing systems, the results are quite obvious. Domain Knowledge is necessary to learn one way or another. Automating learning domain knowledge is not as easy as first thought. Nothing yet beats real-time screentime.

@Scott...Yes, I have a function. The Cronus measure is a function. Here is an example of what happens. Suppose 2 or more cycles with a longer time parameters form a bottom. Another cycle with a shorter time parameter forms a top. The influence of the multiple longer time cycles dominate the smaller cycle; therefore, an acceleration point forms, instead of a change in trend. This is what happened in January 2016 to the Spx. See all of the formed flags and currently forming flags? Through January and February 2016, the Spx formed a hammer and a spring. The January low was the hammer, and the February low was the spring. Note, a hammer is a flag on a smaller timeframe. This was caused by many much larger cycles changing trend. Some of these cycles last formed a top in August 2014. You say the Spx formed higher highs after August 2014. This was AND is a harbinger of things to come. August 2014 through recently was an acceleration point of a much longer timeframe.

@Scott

When a long timeframe cycle forms a high and the market continues UP, WATCH OUT for the party will continue in force. George Lindsey calls this an Ascending Middle Section. Many Thanks to Ed Carlson for hooking me up with his book. I wrote about this on my twitter in August 2015 Folks, the Spx has more high volatility coming later this late Summer/Early Fall 2016. The Spx formed part of a long term cycle from 2000 through 2009, which is why the Spx has been screaming higher with such verocity since 2009.

Note, the top in 2000 to the bottom in 2009. Can you see the bigger pattern? The shorts failures from February 2016 through mid April 2016 have been a direct result of the much longer term cycle reversing to UP. Look at how the market based indicators have been evolving in relation to the Spx at the so called short points from February 2016. This is what happens when longer term cycles step into your puddle.

The original Cronus formulation's motivation was to foresee high volatility and changes in trend points in ANY market. Took me 14 years to understand why I could not definitively state whether these critical points are market highs or lows with the original Cronus formulation. I learned that I had discovered a novel time series analysis technique, which is applicable to any time series. Yes Jim H, this technique is another of my trade secrets. I asked one of my GT profs about publishing it in a journal, and he recommended keeping it private. For those interested, check out Steven Strogatz at Cornell. With insight from Strogatz, I formed the 3 independent market based features.

@John, what you are describing is in effect a very well known FX model. Check out Thomson Reuters IFR and you will find it prominently displayed. It uses vastly different terminology and likely very different calculations but has the same end effect.

Knowing the model quite well and having a decent math background (multiple signals differing in phase, etc. are pretty standard EE concepts) allowed me to immediately understand the points you are making. I do have one real question; I won't have much or any reaction to the answer but am truly curious...

@John

Did you choose the name Cronus as a tongue in cheek way of seeing who is actually paying attention or is that really the name of your system? For those who don't know - Cronus was the son of Gaia, the personification of the Earth and Uranus who represents the sky (among other things). The parents' names (especially the father's) give me cause to ask.

Who can compete with solar, lunar, etc cycles, seeing the markets as having discoverable, observable price functions and being able to effectively say all it needs to work are the correct parameters? Given the terminology you use, it is easy to imagine the possibility you are having a bit of fun. I don't sell things so I tend not to name algos beyond what is necessary to organize them in my code base but I also recognize the need for having a sense of humor in this business.

My comments are all in good fun so please don't take offense if you are, in fact, being serious.

@Scott..Hey Scott, thank you for the link to IFR. I am willing to bet that Thompson Reuters IFR is not a deterministic, astrophysical measure. Can the guys at IFR tell me whether the U.S. long bond will experience high volatility in 2026. I can.

http://mathworld.wolfram.com/Deterministic.html

http://mathworld.wolfram.com/MeasurableFunction.html

Cronus is a deterministic, astrophysical measure not a system. I first saw the Spx high volatility for 2015 in the year 2005 A.D.. I made the high Spx volatility statement on my twitter on from Cronus8, which was first published in 2005.

