Below is the equity curve for an equity strategy I developed recently, implemented in WFC. The results appear outstanding: no losing years in over 20 years, profit factor of 2.76 and average win rate of 75%. Out-of-sample results (double blind) for 2013 and 2014: net returns of 27% and 16% YTD.

So far so good. However, if we take a step back through the earlier out of sample period, from 1970, the picture is rather less rosy:

Now, at this point, some of you will be saying: nothing to see here – it’s obviously just curve fitting. To which I would respond that I have seen successful strategies, including several hedge fund products, with far shorter and less impressive back-tests than the initial 20-year history I showed above.

That said, would you be willing to take the risk of trading a strategy such as this one? I would not: at the back of my mind would always be the concern that the market might easily revert to the conditions that applied during the 1970s and 1980’s. I expect many investors would share that concern.

But to the point of this post: most strategies are designed around the criterion of maximizing net profit. Occasionally you might come across someone who has considered risk, perhaps in the form of drawdown, or Sharpe ratio. But, in general, it’s all about optimizing performance.

Suppose that, instead of maximizing performance, your objective was to maximize the robustness of the strategy. What criteria would you use?

In my own research, I have used a great many different objective functions, often multi-dimensional. Correlation to the perfect equity curve, net profit / max drawdown and Sortino ratio are just a few examples. But if I had to guess, I would say that the criteria that tends to produce the most robust strategies and reliable out of sample performance is the maximization of the win rate, subject to a minimum number of trades.

I am not aware of a great deal of theory on this topic. I would be interested to learn of other readers’ experience.

In this article, I will apply market timing techniques to several popular mutual funds.

The market timing approach produces annual rates of return that are 3% to 7% higher, with lower risk, than an equivalent buy and hold mutual fund investment.

Investors could in some cases have earned more than double the return achieved by holding a mutual fund investment, over a 10-year period.

Hedging strategies that use market timing signals are able to sidestep market corrections, volatile conditions and the ensuing equity drawdowns.

Hedged portfolios typically employ around 12% less capital than the equivalent buy and hold strategy.

In his post on Multi-Market Techniques for Robust Trading Strategies (http://www.adaptrade.com/Newsletter/NL-MultiMarket.htm) Michael Bryant of Adaptrade discusses some interesting approaches to improving model robustness. One is to use data from several correlated assets to build the model, on the basis that if the algorithm works for several assets with differing price levels, that would tend to corroborate the system’s robustness. The second approach he advocates is to use data from the same asset series at different bars lengths. The example he uses @ES.D at 5, 7 and 9 minute bars. The argument in favor of this approach is the same as for the first, albeit in this case the underlying asset is the same.

I like Michael’s idea in principle, but I wanted to give you a sense of what can all too easily go wrong with GP modeling, even using techniques such as multi-time frame fitting and Monte Carlo simulation to improve robustness testing.

In the chart below I have extended the analysis back in time, beyond the 2011-2012 period that Michael used to build his original model. As you can see, most of the returns are generated in-sample, in the 2011-2012 period. As we look back over the period from 2007-2010, the results are distinctly unimpressive – the strategy basically trades sideways for four years.

How do Do It Right

In my view, there is only one, safe way to use GP to develop strategies. Firstly, you need to use a very long span of data – as much as possible, to fit your model. Only in this way can you ensure that the model has encountered enough variation in market conditions to stand a reasonable chance of being able to adapt to changing market conditions in future.

Secondly, you need to use two OOS period. The first OOS span of data, drawn from the start of the data series, is used in the normal way, to visually inspect the performance of the model. But the second span of OOS data, from more recent history, is NOT examined before the model is finalized. This is really important. Products like Adaptrade make it too easy for the system designer to “cheat”, by looking at the recent performance of his trading system “out of sample” and selecting models that do well in that period. But the very process of examining OOS performance introduces bias into the system. It would be like adding a line of code saying something like:

IF (model performance in OOS period > x) do the following….

I am quite sure if I posted a strategy with a line of code like that in it, it would immediately be shot down as being blatantly biased, and quite rightly so. But, if I look at the recent “OOS” performance and use it to select the model, I am effectively doing exactly the same thing.

