I think that there are many quants who understand the limitations of their models, myself included. For example: as much as I would like me to believe that foreign exchange rates evolve according to some long-term mean reversion to PPP model, I know that my country (South Africa) is facing a massive debt crisis and is negatively effected by the slowdown in the Chinese economy. Therefore, it is likely that it will follow Brazil and Russia in a ratings downgrade and the currency will devalue sharply. However, my model tells me that our currency is highly under-valued already according to PPP and I should buy it. Instead I am selling it. It's obviously anecdotal, but I think many quants think this way and do take into consideration extraneous factors.

Stuart, I hear what you say. Indeed, I personally know some quant traders who moderate their own numeric models with observations of this kind. However I am more interested in hearing an answer to a bit different question: what does prevent you (well, not "you" personally, but any quant) from formally describing and then trading these very "extraneous" factors which force them to violate the rules of their own systems? In other words, why do they (quants) prefer to create abstract models trying to include all possible factors which affect price movements instead of splitting this complex task in a number of smaller ones, suggesting simple algos for certain particular market processes?

When volatility increases or some exogenous factor is introduced, the model fails and they have little choice but to abandon the model's decision.

When volatility increases. May I ask you — do you really believe that "volatility" is a self-existing entity? Is it a creature with its own free will? Maybe a God's will? Isn't volatility only the result of a certain process in the market? Why do you state that when volatility increases then the model fails — and what if this model does include certain reaction to changes in volatility?

► Quantitative Easing is one of the economic policy measures that have central banks to try to revive the economy, basically it is to inject money into the market by buying assets.

.

but lately many people talk of QE and I think it is important to explain that

First of all comment that the Quantitative Easing is a measure that Europe is bringing much controversy since the Bundesbank will make your hair stand on end just to hear these two words, is totally contrary to implement these measures, considered desperate measures they can be bad for the economy.

What is Quantitative Easing?

It involves injecting money market aggressively to revive the economy. This is buy the banking assets for a more favorable deposit institutions. Assets that are often bought government bonds or mortgage bonds. To buy a price, therefore, money is injected into the system and increases liquidity for banks to lend money to businesses and families, causing euphoria on the stock is established.

When you have to apply a Quantitative Easing?

The Quantitative Easing is one of the last measures of central banks to stimulate the economy. It is usually applied when:

The economic growth is zero or low inflation risk estanca.- deflacionista.- Interest rates can not go lower.

Applying a Quantitative Easing has two negatives, rising inflation and depreciation of the currency (the Euro in this case)

Europe, late and wrong

Let me comment that the United States applied its Quantitative Easing in 2008 (it has pumped into the system 3.15 billion) and this year the Fed has withdrawn its stimulus, the economy has improved dramatically and now, nine years later,

► It is very risky, because the central bank are constantly wanting to raise rates, and simultaneously causes the affected investors deviate currency, and even worse if injected #QuantitativeEasing Program.

* CAUSES TO EURO COINS fall back, As this inverted billion in assets. Break to investors.

That's what sold it visible over the last five years, since the US, has applied, so it is with the euro. The only beneficiaries are FED.

Alex - volatility is the result of a given "process" however the variables/coefficients used in modelling do become highly unstable as vol increases. You can look at it this way - as volatility of your model variables increase your estimates for these model variables (I use the word estimate because these are usually unobservable values) are less reliable. There is greater probability that your estimates are incorrect.

Even if your model directly incorporates changes in volatility, volatility is unobservable and will always suffer from estimation errors

Alex - using range as a measure of volatility is one approach. Volatility is typically standardized (technically volatility refers to annual equivalents) and I usually work with implied and realized measures (option implied and standard deviation). If I understand your example correctly, you cannot directly compare a daily range with a 20 day range. It would be similar to directly comparing daily and monthly standard deviations - you would be comparing different units of the measure. That's why, unless otherwise stated, volatility refers to an annual quantity.

Actually, volatility can convey quite a bit of information on its own - especially if you work with semi-variances

Antonio, I'm glad that finally we've started from where we all should have started: from definitions. There are plenty of definitions of volatility, I agree. The question is: which of them are meaningful in terms of the market's processes, or, if you want, which have any economical meaning, and which make only mathematical sense, irrelevant to the market. With your definition of volatility, which market process(es) does it represent?

Alex - I view it simply as the market's repricing of risk premia. I prefer measuring this reprising using realized and implied. It can be difficult to compare ranges for different asset classes and even for a single asset class on a time series basis. Ultimately, this repricing should lead to more efficient capital allocation decisions. For example, as equity volatility increases, the cost of equity (capital allocation decision) should increase, thereby lowering the share price.

This would be the economic explanation for what is currently occurring in equity markets. The equity value of firms is being re-priced via the equity premium (risk premiums have been extremely mispriced over the past several years)

Antonio, your definition is understandable, but it is a bit too general: you treat the market as a self-sufficient entity ("market's repricing of risk premia"), and therefore inevitably come to a bit too general conclusions, like "risk premiums have been extremely mispriced". However risk premiums cannot be mispriced on their own; there's no absolute fair value in the market. It only exists in one's model.

very well said :)

Alex - an equilibrium level does exist and is determined by the market. It is another question whether one's model is able to determine what this equilibrium level is. Yes, I did provide a generalized definition - I do this as a profession, so I don't dive into specifics that are tradeable on threads. I'm sure you can understand this :-)

I recently published a short article and chart to my profile which illustrates one method in which I measure these repricings. It is meant to illustrate that you can create models that capture these re-pricings for actionable decisions. Of note, the VIX is down more than 2.5% since I published the post, with the added benefit of having a few days to properly position your portfolio.

Well ! Depends upon alot of other things that are going on also, what makes a trader decide whether to trade intra day and take little profits and trade with tight stops or to give trades a little bit of wider breadth and hang on to them over night. I trade E-minis and here is a perfect example of all the indicators that I use in were in a solid buy mode plus a rock solid entry in March e-mini futures at 3:00 a.m Thursday morning, it was certainly a keeper untill when the market closed today. Its pretty simple "trade what you see ,not what you think"

Zehra, and what does make you decide whether to trade intraday or "to give trades a little bit of wider breath"?

Antonio, what I'm trying to gently suggest you for consideration is an idea that since "equilibriums" and other sounds-like-obvious terms exist only in someone's models, then depending on the degree of importance of this someone to the market we may or may not create very robust models — simply understanding what this important someone is going to do in certain situations. Looking at this angle, why should be overcomplicate the mathematics in our models? Isn't it wiser and simpler at the same time to start from the opposite end — from learning the market itself?