@Scott

I named the measure Cronus because it is particular representation of the Space-Time continuum. Time is measured with changes in space, as in the definition of a second with the Cesium atom. Essentially, changes in gravity creates changes in magnetic fields, which creates changes in electric fields, leading to magnetohydrodynamic interactions.

https://www.cfa.harvard.edu/~namurphy/Lectures/Ay253_01_IdealMHD.pdf

A couple of reads. Scott, Taken's Theorem is relevent too. I understand a bit about EE too, more so from the math side.

http://dc.etsu.edu/cgi/viewcontent.cgi?article=3965&context=etd

http://www.complex-systems.com/pdf/08-1-2.pdf

Take a look at the stable dynamic system on P.3 and revisit the quasi-periodic papers that I pointed out earlier.

http://kodu.ut.ee/~swen/random-stuff/helsinki-time/presentations/takens-slides-2004.pdf

@Scott

Cronus is otherwise know as Father Time. I developed the Cronus models independently from any market. For example, I had a Cronus model sitting on the shelf for 4 years before I found its relevance. It goes with the U.S. long bond. In March 2012, A friend asked me if I had a Cronus model for TLT. I said that I did not know and would have to look. I found multiple critical points in this Cronus model occuring at one time in March 2012. I performed 15 minutes of technical analysis on TLT. Published a long call on TLT as a test. This long call yielded 14% unleveraged in 2 months. I do NOT have the Polarity Keys for the U.S. long bond yet.

You state:

"Who can compete with solar, lunar, etc cycles, seeing the markets as having discoverable, observable price functions and being able to effectively say all it needs to work are the correct parameters?"

@Scott

Market-based historical transforms are price functions in a sense, although a problem is that most are collinear, which means that they bring little if any additional information to bear. Almost surely, the time parameters in the historical transforms are not randomly chosen. A problem with these historical based transforms is that finding the values past the present time is EXTREMELY difficult, meaning the price chart to the right of present time is not realized. I have the Cronus values in perpetuity to the right hand size past the present time. When I was describing the cycles in yesterday's post, I was describing Cronus, not historical market-based indicators. The price function is Cronus. The task is to learn the relationship between Cronus and the historical market-based transforms. The Polarity Keys allow me to definitively state the future highs and lows of an instrument.

@Scott

At the current time, the Polarity Keys are qualitative, so I am able to state the future highs and lows of the Spx for only a decade forward. I have been testing them out on my twitter profile since 2015. I am trying to learn the mathematical Polarity Keys, such that I can state the future change in trend points indefinitely forward and on multiple markets. I discovered another Polarity Key for a short term Cronus cycle at 3 a.m. on August 24, 2015, and I published it at that time. On August 24, 2015 at 15:24, I called a Spx long at 1884. The Spx dropped 103 points intraday on 8/24/2015. I called an Spx short at 1921 on August 25, 2015 at 14:46, which the Spx proceeded to close at 1867 or 54 points lower. Being transparent, the 8/25/2015 Spx short was based on trading skill alone. You can verify at twitter.

https://twitter.com/CronusMaster

I had been looking at the Spx low for August 24, 2015 since March 5, 2015 using the Polarity Keys.

Again, Thank you

@John - It would take quite a bit for me to see the markets as deterministic because that would in a sense obviate all free will in the entire Universe.

When you mention discovering a Polarity Key at a particular time - how did you know you had discovered it? What made that instant in time different from any other and how was that evidenced?

How many turns would you have to call to move beyond theoretically random? When CNBC says "this guy is brilliant, he called the last three whatevers correctly" it always makes me cringe. Given your math background, you likely know full well the small sample it would take to find someone who called virtually anything correctly three times in a row.

Then there is also the obvious question - if you know the SPX "only" a decade forward, why are you not on an island somewhere with enough money in the bank to buy LinkedIn itself? Again, I am not making fun and not playing gotcha, I am just genuinely curious.

John - in which timeframe do you operate ?