That is why it is so important to have a second span of OOS data that it not only not used to build the model, but also is not used to assess performance, until after the final model selection is made. For that reason, the second OOS period is referred to as a “double blind” test.

That’s the procedure I followed to build my futures daytrading strategy: I used as much data as possible, dating from 2002. The first 20% of the each data set was used for normal OOS testing. But the second set of data, from Jan 2012 onwards, was my double-blind data set. Only when I saw that the system maintained performance in BOTH OOS periods was I reasonably confident of the system’s robustness.

This further explains why it is so challenging to develop higher frequency strategies using GP. Running even a very fast GP modeling system on a large span of high frequency data can take inordinate amounts of time.

The longest span of 5-min bar data that a GP system can handle would typically be around 5-7 years. This is probably not quite enough to build a truly robust system, although if you pick you time span carefully it might be (I generally like to use the 2006-2011 period, which has lots of market variation).

For 15 minute bar data, a well-designed GP system can usually handle all the available data you can throw at it – from 1999 in the case of the Emini, for instance.

Why I don’t Like Fitting Models over Short Time Spans

The risks of fitting models to data in short time spans are intuitively obvious. If you happen to pick a data set in which the market is in a strong uptrend, then your model is going to focus on that kind of market behavior. Subsequently, when the trend changes, the strategy will typically break down.
Monte Carlo simulation isn’t going to change much in this situation: sure, it will help a bit, perhaps, but since the resampled data is all drawn from the same original data set, in most cases the simulated paths will also show a strong uptrend – all that will be shown is that there is some doubt about the strength of the trend. But a completely different scenario, in which, say, the market drops by 10%, is unlikely to appear.

One possible answer to that problem, recommended by some system developers, is simply to rebuild the model when a breakdown is detected. While it’s true that a product like MSA can make detection easier, rebuilding the model is another question altogether. There is no guarantee that the kind of model that has worked hitherto can be re-tooled to work once again. In fact, there may be no viable trading system that can handle the new market dynamics.

Here is a case in point. We have a system that works well on 10 min bars in TF.D up until around May 2012, when MSA indicates a breakdown in strategy performance.

So now we try to fit a new model, along the pattern of the original model, taking account some of the new data. But it turns out to be just a Band-Aid – after a few more data points the strategy breaks down again, irretrievably.

This is typical of what often happens when you use GP to build a model using s short span of data. That’s why I prefer to use a long time span, even at lower frequency. The chances of being able to build a robust system that will adapt well to changing market conditions are much higher.

A Robust Emini Trading System

Here, for example is a GP system build on daily data in @ES.D from 1999 to 2011 (i.e. 2012 to 2014 is OOS).

Genetic programming is an approach to letting the computer generate its own program code, rather than have a person write the program. It doesn’t specifically “find patterns” or rules within data structures. It starts with a number of randomly-constructed (as long as they are mathematically valid) sample programs, evaluates how close each one is to achieving what the desired result program should achieve, then steadily modifies the best matches to the desired target program in order to improve their match to the desired target; the original random attempts “evolve” towards a better match by natural selection, the best ones being selected to act as the basis for the next generation of attempts.

A tree representing a candidate formula could be represented as follows:

It basically shows the mathematical operations that will be used in the formula, the order in which they are applied, and what values they act on. When the EL Verifier is analysing a statement like

value1 = sin( X ) / a + b * cos( X )

it has to see work out what order the parts of the statement should be evaluated in, which a person sees immediately; effectively, the Verifier constructs the tree diagram above, so that it knows that it has to generate code to make the computer :

take the value of variable X and pass it through a call to the sin() functio

take that result, and divide it by the value of a

take the value of variable X and pass it through a call to the cos() functio

take that result and multiply it by the value of variable

take the result of step 2 and the result of step 4 and add the

that result is the value of Y for the input value of X

Tradestation optimiser would take a single such tree, defining a fixed formula, and attempt to fit it to the data by varying the values of variables a and b. A Genetic Programming optimiser could do the same, but it also has the freedom to change the mathematical operators and the merge points in the tree, and change the shape of the tree to make the formula more or less complex as well; it can adjust both the parameters to the equation and the equation itself in order to evolve it to a better result.