Alex - the mathematics in my models are on the simpler side - I have seen complex models that are elegant and conceptually attractive, but completely fail in practice. I do believe simple is better and completely agree that understanding market structure is extremely important. That is one of the reasons I like to work with implied measures - the market is telling you what it believes the "equilibrium" level is (or another measure/estimate) - this will partly be the result of market structure.

Great article Alex - spot on. Maybe there was a time where you could crunch some data and output a strategy which could perform in the market. But markets are getting every day more difficult to trade, even for professional traders.

Having a development tool for developing strategies, is not a magic stick, one still need to base the strategy on a trading idea.

Søren, I'm glad that here we share the same point of view. The concept I'm trying to promote here is the result of my own personal development over the past two decades. I also developed strategies which only used a number of technical indicator in some proportions — and even traded them quite successfully for a certain period of time without understanding why. At certain point came the inevitable: I lost quite some money simply because I missed the point when the market regime changed, and I was not able to react to it adequately — simply because I had never known why my strategies worked! For me it was a turning point when I started to develop my own method of trading strategies design — which is based on market analysis, and not on abstract mathematics. And I can see today that again a lot of traders with prior background in mathematics try to do exactly the same as I used to do, I'm humbly trying to draw their attention to the physical world, and not to some math abstractions.

There are a lot of quants that understand causation or why certain behaviors occur in the market. To use your analogy, there are some that are trying to predict the cab's destination by looking at the driver's incentives. Causation can be obscured by the process of creating trading signals since they rely on price history, yet it is easier to predict where the couple is headed given the cab's recent turns.

Andrew, your words is the sweetest music for my heart. Could you, however, give any idea about the actual quantity of this "a lot"? Maybe a proportion to others? And of course the reasoning behind this proportion.

Funny how it sounds, but the better the idea I'm trying to code, the more simplistic the code. And vice versa. I can't imagine any limitations of existing automated trading platforms for directional strategies other than simply technical (like an attempt to trade portfolios of 1000+ stocks or if your strategy requires a certain latency for example).

Oscar - there is actually one method to this "madness", which is used broadly by discretionary traders, but not by automated trading

- many experienced traders look across the markets and instruments before they decide to place the trade. However, such is very difficult to implement in an automated trading system - simply cause there dosnt exist development platforms which contain this, or they are so expensive that only big market players can afford to use them.

Oscar, sentiment analysis is a great thing, only you made a bit of a confusion here: you took immediate reaction to extra-market information, like words uttered by important talking heads, for sentiment, which is a relatively long-term process. For example, very frequently you can tell with a good degree of probability where price is going to go after a certain important news — based on the analysis of the sentiment formed before it. My experience somehow suggests me that there's no such thing as "100% unexpected" price movement; Soros also was not the first one to drive the market down, it's been in the bearish phase for quite some time before.

@Soren, I totally agree with your thoughts. Strategy development is a art of which is developed over many years of real test of success or failure experiences.

Over this time the developer develops a cognitive bias of what, when, who, where,why and how this knowledge can be put to code. These methods

when applied correctly, provide results many multiples of the very low industry standards of today. Typically today, the quants of mathematics

come up with a strategy on a single instrument of which if absolute data was used in the presentation of results it would show to be a instant failure.

Simply said, if you do not have a completely automated strategy with exact entry and exits of which can easily trade 50 to 100 instruments

on the exact same tick, then you do not have a true working knowledge of the markets. It is all about the personality of the markets instead of

some super complex mathematical algorithm. The market personality changes

on a daily bases of which takes hundreds of lines of code to put these changes into the much broader perspective of the results of this single

day personality change.

The aspect of just a few advanced mathematical algorithms to develop a extremely robust strategy is slim to none.

The absolute quality test of a strategy is to be able to apply the strategy to 50 to 100 instruments of anyone's choice of instruments

and over the total portfolio beat any thing in the retail market place today.

At this level you are no longer a quant, you are a master strategy developer. Two completely different worlds when it come to completely automated

strategies.

All the best trading to all

Alex, after taking a poll of contacts (rather small sample mind you), it seems we all believe that we are unearthing some idiosyncrasy caused by some regular market behavior. Whether this still exists in the final system after numerous refinements and iterations is the real question: is the original exploitation of the anomaly still driving alpha? I guess that is best determined by the results and not necessarily our opinion of our methods.

Andrew, I do agree that if someone makes money in the market then it doesn't matter really how he does it. The question here is mostly in the ability of this individual to adequately react to changes to the exploitable patterns and the market regime in general. In case there was no clear idea behind the system chances that the appropriate measures will be taken in a timely manner are very low really, and the stories of 2008-2009 and 2013-2014 clearly demonstrate it.

Agree with Andrew and Alex: how many small time day traders earn their living year after year. On the floor they can beat the spread and split the market orders. But off the floor ? I remember 10k Jim. He did a huge amounts of trades, he's still around. I am an equity option wing trader acting on news, still around. I need my tools, but I am not trying to make money on understanding all markets. Not even interested in understanding markets. In physics, the mathematics will uncover causalities. In markets, its not the same hypothesis. I found that the method is more important than the forecast. Hence its better to be smart than lucky.

Absolutely correct Alex and Stephane. Knowing the reasons why something works is really secondary to if it works. Yet knowing why it works will signal that a system has ventured beyond its "sell by date". Another preventive measure if the why is unknowable is to be on the look-out for market regime shifts. As you said earlier in this string, a market regime shift rarely comes out of nowhere and this should be enough time to pare back positions.

@ Alex, it is nice to see that somebody takes sentiment analysis as an adjunct to trading and furthermore interprets in a relatively longer time frame. I do believe lots of firms take it in a more immediate time frame, particularly those trying to anticipate short term price action. Be what it may, it is here to stay and we all have to live with it.

So there are some abvious problems to solve in order for future developers not to use a decade or more to gain experience, and perhaps even build his own development environment:

Problem 1) There are only 6 base technical indicators, we need also to solve the problem of how to develop custom indicators.

Problem 2) Since technical indicators are developed using a timeframe, we need to solve the problem of how to develop custom timeframes/charts.

Problem 3) Since most successful discretionary traders look across markets and instruments, we need to copy that --> we need an development environment which is cross market aware, in any aspect.

Problem 4) We have concluded that programming is too slow to gain experience and test a high count of ideas ---> we need to replace programming.

Problem 5) We need to provide a solution with at its best is free to use.

What do you think - a dream, or possible ?

1) Why do we need indicators at all?

2) Why do we need to use time frames?

3) Not a problem for any modern software, even retail.

4) Programming is best at flexibility, while learning programming in general purpose language may take time. Therefore domain-specific languages like EL is the way to go. In my case actual coding takes only less that 1% of time I spend on the development of a new idea.