@John B - More questions in partial answer to your answers to my questions:

3) Zero

4) I take it that means Philip prefers his own platform to the one you subsequently gratuitously introduced?

5) Have you ever considered "monetizing Cronus" by trading it?

6) You've taken profits already? How many contracts? Where was your stop?

7) No doubt off topic, but I'm intrigued by your "weather forecast". Has "MUCH LATER" arrived yet?

@Scott

Use a RUNS test

http://www.itl.nist.gov/div898/handbook/eda/section3/eda35d.htm

Since Cronus is DETERMINISTIC, the sample size is small to alleviate my concerns. We are no longer talking about a purely stochastic process for the output (prices). I have also looked at the 1920s too. I have ZERO interest being on tv.

@Scott

you ask:

"Then there is also the obvious question - if you know the SPX "only" a decade forward, why are you not on an island somewhere with enough money in the bank to buy LinkedIn itself?"

The old Cronus formulation allows me to see any extremal events with certainty in perpetuity. The decade is related to the polarity key for the 80day. I believe that it chains to go farther into the future. However, I am searching for a strictly mathematical polarity key, such as through symbolic regression. This would allow me to see in perpetuity. To answer your question, 2 words are neccessary, Research and Development. R&D comes at a

cost. I cannot trade and do R&D at the same time. I was asked some very difficult questions in 2009 by a fellow with 10 digits worth. He initiated contact with me from Mark Jurik's group. Most here would think that he asked me crazy questions. I was again asked similar questions by a global head.

@Scott

Again, I had to answer no to him too. The first time that I was able to answer some of the questions affirmatively was March 2015. We have a HUGE event coming before the end of this decade, which makes 2008 look like a molehill relative to amplitudes.

Second, I have shortcomings too. I cannot do everything alone. I truly understand the idea of comparative advantage. I need guidance and leadership too. Now, I am shopping for a team with a strong leadership model in place. I have a very good map, and the map can be implemented in many ways. This is the Execution part of the ACE method. I have a much better understanding of what I do not know, and I am seeking guidance in these areas.

@Jim H...you ask:

"4) I take it that means Philip prefers his own platform to the one you subsequently gratuitously introduced?"

Yes, his group has developed their own platform.

"5) Have you ever considered "monetizing Cronus" by trading it?"

I first traded Cronus in 2002 after I presented it at GT. My GT teacher became my client at the end of the semester.

In 2011, one of my contacts asked me about my intentions of monetizing Cronus. So he actually planted the idea. He means on a large scale. I failed to understand what he meant at first.

"6) You've taken profits already? How many contracts? Where was your stop?"

Do you see an exit for the trade on my twitter yet? NO Some of the trades on the profile are daytrades too. Feel free to verify the time and price stamps. I see a potential for the January 2016 trade around 1870s to yield 1600 Spx points. I can see a projection to this area.

"7) No doubt off topic, but I'm intrigued by your "weather forecast". Has "MUCH LATER" arrived yet?"

Not by a long shot, this is a pattern that I have found reliable since the 1700s. I first recognized this pattern in December 2012. This pattern arrives twice a generation. This pattern also allows for timing of depressions. The blip in 2007-2009 was a Depression not a recession, and they have happened like clockwork since at least 1700. Read about the Depression of the 1890s and the similarities to the last one.

@Soren L...I have operated down to the one minute timeframe. Why?

@John, to provide the missing information - "have operated down to one minute" dosnt tell much. Your interesting talk about patterns and cycles, in which timeframe do you see these patterns ?

Correction to "Look at Scientific American's December 2015 issue "Wacky Jet Stream"" Make that December 2014. of Scientific American

@John it sounds like you are describing a Markov Chain but I am a bit confused. If it is deterministic and if I understand you correctly, you are implying you have a closed form function, why the need for R&D? I can see the need for computing power but it would seem that the function itself would be self describing (or what is sometimes referred to as self referential) in a way that would allow you to trade directly from it.