For a mathematical curve fit, a GP optimiser would evaluate each individual tree by applying all the measured X values to the tree’s inputs, compare each output to the measured Y values, and sum a measure of the error over all the data; that sum would be the measure of how well the current tree matches the measure data. The “genetic” part of the name derives from the way it tries to evolve the population of trees its using to find the best.

The main evolution technique is “crossover”. When two parent animals create offspring, each offspring will get part of its DNA from one parent and part from the other; improvement of the species happens if some of the offspring get DNA component combinations that suit the environment better than their parents are suited. The GP optimiser emulates this process by selecting two parent trees, and swapping a section of one of those trees with a section of tree from the other parent, to create two offspring. Eg given parent trees

representing equations

value1 = sin( X )/a + b * cos( X )

and

value1 = cos( X ) / a + b * sin( X )

the offspring might be

representing equations

value1 = sin( X )/cos( X ) + b * cos( X )

and

value1 = a / a + b * sin( X )

Those specific changes are unlikely to both be an improvement, but that’s the way with random processes; the changes made aren’t guided by any sort of principle, its just a case of “change something, anything, and see if its any better”.

A secondary change process that can be used is “mutation”, in which something about a single tree is simply changed, not swapped. This is intended to introduce diversity, so that if none of the current trees is a particularly good performer, there’s a chance that something radically better might be brought into the pool.

The push trying to steer the evolution towards a better result comes from deciding which parents are allowed to create offspring. The original idea was that all the current trees were ranked in sorted order of their fitness, the worst ones were removed from the population to be replaced by new offspring, amd the trees that were the best performing are selected to be parents – so the weak die, and the strongest breed, hoping their offspring will be at least as good as the parents.

One reservation I have about a product like Adaptrade Builder is that it doesn’t follow this original pattern. It chooses “a few” (2 by default) trees to be considered as parents, by entering a “tournament” and the best tree in the tournament is selected as a parent. This seems to me to reduce the bias towards breeding strength with strength, but I’m no expert.

Rather than being simply mathematical, Builder seems to generate tests for entry and exit orders. It takes arithmetic and comparison operators for granted, and allows trees to be built from technical indicators rather than mathematical functions like sin() and cos(). So where an EL programmer might write

if average( Close, fast ) crosses above Average( ( High + Low )/2, slow ) and CCI( length ) > overbought then buy

Builder would have a tree

from which an offspring might be generated as :

to use a Buy test

if average( ( High + Low )/2, fast ) crosses above Average( Close, slow ) and CCI( length ) > overbought then buy

The structure of the test to go long has changed, but in a random rather than the guided way a human might do when trying to develop a strategy.

One of the frustrating aspects of research and development of trading systems is that there is never enough time to investigate all of the interesting trading ideas one would like to explore. In the early 1970’s, when a moving average crossover system was considered state of the art, it was relatively easy to develop profitable strategies using simple technical indicators. Indeed, research has shown that the profitability of simple trading rules persisted in foreign exchange and other markets for a period of decades. But, coincident with the advent of the PC in the late 1980’s, such simple strategies began to fail. The widespread availability of data, analytical tools and computing power has, arguably, contributed to the increased efficiency of financial markets and complicated the search for profitable trading ideas. We are now at a stage where is can take a team of 5-6 researchers/developers, using advanced research techniques and computing technologies, as long as 12-18 months, and hundreds of thousands of dollars, to develop a prototype strategy. And there is no guarantee that the end result will produce the required investment returns.

The lengthening lead times and rising cost and risk of strategy research has obliged trading firms to explore possibilities for accelerating the R&D process. One such approach is Genetic Programming.

Early Experiences with Genetic Programming
I first came across the GP approach to investment strategy in the late 1990s, when I began to work with Haftan Eckholdt, then head of neuroscience at Yeshiva University in New York. Haftan had proposed creating trading strategies by applying the kind of techniques widely used to analyze voluminous and highly complex data sets in genetic research. I was extremely skeptical of the idea and spent the next 18 months kicking the tires very hard indeed, of behalf of an interested investor. Although Haftan’s results seemed promising, I was fairly sure that they were the product of random chance and set about devising tests that would demonstrate that.