5) Why the best solution should be free? Why anything at all should be free?

1) Indicators, or rather criteria derived from indicators, are the input to any kind of trading, discretionary and automated. Wherein an indicator and a criterion can be anything. So how are you going to trade without inputs ?

2) We need to format the data, as most technical indicators are calculated using data formatted as a timeframe.

3) I think it is a problem.

4) I do not agree on this view. Much more sophisticated solutions can be developed using visual tools. Take for example web sites and desktop application, they would be very primitive today, if it wasnt for visual development tools. Even the most die hard programmer use visual development tools developing desktop solutions.

5) I was thinking in the perspective of the users - not the vendors :-)

.. but perhaps not the right forum to pop such disruptive ideas and suggestions.

Søren, I should disappoint you: the most impressive web sites are still coded in command line tools. It's another story that today anyone can take a ready made template and indeed create yet another similar to others, but nicely looking site in minutes. I hope this is not what you'd name a "sophisticated" solution. But indeed, this discussion doesn't seem to be appropriate for this thread, maybe you'd want to start a separate one.

As to free solutions — I am a true believer that anything free is only demoralising and depraving. This is what life has taught me. However I admit there could be alternative experience.

There are several platforms out there which are free - where the tech provider share revenue with the broker, not affecting commissions or the spread. The vendor makes money - they user gets the product for free - a win-win as I see it.

Soren, are you familiar with TradingMotion?

You mean if I am familiar with; Collective2, Etoro, Zulutrade, Tradency, Mirrortrader, Darwin, Yahoo Japan, Saxo Bank Trading Floor, FXPro, Ayondo, MyFXBook, CopyOp ...... etc etc etc ?

Personally, I would be very reserved with sentiment data from Social Media. To me this data would be very relative in nature and very easy to miss interrupt.

Basically, the data is on what people say verses what they actually do or have done. At the moment, I can not think of a way to qualify the data to a more

absolute value of which can be properly analyzed.

To bring it back to the original post by Alex - imo the investing strategies can't be 'unidimensional' anymore. Just technical, or just fundamental etc. - with so much information bombarding investors from everywhere, many factors are into play. Sentiment data is one way to bring some additional depth, but so are the options markets data, the analysts estimates etc.

And who knows, maybe even astrological readings:)

Oh, astrology! Thanks for reminding me, I will post my "astro trading strategy", hope it will be as fun as my previous posts in the "Trading as religion" series.

Alex, there was a mutual fund based on astrology created in the 80-ies and it outperformed the market ... for a year, and never since :)

Would love to see your 'trading as a religion' series

Well, as to the complete series I'm afraid it's only available now there: http://trading-as-religion.com, but the latest one is hot off the press on Linkedin. Do promise to post the astro research soon :)

Right now world is moving towards chaos, so we can find substantial jump in volatility and premiums for next 1.5 years. it ll be interesting to see how the LT Quants perform. one interesting thing about intraday is lower risk reward can be adjusted with higher leverage with a good strategy.

Pradeep, I highly welcome more detailed and especially numerically supported explanations on risk/reward ratios depending not on the chosen risk management by the strategy designer, but on the data resolution and/or holding period.

Antoine, the idea of this thread (and this group in general) is not to sell anything to anyone. Therefore I will refrain from posting explicitly promotional materials and will expect the same from others. Any interested party can always ask questions in private messaging.

@Atanas - "as an additional data point in a broader strategy" - we are at the same page, I fully agree, and am in fact looking for two additional inputs to what we are doing:

1) Market news on FX.

2) Market sentiments on FX.

3) Same for oil and gold.

Anyone who can deliver such, whith historical data, please contact me.

@Antoine, that was the second time that you have tried to draw Alex out. Now, with the information that you have posted in your

background about the performance of your fund.

120% annual return may or may not be impressive. It strictly depends on the mix of instruments in your portfolio as well

as the total number of instruments that you actively trade on a daily bases.

25% peak to trough draw down is horrible. No modern day quant would allow much more than about 15%.

I will be very honest with you about live trade results. I have not live traded my strategies since the spring of 2012.

It has taken me this length of time to recode my strategies to be compatible with the new versions of TradeStation.

ball is now in your court

Soren, yes, seems we are on the same page but trading different instruments. I really dont know if there are FX sentiment databases and wether they work at all.

Alex, I enjoyed very much your trading as a religion series, the main point I took out of it (and agree very much with ) is to not take any 'everyone knows that..' statement at face value without validating it with specific , actionable data. Inthis respect, and if I understand correctly your post, I am not convinced that you need to follow the big institutions entry/exits long term. They can move prices short term, but their overall returns performance is alarmingly low - in fact if you bet against them, you would have a better statistical chance.

About moderating the information posted on the forum. I have no issue with a no performance information allowed to be posted. Once, I was aware of this rule I would have instantly deleted my post and I did. It was totally unprofessional with what was in a email and posted on the forum about my character . It was not about that I posted

Performance Information, it was what the Performance Information showed.

all the best

For the record - the terms are often messed up by those who do not understand fully what they are doing:

-----------------------------------------------------------------------------------------

Algorithmic trading, is executing a large order in the market by using an algorithm.

But, is also considered to be manipulating LMT orders to influence the market.

This kind of trading is typically about milliseconds.

----------------------------------------------------------------------------------

High Frequency Trading; is about fiber cables, radio towers, to get information first, also milliseconds.

---------------------------------------------------------------------------------

Automated Trading - also called "rule based trading", is above seconds, or even minutes.

If ( MACD > 0 ) then buy.

As EU specify in their definition; "turn the strategy on, and go to the beach".

EU specify 500ms as the border between Algorithmic and Automated Trading. So, if you hold a position more than 500ms, then you are using Automated Trading.

If you hold a position less than 500 ms, then you are violating the upcoming EU regulation, and many other regulations already in force.

For the record: my strategy is a portfolio strategy.

No limit orders.

Typical trade is hours long

I think Quants must be open minded to include any possibility. The lack of flexibility is directly attributed to a Quants teaching and experience. Sad thing is most Quants were taught by tenured professors who actually have no experience at all. Truly you can make money in the markets at any participation level. It's a matter of resources and capable application.

Group, please allow me to say some opinions based on over 15000 hours of screen time developing

custom indicators and custom strategies. And by no means am I claiming my opinion are the only

correct to trade the markets, they just fit me and no one else.

Soren, I respectfully disagree about a strategy has to work over different time periods with a explanation.

Typically, a strategy that can work over different time periods are what I call indicator based and some changes

has to be made on the input values to change from one chart period to another or over many different instrument.