Quite possibly I am missing something basic in what you are saying. Also, I am not sure I was clear in one of my earlier questions to you. If you did a simple null hypothesis test, how many instances would be required to disprove the null hypothesis and have you done this or its equivalent. I saw you mentioned a runs test but I am not sure that would be an appropriate measure for this situation.

233 replies and we are still nowhere. Perhaps the best way to answer the original question posted: "Why am I not interested in your most generous offer?" would be simply to say: I am just not interested, please find someone else to help you out. There is not need for scientific validation or anything else, this is a matter of personal judgment and nothing else.

@Oscar - my original intent was to explain why few if any trading professionals are interested in the latest "disruptive" idea. Years ago I read a book by Al Brooks on price action and he had a statement in the book that I found interesting but didn't fully comprehend at the time. He said, he sometimes would offend people because they would approach him at industry functions with a "brilliant new idea" and he would politely explain that he hadn't even fully explored all the nuances of his existing trading let alone his notebooks full of his own new ideas and he had no time for a completely new approach.

@Oscar

All this is not to say professionals like him don't want to have discussions about the market in general or even specific aspects of trading and algorithms, just that someone with a couple of weeks thought about a "new approach to the markets" shouldn't expect that 1) it is in fact new, 2) that their approach could supplant the trader's existing approach and most importantly 3) that the trader would be so taken by this new approach that he would drop what he is currently doing (including trading) and form a partnership to split the proceeds of this new idea while doing 90% of the work required. It is somewhere between common sense and are you kidding me.

@Scott, you have asked about hypothesis testing of the signals. Let's keep in mind that the signals are generated by changes in celestial bodies position in space, which eventually lead to energy changes on Earth. In roughly 2 billion years, human life on Earth will be impossible without an energy shield because the sun's sphere increases with time. Cronus is simply an energy measure. Nothing more, nothing less. I presented a simple explanation of the mechanisms of how the signals are generated in an earlier post. The one with the magnetohydrodynamics. Note: LOTS of handwaving here. First, large astronomical bodies change position in space. This change in space creates gravitational waves. Read changes in gravity. These changes in gravity create changes in the characteristics of stars. Do you really think that

the 22 year solar polarity cycle is coincidental in magnitude, folks? The changes in gravity creates coronal mass ejections CMEs. Read plasma.

@Scott Read magnetohydrodynamics. Think changing magnetic field caused by the changing

gravitational fields...Creates CMEs. Changing magnetic fields create changing electric fields. RCA Labs did studies on this in the '40s, especially important before the advent of the satellite.

Scott, regarding your question. You may be thinking about causality. Judea Pearl has done seminal work in causality studies.

http://bayes.cs.ucla.edu/BOOK-2K/neuberg-review.pdf

http://bayes.cs.ucla.edu/jp_home.html

Also using graph theory. Multiple hypothesis tests are available to ascertain whether an actual DETERMINANT has been found. Also tests for causuality exist, such as Granger.

https://support.sas.com/rnd/app/examples/ets/granger/

http://schwert.ssb.rochester.edu/message.pdf

@Scott..You ask about R&D. This knowledge was not absorbed through osmosis. Take a look at some of the articles that I presented. The one on quasi-periodical cycles is dense. Multiple nobel prize winning ideas are embedded within Cronus.

I figured that, since these people won a Nobel prize, the information is highly valued. My task was how to apply it after I figured out what they were saying. Another idea: How is the Fractal Market Hypothesis related to social networks of Watts? How can this information be reduced to a single value? Most floor traders could explain how if they understood the math.

@Scott

Now, I am focusing on trying to figure out how these 3 independent historical indicators represent market regimes and defining the phase changes between regimes. Easter egg: only 4 regimes exist. I have hidden a few easter eggs in these posts. This one was just point blank. These 4 regimes are also exhibited in the Vix. Second, I am searching for the mathematical Polarity Keys for Cronus. Cronus R&D: I already know a few different quasi-periodical cycles (QPC) and which ones goven which markets. Different QPCs govern different markets. Read the US long bond Cronus measure in a previous post here. I start with the QPC. Find which market they govern. Find the appropriate time parameters for the respective market. Think overshoot and lag issue. Find the Polarity Keys for the particular market. Find the relation between the Cronus measure and the market.