One of the challenges I devised was to create data sets in which real and synthetic stock series were mixed together and given to the system evaluate. To the human eye (or analyst’s spreadsheet), the synthetic series were indistinguishable from the real thing. But, in fact, I had “planted” some patterns within the processes of the synthetic stocks that made them perform differently from their real-life counterparts. Some of the patterns I created were quite simple, such as introducing a drift component. But other patterns were more nuanced, for example, using a fractal Brownian motion generator to induce long memory in the stock volatility process.

It was when I saw the system detect and exploit the patterns buried deep within the synthetic series to create sensible, profitable strategies that I began to pay attention. A short time thereafter Haftan and I joined forces to create what became the Proteom Fund.

That Proteom succeeded at all was a testament not only to Haftan’s ingenuity as a researcher, but also to his abilities as a programmer and technician. Processing such large volumes of data was a tremendous challenge at that time and required a cluster of 50 cpu’s networked together and maintained with a fair amount of patch cable and glue. We housed the cluster in a rat-infested warehouse in Brooklyn that had a very pleasant view of Manhattan, but no a/c. The heat thrown off from the cluster was immense, and when combined with very loud rap music blasted through the walls by the neighboring music studios, the effect was debilitating. As you might imagine, meetings with investors were a highly unpredictable experience. Fortunately, Haftan’s intellect was matched by his immense reserves of fortitude and patience and we were able to attract investments from several leading institutional investors.

The Genetic Programming Approach to Building Trading Models

Genetic programming is an evolutionary-based algorithmic methodology which can be used in a very general way to identify patterns or rules within data structures. The GP system is given a set of instructions (typically simple operators like addition and subtraction), some data observations and a fitness function to assess how well the system is able to combine the functions and data to achieve a specified goal.

In the trading strategy context the data observations might include not only price data, but also price volatility, moving averages and a variety of other technical indicators. The fitness function could be something as simple as net profit, but might represent alternative measures of profitability or risk, with factors such as PL per trade, win rate, or maximum drawdown. In order to reduce the danger of over-fitting, it is customary to limit the types of functions that the system can use to simple operators (+,-,/,*), exponents, and trig functions. The length of the program might also be constrained in terms of the maximum permitted lines of code.

We can represent what is going on using a tree graph:

In this example the GP system is combining several simple operators with the Sin and Cos trig functions to create a signal comprising an expression in two variables, X and Y, which may be, for example, stock prices, moving averages, or technical indicators of momentum or mean reversion.
The “evolutionary” aspect of the GP process derives from the idea that an existing signal or model can be mutated by replacing nodes in a branch of a tree, or even an entire branch by another. System performance is re-evaluated using the fitness function and the most profitable mutations are retained for further generation.
The resulting models are often highly non-linear and can be very general in form.

A GP Daytrading Strategy
The last fifteen years has seen tremendous advances in the field of genetic programming, in terms of the theory as well as practice. Using a single hyper-threaded CPU, it is now possible for a GP system to generate signals at a far faster rate than was possible on Proteom’s cluster of 50 networked CPUs. A researcher can develop and evaluate tens of millions of possible trading algorithms with the space of a few hours. Implementing a thoroughly researched and tested strategy is now feasible in a matter of weeks. There can be no doubt of GP’s potential to produce dramatic reductions in R&D lead times and costs. But does it work?

To address that question I have summarized below the performance results from a GP-developed daytrading system that trades nine different futures markets: Crude Oil (CL), Euro (EC), E-Mini (ES), Gold (GC), Heating Oil (HO), Coffee (KC), Natural gas (NG), Ten Year Notes (TY) and Bonds (US). The system trades a single contract in each market individually, going long and short several times a day. Only the most liquid period in each market is traded, which typically coincides with the open-outcry session, with any open positions being exited at the end of the session using market orders. With the exception of the NG and HO markets, which are entered using stop orders, all of the markets are entered and exited using standard limit orders, at prices determined by the system

The system was constructed using 15-minute bar data from Jan 2006 to Dec 2011 and tested out-of-sample of data from Jan 2012 to May 2014. The in-sample span of data was chosen to cover periods of extreme market stress, as well as less volatile market conditions. A lengthy out-of-sample period, almost half the span of the in-sample period, was chosen in order to evaluate the robustness of the system.
Out-of-sample testing was “double-blind”, meaning that the data was not used in the construction of the models, nor was out-of-sample performance evaluated by the system before any model was selected.