I only use two or three indicator based code and they are very customized and this over a few thousand lines of code.

I do not need to change a single input value,variable or single line of the code in the signal generator to trade

as many instruments as you so please. I typically use one minute chart for two reasons. I get better pen point entry / exits

(this has a major effect with draw downs both per trade as well as peak to trough)

and with one minute bars I can achieve my minimum requirement that the strategy has to be stress tested over

500,000 bars (using days, weeks, month and years instead of bars is no where as efficient) as well as over a minimum of two asset

with a total of at least 20 different instruments.(pleased notice I did not say optimized)

Alex,

Over the last few years there have been some studies done as to delaying your entry / exit signal into the exchanges

with positive results. The timing of when a signal is actual sent to the exchange can be rewarding in some styles of strategies.

I have been doing this for over 10 years.

There has been a lot of HFT shops closing in the last year or so. I have never found a advantage what so ever in “tick trading”.

Please, I am in any way trying to solicit any thing from any one.

sorry, the number of bars that I mentioned above should have printed five hundred thousand

I think we are at the same page James, every developer and system designer has his method, and the comments I make are often related to the - lest say "style" - of development I represent, not to go any further into this "style" of development.

As to the delayed entry, that very much depend on the strategy you use - if you jump on a fast move, the delay matters, we measure the spread, and the difference between the price when the order is send, and the filled price - and it is very clear, that slippage is quite high when you jump into a market move. Else the slippage is quite uniform around zero (if you use the right broker).

You can also skew the time, and close bars 5 seconds off the minute, to perhaps get a better execution, or a bar close before the competition.

Ask any prop trader if it matters when you enter a trade on the bar close, or later in the 4H timeframe.

James, I am not 100% sure what you mean, but I believe the issue in question (with delayed entries) can mostly be attributed to local corrections after a decent move which normally happen within the first minute after such a move. Since there's plenty of strategies exploiting price movements of this kind (breakout, momentum, "price action" and so on) no wonder that most of them benefit from such a delayed entry. However if you make such a delayed entry a permanent rule without proper testing it may play a very bad trick with you, because in certain cases it may improve the entry price, but in certain cases you will suffer from a significant slippage, and to gather and analyse reliable statistics about these outcomes is in fact quite a problem.

As to whether high frequency or any other technological advantage may or may not remain an edge — well, it's an open question. I even recently have written an article for Tabb Forum: http://tabbforum.com/opinions/is-the-future-indeed-so-algorithmic.

The problem with intraday is that you may miss the gaps between days.

Sounds like the topic describes what traders do: technical indicators, candles, etc. This not exactly what a quant trader does. A quant is a specialist of quantitative risk management, and does not take proper directional positions, sometimes he/she can include a little bias in the strategy, based on a personal sentiment of the market, but rarely he/she will have crazy exposure where the book can be smashed by just one ot two big moves.

Everybody can, with some basic math knowledge, build an algo based on a pattern observed in the past, no need to be quant for that.

Quant traders are just quants with programming skills, developping their own programs to execute the trades.

Not so much if intraday or not, but rather whether best to trade instruments which has gaps or those traded 24/5 - the issue that your unattended automated system is having an open position in a closed market, to make sure to exit every day before market close, that the tech.

Jiliali - quant or not, all use a set of criteria as input, those criteria are derived from something, which in general are called an indicator. An indicator is calculated using a formula, which can be anything. A criterion can be anything - is it Monday today ? true or false. It might be possible to develop without formatting data into timeframes, but for sure at least one input derived from a formatted timeframe can contribute positively to a strategy - unless we talk about algorithms in its correct industrial standard term; to execute a large order in the market / market manipulations etc, QUANT is for me not algorithmic traing even you could probably use some quant methods in an industry term algorithm.

Jilali, the topic doesn't describe what traders do. The topic doesn't describe anything at all. The topic just raises a question about the foundation behind any trading strategy in general and any part of it in particular, where risk management is no exception. For example it's common place now to hear that "you should risk no more than 1% per trade", and it's funny that really a lot of people blindly do it only because they were told, without even an attempt to find an explanation.

What I can clearly see is that almost none of the participants in this discussion with a remarkably rare exception can stay in really general terms. Eventually the discussion just goes down to any particular examples, while the original question was whether quant (systematic, algorithmic, numeric, formal, rule-based, whatever-you-name-it) traders really understand to which market processes their quant (algorithmic, numeric, formal, rule-based, whatever-you-name-it) models correspond, or not. Looking at quite a few answers I tend to come to a sad conclusion that the answer in most cases is negative.

Gaps are for some discretionary traders important, indicate a major market move. That makes them important for automated trading also, if you want to convert the trading style into automated. But trading gaps is a difficult art, and probably difficult to implement in automated trading.

There are differences between the strategies used discretionary and those used automated, simply cause of ease of development. It is for strategy development of an ATS a matter of minimize development time, the complexity of development and the complexity of the strategy itself.

The decision of which instruments to trade, which timeframe(s) to use, which data to use, has a major influence on the development time, how many ideas you can develop and test. Using for example minute data as lowest timeframe provide other types of strategies, equally performing as the strategies developed and tested using tick data. Taking profit at a bar close instead at a fixed exit P/L is equally reproducible.

Gaps do not indicate "major market moves". In different markets they mean a bit different things, but in forex for example they may mean two simple things: that either over the weekend there was an important news, or simply a continuation of the movements started on Friday, I hope no need to explain which events might initiate such a move. In stocks gaps can be seen more frequently because of the limited duration of the trading session and the fact that quite a few key economical news are released before the market open.

Trading gaps is no art at all, they are the easiest to trade, and even source codes for strategies based on fading the gaps are available in public domain. It's interesting that almost 90% of these open source strategies work reasonably well.

@Alex I always enjoy the discussions you are a part of. You and I trade diametrically opposed strategies and I must confess that the longer term strategies seem more like magic to me than anything. My shortest trade recently lasted 9 milliseconds from order entry to execution (it was profitable) and I am out of 99% of all trades in less than a minute. I stick to where my edge is; boring but profitable.

As Bruce Lee said, I do not fear the man who knows 10,000 moves, I fear the man who has practiced one move 10,000 times.

@James you are correct about a lot of HFT shops closing up but what may not be common knowledge is that most closed not because the trades were no longer profitable but rather because the trades weren't profitable enough to cover the huge costs in infrastructure required to be competitive. I choose the path of not trying to outrun the bear, but just being able to out run the average trader.

Once I asked a trader to explain me the latest movement of the gold price. He answered me with the question: do you know how to use that price movement.

So, may be the question is: how many traders know how to use the price movement for making profit.