@Scott

Here are some "crazy" questions about the Spx that I have been asked. What will be the trend in the 2050s and 2060s A.D? Cannot answer, except for the depression and when will occur. When will the current market top out in this decade, relative to the recent economic cycle? Can answer. What will be going on in the 2020s? Can answer out to 2026. After 22 years, I can finally answer this question.

@John, it will take me some time to fully go through your posts and digest the information in them but I do want to share a simple thought although it is something you have already considered.

In some ways what you are describing is N dimensional space and your difficulty in some areas of alignment may be at a high enough dimension that the convergence appears in an area (dimension) you aren't looking at i.e. you are looking at point in time t in the SPX whereas the event is occurring at t1 in a completely (yet weakly correlated) market (or even some other type of event like a political one).

If you think of traffic cones being stacked such that they can seemingly overlap in physical space at the lower dimensions but in fact do not due to their hollow nature. While the proximity may not be obvious on the surface, at higher dimensions, the connections are there. Events can operate in the same manner. This isn't a great analogy but it is the best I can do in the limited space here.

@Scott....Here are some articles

first, on your action at a distance...first exposed to this idea in 1994. thought that it was garbage, until I saw it applied firsthand.

http://www.nist.gov/pml/div686/20151105loophole.cfm

http://www.nature.com/news/quantum-physics-what-is-really-real-1.17585

second, future affecting past.

http://secondnexus.com/technology-and-innovation/physicists-demonstrate-how-time-can-seem-to-run-backward-and-the-future-can-affect-the-past/

http://www.nature.com/nphys/journal/v11/n10/full/nphys3414.html

Would have not believed it until I experienced it firsthand. This time with a loss. Made me a believer in a heartbeat in 2009.

Aside:

Political events have no bearing on markets. J.M. Hurst shows this to be true in "Profit Magic of Stock Transaction Timing". Also, I predicted at GT on 8/23/2001 a 25-30% Spx decline for September 2001.

The jet stream article is actually in December 2014, "Scientific American" I verified today.

http://www.scientificamerican.com/article/a-wacky-jet-stream-is-making-our-weather-severe/

Some comments - I say in this high timeframe of stocks, you would look elsewhere to try predict the future - you cannot use math on the stock market in this timeframe, where it 60% depends on the macro economy, 30% on the surrounding market, and perhaps 10% on the stock chart you look at.

Check this conceptual graph illustrating what you can do when you use your head instead of math ;-) http://www.tickcom.com/examples/stocks_14_12.png

Besides, you cannot use math on daily bars, to few data to be able to conclude anything.

As to the dimensions of timeframes theories, if you are in a low timeframe, the higher timeframes are lacking data, well above your stop loss and should be used with great care, since using information from higher timeframes may disturb your core strategy.

@Soren...You state timeframes. Which timeframes do you refer? Also, you are aware that markets are self-similar?

Here is a task. I present you a market chart with timestamps removed, and you tell me which timeframe that it comes. You can do this? The current understanding is that this task is not yet doable, given the self-similarity characteristic.

You state:

*

you cannot use math on the stock market in this timeframe, where it 60% depends on the macro economy, 30% on the surrounding market, and perhaps 10% on the stock chart you look at.

*

Far from being true. Perception generates pricing. You understand that the stock market leads the macro economy by 6 to 9 months? The economy does not generate pricing. Consumers choices are dictated by perception, which dictates pricing.

@Soren

you state:

*

As to the dimensions of timeframes theories, if you are in a low timeframe, the higher timeframes are lacking data, well above your stop loss and should be used with great care, since using information from higher timeframes may disturb your core strategy.