Performance results are net of trading commissions of $6 per round turn and, in the case of HO and NG, additional slippage of 2 ticks per round turn.

(click on the table for a higher definition view)

The most striking feature of the strategy is the high rate of risk-adjusted returns, as measured by the Sharpe ratio, which exceeds 5 in both in-sample and out-of-sample periods. This consistency is a reflection of the fact that, while net returns fall from an annual average of over 29% in sample to around 20% in the period from 2012, so, too, does the strategy volatility decline from 5.35% to 3.86% in the respective periods. The reduction in risk in the out-of-sample period is also reflected in lower Value-at-Risk and Drawdown levels.

A decline in the average PL per trade from $25 to $16 in offset to some degree by a slight increase in the rate of trading, from 42 to 44 trades per day, on average, while daily win rate and percentage profitable trades remain consistent at around 65% and 56%, respectively.

Overall, the system appears to be not only highly profitable, but also extremely robust. This is impressive, given that the models were not updated with data after 2011, remaining static over a period almost half as long as the span of data used in their construction. It is reasonable to expect that out-of-sample performance might be improved by allowing the models to be updated with more recent data.

Benefits and Risks of the GP Approach to Trading System Development
The potential benefits of the GP approach to trading system development include speed of development, flexibility of design, generality of application across markets and rapid testing and deployment.

What about the downside? The most obvious concern is the risk of over-fitting. By allowing the system to develop and test millions of models, there is a distinct risk that the resulting systems may be too closely conditioned on the in-sample data, and will fail to maintain performance when faced with new market conditions. That is why, of course, we retain a substantial span of out-of-sample data, in order to evaluate the robustness of the trading system. Even so, given the enormous number of models evaluated, there remains a significant risk of over-fitting.

Another drawback is that, due to the nature of the modelling process, it can be very difficult to understand, or explain to potential investors, the “market hypothesis” underpinning any specific model. “We tested it and it works” is not a particularly enlightening explanation for investors, who are accustomed to being presented with a more articulate theoretical framework, or investment thesis. Not being able to explain precisely how a system makes money is troubling enough in good times; but in bad times, during an extended drawdown, investors are likely to become agitated very quickly indeed if no explanation is forthcoming. Unfortunately, evaluating the question of whether a period of poor performance is temporary, or the result of a breakdown in the model, can be a complicated process.

Finally, in comparison with other modeling techniques, GP models suffer from an inability to easily update the model parameters based on new data as it become available. Typically, as GP model will be to rebuilt from scratch, often producing very different results each time.

Conclusion
Despite the many limitations of the GP approach, the advantages in terms of the speed and cost of researching and developing original trading signals and strategies have become increasingly compelling.

Given the several well-documented successes of the GP approach in fields as diverse as genetics and physics, I think an appropriate position to take with respect to applications within financial market research would be one of cautious optimism.

This is a follow up post to my post on the Mathematics of Scalping. To illustrate the scalping methodology, I coded up a simple strategy based on the techniques described in the post.

The strategy trades a single @ES contract on 1-minute bars. The attached ELD file contains the Easylanguage code for ES scalping strategy, which can be run in Tradestation or Multicharts.

This strategy makes no attempt to forecast market direction and doesn’t consider market trends at all. It simply looks at the current levels of volatility and takes a long volatility position or a short volatility position depending on whether volatility is above or below some threshold parameters.

By long volatility I mean a position where we buy or sell the market and set a loose Profit Target and a tight Stop Loss. By short volatility I mean a position where we buy or sell the market and set a tight Profit Target and loose Stop Loss. This is exactly the methodology I described earlier in the post. The parameters I ended up using are as follows:

Long Volatility: Profit Target = 8 ticks, Stop Loss = 2 ticks
Short Volatility: Profit Target = 2 ticks, Stop Loss = 30 ticks

I have made no attempt to optimize these parameters settings, which can easily be done in Tradestation or Multicharts.