Scott, there's still one thing we have in common: we both understand the very essence of the market process(es) we're exploiting. The fact that they reside on different time frames is just secondary. That's what makes a huge difference between both you and me, and, say, a typical "quant" defending his strategies with a help of signal-noise models, Bayesian logic and other stuff.

Alex, exactly what are signal-noise models?

The question is whether the trader understands what he's doing or not. Sorry, but I don't know how to formulate the question in an alternative form.

James, I mean models and the mere concept of the market as consisting of some useful information and useless. The latter is most commonly called "noise". I guess the term originates in radio physics where indeed we have useful signals and noise and the main task is to filter the former from the latter. However if we look at the market as a superposition of human actions, and not as a natural phenomenon independent of human will, we should immediately admit that there is no noise as such in the market, and therefore there's nothing to "filter".

Alex - more than 70% of the executions in the market are not by humans, but by computerized systems.

Of cause there are noise in the "charts". It is the noise which generate false signals to our trading strategies - too much noise, too many false signals. But if you want to argue that noise in it self is also an information, no problem with me - what ever works for you.

The fact that the executions are made by robots doesn't mean that robots actually design themselves. And if you define noise as the component of the price time series which generates false signals in your system doesn't mean that there isn't someone else who very successfully uses this "noise".

Ok. For myself, I think I do understand what I am doing. However, I don't claim I understand the market, but I am swimming with the wave. At certain times I survive, dive or sink, altogether!

Alex, I would be inclined to think that anyone that has studied the markets over a period of time will have the opinion that human will

is the largest factor in the markets. In the market, one of the basic concept’s is one human’s will to make money at the expense

of another. The very basic relationship in nature of “Who is the predator and who is the prey.” I very quickly learned I was

the prey or part of the food chain for the predators. One of the basic survivable technics of a human is to know who the predator is

and what their actions may be. The prey then survives by their reaction to best interrupt the signal from the noise.

IMVHO, Predators are in control of the markets at any and all times. They control the signals of market direction change as well as the

noise in the market. They control the mathematical relationships of the markets when they express “their human will” with

super computers using mathematical algorithms to not only lead the prey into traps, but to cover their intentions and even

more important who they are.

Please allow me to try to answer some of your questions with my own thoughts and opinions.

First, I am absolutely not qualified to be judge and jury of someone else's knowledge on how they trade the markets.

Why so many people recommend smaller position size in a individual instruments? This may come from computer analyzes of

of some of the many different models, methods and algorithms used in the markets. Perhaps for the short term trader, these results of relative information

produced showed that it was a more effective use of capital to have very small positions per instrument and then trade

many more instruments.

I am not a mentor or guru, nor am I soliciting anyone for any purpose.

all the best

Alex, thank you for the kind words. We definitely trade on different time frames and while I cannot speak for any other traders, I can say I stick to trades I thoroughly understand. To do otherwise would seem to be the equivalent of trying to cross a busy street with your eyes closed and ear plugs in.

I like to think about time frames in terms of the characteristics of a move that catches a broad category of trader's attention. Large hedge funds wouldn't notice a 10 pip move in FX unless possibly at a critical juncture (i.e. an option defense price) but I sure would. If you repeat this process thinking about what type of trader would notice what type of move (characterized by time, price, volume, whatever strikes you), you may find a somewhat orderly progression with some practical application or you may see it all as nonsense.

For me, this is helpful when designing the critical "don't do stupid stuff" part of every good algorithm.

Alex, you seem to have quite a negative view of quants, that's your point, not mine (looks obvious...). As Søren Lanng pointed out, quants are not algo traders, in the common sense of the term (looking for latency inefficiency, etc.), for which I can understand it sounds like cheating on the rules; although personally I don't have any problem with them.

My point was, that you describe (or "talk about", if you are more comfortable with) quants like people pretending to be able to predict the market. And this is Wrong. If you have met quants telling you that they have super complex models, using whatever super advanced stochastic methods, that make them beat the market every day, you are not obliged to believe them; as well are you are not obliged to make a generality out of it.

I've never heard a quant saying "You should risk no more than 1% per trade" neither. The risk you take is based on the inherent risk (it's intentional that I don't say volatility...) of each instruments you trade. Again, people saying this are not quants. Every quant has his own methods, I personally use volatility based position sizing and kelly criterion to determine the size of my positions. Nothing human driven here, only mathematical logic. You think it's useless (quote: "That's what makes a huge difference between both you and me, and, say, a typical "quant" defending his strategies with a help of signal-noise models, Bayesian logic and other stuff."), not me.

Talking about signal-noise (using your terms): noise is the best enemy of quants. I can't imagine a quant proposing a model saying: "Well this is my model, I can explain 60% of the movements and the remaining 40% is white noise."

I hope it was clear, when you first wrote your post, that you would probably be talking with people who disagree with you. I have a little doubt.

There is no noise in the market, and every (sic!) tick has its meaning. The fact that your model cannot explain it doesn't make it noise. You might be interested in filtering out some data that you're uncomfortable to work with and you call it "noise", but then Scott Boulette comes in and squeezes a lot of profit out of this "noise" — simply because he understands its nature.

Jilali, of course I forgot to add many popular and less known approaches to the list — so to prevent a possible argument that real quants today don't use stochastic models and/or Bayesian logic, please add to this list anything that comes to mind — from machine learning to crystal balls.

Alex - You posed a slightly different comment since I last posted. "So far I haven't seen any reasonable argument why we should prefer stochastic models or Bayesian whatever to simply studying real cash flows which actually drive the price."

From a fundamental investment approach, you are correct. I'll make one small change to your comment though: "future" cash flows drive prices. You still have to provide estimates for your variables (i.e. fundamental factors that will drive future cash flows) and your model will be subject to a number of forecasting errors. These models do have the benefit of being less abstract and thus easier to explain.

If you are doing any work with derivative strategies, you will have very little choice but to use stochastic modelling. However, using such a model to predict or forecast a physical security (stock) price would not be recommended. The counter argument will be that prices follow a random walk process, which I disagree with.

Antonio, let me try to put it this way: the most reasonable path for the research is to research which exactly stochastic models is used for the derivatives pricing by those who form the price. And then trade against them.

As to the random walk process — recently there was a discussion in our group following a paper where quite acknowledged and reputable scientists proved that the market price time series don't follow the random walk, so we can assume this question as answered.

Alex - good to hear that I am in good company regarding the rejection of the random walk :-)

Conceptually, I agree with your strategy regarding trading against them; much more difficult in practice though.

As always, enjoyed the exchange of thoughts.