*

More statements that require much more forethought before declaring true. The higher timeframes dominating the lower timeframes is where core strategies of lower timeframes are disturbed. This is vital information. Been there, done that, and got the t-shirt. Read loss. Look again at what was being said on marketwatch in January 2016. I posted some examples on my twitter. What people are currently seeing is the influence of a higher timeframe dominating.

https://twitter.com/CronusMaster

Another example is the May 2011 Spx high. I did a Cronus publication prior to that Spx short post.

@Soren

A couple of key questions is: where is the longer term cycle in its period? The critical times are when they are changing trend (CIT). Will the CIT be dampened?

http://www.entropy.energy/scholar/node/damped-harmonic-oscillator-energy

Consider why the dampening is occuring.

Folks, markets exist in a conservative field. Consider the significance of this statement. I discuss this in more detail on Jurik's yahoo group.

Folks, I see that some of you have me on your twitter lists. I have been known to restrict access at extreme vol, which means that only followers can see my posts not lists.

@John, if a higher order cycle influences the cycle I am effectively trading, I don't need to know unless my exits are so large they are affected by the higher order cycle or if the cycle definition is so precise (a point vs an area) that I know I am going to have to cross it to exit. Other than those two situations, my algos are not of a type to be affected.

There is a simple axiom that works in trading just as well as it does in many walks of life - micrometer, grease pencil, ax. Never measure more precisely than you can mark and don't confuse the precision of your marks with the precision with which you can execute. Back in the 4th grade (or thereabouts), we learned that multiplying a number with 2 decimals by one with 3 decimals technically gives you 6 decimals but you still only have 2 decimals of accuracy.

@John - it is a scientific fact, that you do not have sufficient empirical data in such high timeframes to reach any other conclusion, than you do not have sufficient empirical data.

@Scott...Please, clarify and color your statement below. Thank you.

*

if a higher order cycle influences the cycle I am effectively trading, I don't need to know unless my exits are so large they are affected by the higher order cycle

*

@Soren L.

You state:

*

it is a scientific fact, that you do not have sufficient empirical data in such high timeframes to reach any other conclusion, than you do not have sufficient empirical data.

*

You are speaking my language now. You state scientific fact. Where is the proof? 2 simple words that I heard often in grad school. Prove It. I am presuming that if you cannot provide a rigorous proof of your claim, you can provide a link to a scholarly article available to read, as I have been doing. I want to see a proof.

Some general comments:

I have nothing to sell anyone and I am not connected to any internet site that wants to sell anything. I do not have a need or want for someone to “help me”.

I have well over 10000 hours of screen time studying the markets.

My approach to completely automated strategies (trading both long and short) uses heuristic methods and algorithms(not indicator based).

I can work from a single tick or a single second. (Intraday strategies) that go home flat every day.

I do not use Stop Loss/Profit Targets what so ever.

I never add to a losing position.

I am not in favor of be exposed in the markets over 25% of available time to trade the instrument.

My methods are only suitable for me and not for others.

see next post

Soren has said “ I say in this high timeframe of stocks, you would look elsewhere to try predict the future - you cannot use math on the stock market in this timeframe, where it 60% depends on the macro economy, 30% on the surrounding market, and perhaps 10% on the stock chart you look at.”

I agree with this statement for the most part. I would like to further define the statement with my personal opinion.

Any tradable instrument in the markets that is worth trading for a profit (intraday) is closer to 90% about the surrounding markets and about

10% to the instrument being traded.

When using data from several time frames the information flow is typically from the higher time frame to the lower time frame.

I use the very opposite approach.

When your personal approach to trading the markets is well outside the typical approach , then there just is not any scholarly articles available to read.

just my thoughts,

all the best

@James H...

You state:

*

I do not use Stop Loss/Profit Targets what so ever.

*

This statement means that 100% of your trades are profitable by definition. An exit of any type with a loss is a stop loss.

@John, yes, in the strict definition you would be correct. A better description of my approach(model) is that my strategies are signal based across

several instruments and the amount of loss or profit (per instrument or portfolio position) is not part of the algorithms. Simply said, “it is what it is”.

John, please consider that when I first came into this forum and posted some information that I was very quickly “labeled” as a fraud or even worse.

all the best