Great topic! I back-tested more than a hundred algo strategies on the past 30- year timeframe and I am sticking with day trading. When you go with longer periods of time the outcome predictability substantially deteriorates. The probability of substantial event occurring with a company stock, I have longed or shorted, is growing exponentially with every day. It's true that this substantial event has a probability to reflect positively on my return in half the cases, but the volatility grows exponentially by every day you maintain a position. Since I trade on statistics, the volatility and predictability are the keys for stable return. My current strategy trades 24 day/swing algos simultaneously and returns 50% per month. I was not able to achieve similar results trading over longer periods of time. From my experience shorter periods are safer for algo traders.

Sergey, I take my hat off to anyone who can make almost 1300% a year, especially trading stocks. Could we, the rest of the group members, hope that you gave us just a slightest tip about the direction to which we should head our researches to reach at least one tenth of your performance?

Alex , yes I am testing my strategies with real money. The strategy trades major US equities (NYSE, Nasdaq, AMEX) with at least $5 share price, decent market cap, and decent daily volume.

Current version is optimized to trade up to $10M a day. It opens 6 new positions daily on average. I can dial daily investment up by making investment parameters less restrictive. The return will slightly go down, as well as volatility.

There are 24 algorithms optimized on the past 20-year set of data to maximize return with a penalty for volatility component.

The big part of optimization was to decorrelate the algos, i.e., if two algos perform alike, one of them becomes a gap-filler, so it will pick up the slack during the periods of bad performance of the first algo. This allows for less volatility and more stable return.

The strategy is fully automated and runs on multiple dedicated servers in a data center with a great Internet connection. There is a large number of components in the system responsible for querying various data sources and executing the trades. Because some of our positions may be 10-50K of shares, the program splits them into small batches and feeds them to the market without alerting other traders of that instrument.

I am still working on the execution component so that I can start trading larger sums without running into bad fills.

Frankly some of the things I am doing, I haven't seen discussed publicly. The platform is the result of many years of research and programming. I didn't see any other strategy by anyone that performs as well as this one.

I posted the theoretical numbers for the strategy in this file https://docs.google.com/spreadsheets/d/1FmWDsxBQX5YCR-fxWD_j6oL3HiyNKfF34qVEcSMxc_M

The returns are cumulative w/o leverage.

Alex, interesting references to "the market regime change." In my approach I try to build the system that works well in any market conditions. For that I optimize it over extended period of time and combine together many algorithms, which perform well in a long haul even though they saw some bad times as well. When I look at the daily picks my program makes I can spot the next day losers pretty well because of particular algorithms that don't perform well lately. The problem is I cannot predict "the market regime change" and I even stopped doing that. I let the bad algorithms run with the good ones because tomorrow they may switch places. The one which was good becomes bad, and the one which was bad becomes good. The idea is that as a combination they have been tested to work well in any market conditions.

Sergey, I am not sure if this discussion is the appropriate place to advertise your programs. However if you don't mind discussing certain particular aspects of your developments it would be very welcome. For example, I am all ears to hear how you manage to efficiently use partial fills for 10K+ shares for your news trading algorithms. I think it will be very educative for most of people in this group.

To execute the large orders, I use consecutive LMT, MKT, MOC and LOC orders with quantities lesser than current bid or ask size.

Thank you, Sergey, I do not have more questions.

Seregy, very interesting concept and methods. Have you tried to lay off large positions across multiple brokers? I have no idea if this will help today or not.

all the best

http://gobinary247.com/lps/landing-two/index.html

James, this interesting idea I published almost 2 years ago and there's still the link to it in my profile: http://tabbforum.com/opinions/formalizing-risk-diversification-new-opportunities-in-portfolio-design-for-the-buy-side. Executing large positions in stocks via multiple brokers may only make the average price worse because of the latency. If you mean OTC instruments it might be a slightly different story, but in any case we can't seriously discuss "50% monthly" returns.

James, since I don't send to dark pools, all orders are showing up on the tape. I am not sure if spreading them across the brokers would help. Plus only few brokers have decent APIs. Also many brokers don't give you MOO, LOO, MOC and LOC orders. Thank you for suggestion!

Alex, I understand that high returns are hard to believe. That's why I started publishing the trades as I open positions. I know building a public track record will take some time. But at least you can independently verify since I started publishing trades a week ago. You can see that algo's weekly performance was around 14% cumulative, which is in line with theoretical model. https://twitter.com/zQuantAlgo

Sergey, thank you for your update. Unfortunately I am probably a wrong person to sell your product to. I doubt if I might be interested in a strategy which assumes trading news in stocks filling 10K+ shares using market and/or limit orders. I also doubt if anyone might. However good luck, but please use a separate thread of discussion for your promotion.

Alex, I didn't mean to barge in. Just trying to answer some questions. I need some help /opinions from experienced trades. I will post my questions in a separate thread.

I don't think a quant shop would have one trading system. I would think quant firms would have many different systems that trade different styles, and some are more aggressive than others.

All this is just dream talk - so when are you going to get real ?

Muhammad, aggressiveness is a term of money management, it's not related to the logic of a trading strategy. Of course trading firms, and not necessarily only quant, trade many assets using various strategies, however the resulting portfolio is always very well balanced not according to the "trading styles", but to the value each asset and/or strategy deliver to the portfolio in terms of its robustness and risk diversification.

Nevertheless we've again deviated from the main subject of the discussion. It seems that with only a remarkably rare exception the discussion participants are not willing to discuss the reasons why their strategies are in the market or out of it. I'm starting to feel that I have to come to an unpleasant, but inevitable conclusion...

I think FX trading got a little harder now because of all of the turmoil caused by policy shifts and currency wars. Until recently FX trading was probably easier than stock market trading. Now maybe not. Even though stock market looks more random than FX, it has things that clue you as to the immediate direction. Support and resistance are common to both FX and stocks, but vix, trin, tick, ad line, market profiles, diversions between markets, volumes. can all be used to time a short term trade. I have a lot more success picking the short term bias than the long term bias, that's why I trade short-term.

although trin isn't as useful as it used to be. Nowadays elevated trin that continues to rise can coexist with a rising market, which was impossible before 2008.

The shift in FX market is real, an FX trader must realize the markets of yesterday are not going to be similar to the markets of today and beyond, until the calm return and the fear abate.

Partially agree with Muhammad on multiple trading systems. A quant shop should run multiple algorithms, but they all can be managed from a single trading system/platform. I pay for subscriptions to various databases and APIs, replicating them will go against my bottom line. Also maintenance of multiple systems becomes a nightmare. Multiple algorithms reduce volatility, provide stability, improve predictability, allow access to better liquidity (I know for many retail traders it's not a concern though).

Maintenance becomes a nightmare only in case you don't know how to do this maintenance. Normally developers work closely with risk managers to compile the list of procedures, both routine and emergency, which cover any possible situation and mitigate both systemic and operational risks. Of course if you didn't do that then it indeed may become a nightmare.

Some short term traders trade the open or the first 90 minutes. What would stop you from playing the open and also swing some other trade, and trade everything in between?

Muhammad, is it indeed a question to me?

FX traders didn't have to deal with currency wars before. This would be a major change

FX traders didn't have to deal with currency wars before. Currency wars is a major change. The advantage FX trading had over stocks trading was, trends used to last longer. Nowadays, countries announce new policy every other day affecting the FX landscape, making it rougher, I think.

I agree with you Marc that systems need to adapt, actually we said that earlier when we were talking about having several systems not one system. There was opposition in the group to the idea of having more than one system, some suggested it would be too complicated.

Muhammad, could you please be a bit more specific in your statements? For example, I would be happy if your explained what a "currency war" is in your terms, how it affects the markets and why, and which particular groups of traders suffered from them? This also strongly relates to what I said about understanding what's going on in the market.

Marc, very true about the relationship between strategy complexity and its stability. Unfortunately although simple strategies mostly are very stable against changes in market regimes, their returns are just slightly above acknowledged benchmarks, and therefore even investors tend to look for less stable but potentially more profitable solutions, not saying about private traders. As I have to talk to many on them almost on daily basis I can see that 90% of them are still convinced that fx opens an opportunity to make thousands of percent of return per year with zero risk. The rest 10% are convinced however that it's totally impossible to systematically make profits in this market.

The advantage I always heard from Fx traders was that FX holds trends for longer and it doesn't fake-out as much as equity markets. It seams to me now with all the uncertainty surrounding the economies of the world and the more frequent interest rate changes by the various CBs, and the more frequent statements and claims by these CBs, the FX markets relative calm is something of the past. They (FX traders) are in an era of constant tectonic shift now, which would/could affect systems backtested against data 2 years old and older adversely. Note the FX markets calm here refers to its relative calm compared to equity markets.

Marc, I am not surprised to see it because of the very problem which I tried to discuss in this thread and which you fortunately understand — but still a lot of traders don't. The problem of changes in so-called market regimes is mostly used as an excuse for severe losses which simply are caused by insufficient understanding of what is going on in the market on short, medium and long term — and of course this is very, very far from all that retail nonsense like "currency wars", "central banks manipulation", "currencies are more trending than Middle-East refugees to Europe" and so on: these only can add to the collection of meaningless terms. I do have a number of semi-formal models which explain changes in the market regime, and thanks to them I successfully prevented a couple of accounts in 2013-2014 from running into very deep losses, and I just hoped to discuss serious things with serious people. Unfortunately what I saw was a collection of truisms and a strong sales pitch.

And just one more note on predictability: if someone says that a certain market is unpredictable this can mean only one thing — that the very model he used for predicting doesn't work. The fact that 99% of researchers believe that there are only two models in the world — his own and an incorrect one — doesn't make markets less predictable. At the same time it doesn't make them more predictable.

Abdullah, I base my judgements on a number of observations. First, my personal contacts in the industry. Second, various places for discussions, mostly outside of Linkedin. Third, I regularly monitor publications, both professional and academic. And fourth, I am no less familiar with the performance of quant traders in 2012-2015 than anyone else.

This makes me curious to know the performance 2012-2015 part indeed, please, and how they calculate it. ~smile~.

But you didn't put your findings

Alex.

1- do they consider trading with original deposit each month, or they consider trading with original deposit + its profits, since comulative trading brings different results.

2- Could you give me the minimum and maximum performances since 2012 that were considered "very successful"?

As well as their highest acceptable sharpe ratios "drawdown". And, do they use Modigliani ratios?

Marc, my observations indicate that there is a kind of a trade-off between the "universality" of a model and volatility of its returns: the more "universal" is the model, the greater and longer the drawdowns essentially. Did you analyse your model against other, market-specific models? If yes, I am really curious to hear the comparison of some key metrics, especially recovery rate, return on drawdown and average annualised return to standard deviation of annualised returns ratio.

Abdullah, I am very proud and pleased with an honour to make a free research for you, I am only a bit short of time right now.

Alex, first, I owe you and others an apology. After reading your post, my first thought was why did you post that. I reread my post and I can easily see

how my poor use of the English Language change the intent of what I meant completely. My using your name was not meant to address you personally

but simply catch your “eyes”. My use of “you” was meant to be addressed to all. My intent was to try to express, with a example, of someone (myself)

is forced to use a“collection of meaningless terms” as many would think. Often, when I am trying have communication with fellow traders I get

“caught between the bark and the tree”. I take it on the chin and move on.

Alex, I truly did not take that as being personally directed to me at all. In fact, that is a very fair description of how I have to communicate.

all the best

At the risk of sounding snarky - in my experience 90% of all "traders" don't actually trade

"And fourth, I am no less familiar with the performance of quant traders in 2012-2015 than anyone else." This sentence you answered me made me asking for the above. Sorry if I understood wrong.

Scott, if by chance your thought was about me, then I have no secrets that I have not live traded since the spring of 2012 because my strategies became

incompatible with the new versions of TradeStation at that time. Honestly, with the events in the last couple of week, maybe never using the TradeStation platform. It is what it is.

all the best

@James, the comment wasn't directed at anyone, it was prompted by a variety of posts I have been reading lately in several different groups. I am all for an open discussion about anything related to algorithmic trading (or just about any aspect of trading); presumably everyone active in the group is here to learn (or possibly teach and hopefully not sell) and I am completely open to answering decently well thought out questions where I have the appropriate expertise.

What I do find amusing from time to time are "traders" talking as though they are an expert on a subject, all the while saying things that people who trade for a living know with 100% certainty are just not how things work. I am not talking about a difference of opinion, I am talking about being just plain wrong.

It reminds me of a little kid swearing he didn't eat the chocolate cake with chocolate all over his face and thinking the parent is buying his story because he can tell they are trying not to laugh.

Scott, I bet we should see now a good number of posts requesting you to publish your track record, or bank statement, or whatever to prove that you actually trade. Amazing how it is, but it only reminds me of a great Russian absurdist Daniil Kharms: he once wrote a series of extremely small "plays" which basically consist of a simple dialogue like this:

Physicist: I am a physicist.

John Doe: And I think you're piece of shit.

Physicist can't utter a word and falls dead.

Curtain.

I absolutely don't know how to behave in situations like this. Do you?

At first glance this appeared to be about multiple time frame analysis, but it seemed to morph into a rant against quants. My advice is everyone has their own unique trading style, and each trader needs to figure out what that is. Are you a day trader, swing trader or position trader. Once you figure out how trading will fit in your lifestyle, then you can choose what kind of trader you wish to be and that will dictate your time frame analysis. If I'm doing Globex trades on financial futures, I'm on a 60 minute chart for trend, 15 minute chart for defining supply and demand levels and a 5 minute or 3 minute for execution. If it's an option trade on a stock or ETF, then I'm on a monthly/weekly for trend, daily for supply and demand zones and a 240 minute or a 60 minute for execution. It all depends on the asset class and type of trader you are (i.e. you don't day trade options due to Theta). To answer your question at the end, call the guys at Renaissance Medallion Fund.

Christopher, thank you very much for your advice. May I only ask you what you base your advice on? What are the fundamental economical reasons which urge you to use the timeframes you mentioned? Why 60 minute and not 50? Not 70? Not 100? Why at all 3 time frames and not 4 or 2 or 5 or whatever else?

Group, personally I welcome open and respectable discussion about anything that I post. I am only here to learn.

Scott, I have a high regard for your success in the structure and methods that you use to trade the markets. I would be inclined to think that you are producing more efficient results than the market makers do. My question is what method do you use to compare your own strategies as to the best one to use? example: Van Tharp

all the best

@James - Possibly undermining whatever esteem you hold me in, the answer is - I don't try to determine what works best, I only try to determine what works at all. Most everything I do is at its base, a guess. It is an informed guess but still a guess.

I guess as to what markets to trade and what times to trade them, I guess as to what parameters to use, I even guess as to whether my code will work. I found a bug yesterday that had been hanging around for close to two years and I only found it by accident because I was taking some latency timings in an area of the code where I normally wouldn't be.

I test in simulation only until I determine the code basically works as intended; after that is it all live trading with real money at the smallest size that will provide me with the information I am testing for.

I get real nervous when someone tells me they absolutely know, especially that they know with enough certainty they can tell someone else what they should be doing.

Hi Alex - I use time frames that work for me. There is nothing wrong with using any of the time frames you threw out above. No two people trade alike. The main point is to be able to see the forest through the trees. I don't want to take a long or a short on a 15 minute chart if I don't know where I am on the curve (generally a weekly chart). I don't want to short on a 15 minute chart if I am close to a demand zone (area I feel institutional buy orders reside) on a daily or weekly chart. The reason is that, in my experience, higher time frame zones hold better than shorter time frame zones. Levels from a higher time frames have to be chewed through on smaller time frames and that frequently causes price to reverse in the short term. Trading is all about identifying where institutional order flow resides on a price chart, getting behind it through confirmation entry on a reversal pattern(not in front of it) and catching a good piece of the move. Trade safe!

Christopher, well, this is what I expected to hear. "This works for me". An answer as universal as useless. What I honestly cannot understand — and this is exactly the main subject of this discussion — is why trading is perhaps the only domain of human knowledge/activity where 99% of argumentation is based not on the physical reality observations, but on personal judgement, or even worse, feelings.

Thus said, for example, if someone says "I enter the market because I can see an imbalance between specialist and commercials positions" it makes sense. If someone says, "I exit the market because normally around this date we may expect an options barrier protection" this also makes sense. If someone says, "I do something because I can see certain footprint of a major market participant who uses an iceberg/liquidity seeker/whatever algo" it does make a lot of sense. But when I hear just "I do this because it works for me" or "I use this because it better shows me trends/support/resistance/double shoulders/whatever" then it doesn't make sense, and this is exactly what I tried to draw attention to with this discussion.

Hi Alex, polarity of opinion is what makes markets. Perhaps I read through your post too quickly and missed the point you were making, if so, I apologize, and that is on me. Having said that, I have a rules based system which takes emotion out of the equation. I trade what I see (or what the charts offer), not what I think or how I feel. I spent a decade as a fixed income derivative trader at JPMorgan writing out-of-the-money put options on fixed income portfolios in Defined Contribution plans as well as on Bank-Owned Life Insurance portfolios. I've adapted my experience in the OTC markets to the exchange traded markets (with the exception of Spot FX), and I have a trade system that works for me. If the quants, or smart money, have it all wrong, and you have it all right, just trade against them, transfer the wealth into your account, and be happy you were "ignored".

Alex - you may be beating a dead horse here :-) Quants that do understand why they are taking a certain action will never disclose that in a public form (even in a private forum if the compensation isn't sufficient enough), so you have reduced your audience to quants that only do what you are critiquing. I am generally vague in my public responses for that very reason - why would someone (other than for academic purposes) publish valuable information for no fee?

One thing that strikes me in this exchange is the range of mechanisms for explaining the reasoning of a trade entry/exit. This appears to go from "I have a feeling" to reliance on hard numbers. While it might seem as though I would come down squarely on the side of hard numbers, that isn't entirely the case.

I took what I think of as a systematic approach to this problem. I chipped away at quantifying what my eyes could easily see. I threw away thousands of lines of code over the years just trying to accurately capture a few simple feelings - fast, strong, long time, far, etc. Once I adopted that framework, I worked on moving away from "fast" to "N price changes per second for P seconds extrapolated out to T time period".

It is possible to quantify most things and it is possible to trade them successfully before you can fully quantify them. Black box is the goal but gray box is often the reality.

Group, please understand that I hold high regards for everyone that has posted or will later post. What ever your methods or perspective is absolutely the correct way for you trade the markets. What ever my method or perspective to trade the markets should never be a factor

in your method or perspective to trade the markets or your mine. There is no possible exchange of thought in a debate. For myself, I try to stay in a win/win situation at all times. I care little if you better my levels of performance on any tradable instrument. If you do better me, then it simply presents a personal challenge to myself to understand how or what you used to better me. The challenge should never be between you and me, it simply is

between me, myself and I to improve my own personal knowledge and skill set to improve the efficiency of my own trading methods.

all the best

Behavioural sciences do create an added value because they challenge the mental models traders use. No problem for me if you really want to see them as 2 different topics. But when building and implementing a trading strategy I would prefer to interlink both of them.

BTW, what you mentioned about 'the international brokerage' topic is indeed something that is already been done by bigger banks and hedge funds.

Yeah I know, while I am dieing to find only one quant who can automate my manual successful rules into a system, using C++ with FIX, statistics and backtesting !

Well I can always help you out in that regard ;-)

i sent you an invitation

Who's next? What are the charachteristics of an ATS programmer on an LP level who is passionate to open a workshop with traders who have successful tracking records?

We might open a topic for that.