What do we mean by volatility being above our threshold level? I use a very simple metric: I take the TrueRange for the current bar and add 50% of the increase or decrease in TrueRange over the last two bars. That’s my crude volatility “forecast”.

The final point to explain is this: let’s suppose our volatility forecast is above our threshold level, so we know we want to be long volatility. Ok, but do we buy or sell the ES? One approach ia to try to gauge the direction of the market by estimating the trend. Not a bad idea, by any means, although I have argued that volatility drowns out any trend signal at short time frames (like 1 minute, for example). So I prefer an approach that makes no assumptions about market direction.

In this approach what we do is divide volatility into upsideVolatility and downsideVolatility. upsideVolatility uses the TrueRange for bars where Close > Close[1]. downsideVolatility is calculated only for bars where Close < Close[1]. This kind of methodology, where you calculate volatility based on the sign of the returns, is well known and is used in performance measures like the Sortino ratio. This is like the Sharpe ratio, except that you calculate the standard deviation of returns using only days in which the market was down. When it’s calculated this way, standard deviation is known as the (square root of the) semi-variance.

Anyway, back to our strategy. So we calculate the upside and downside volatilities and test them against our upper and lower volatility thresholds.

The decision tree looks like this:

LONG VOLATILITY
If upsideVolatilityForecast > upperVolThrehold, buy at the market with wide PT and tight ST (long market, long volatility)
If downsideVolatilityForecast > upperVolThrehold, sell at the market with wide PT and tight ST (short market, long volatility)

SHORT VOLATILITY
If upsideVolatilityForecast < lowerVolThrehold, sell at the Ask on a limit with tight PT and wide ST (short market, short volatility)
If downsideVolatilityForecast < lowerVolThrehold, buy at the Bid on a limit with tight PT and wide ST (long market, short volatility)

NOTE THE FOLLOWING CAVEATS. DO NOT TRY TO TRADE THIS STRATEGY LIVE (but use it as a basis for a tradable strategy)

1. The strategy makes the usual TS assumption about fill rates, which is unrealistic, especially at short intervals like 1-minute.
2. The strategy allows fees and commissions of $3 per contract, or $6 per round turn. Your trading costs may be higher than this.
3. Tradestation is unable to perform analysis at the tick level for a period as long at the one used here (2000 to 2014). A tick by tick analysis would likely show very different results (better or worse).
4. The strategy is extremely lop-sided: the great majority of the profits are made on the long side and the Win Rates and Profit Factors are very different for long trades vs short trades. I suspect this would change with a tick by tick analysis. But it also may be necessary to add more parameters so that long trades are treated differently from short trades.
5. No attempt has been made to optimize the parameters.
6 This is a daytading strategy that will exit the market on close.

So with all that said here are the results.

As you can see, the strategy produces a smooth, upward sloping equity curve, the slope of which increases markedly during the period of high market volatility in 2008.
Net profits after commissions for a single ES contract amount to $243,000 ($3.42 per contract) with a win rate of 76% and Profit Factor of 1.24.

This basic implementation would obviously require improvement in several areas, not least of which would be to address the imbalance in strategy profitability on the short vs long side, where most of the profits are generated.

In this post I am going to explore how, starting from the assumption of a stable, Gaussian distribution in a returns process, we evolve to a system that displays all the characteristics of empirical market data, notably time-dependent moments, high levels of kurtosis and fat tails. As it turns out, the only additional assumption one needs to make is that the market is periodically disturbed by the random arrival of news.

NOTE: if you are unable to see the Mathematica models below, you can download the free Wolfram CDF player and you may also need this plug-in.

You can also download the complete Mathematica CDF file here.

Stationarity

A stationary process is one that evolves over time, but whose probability distribution does not vary with time. As the word implies, such a process is stable. More formally, the moments of the distribution are independent of time.

Let’ s assume we are dealing with such a process that have constant mean μ and constant volatility (standard deviation) σ.

Φ=NormalDistribution[μ,σ]

Here are some examples of Normal probability distributions, with constant mean μ = 0 and standard deviation σ ranging from 0.75 to 2

If we assume for the moment that such a process is an adequate description of an asset returns process, we can simulate the evolution of a price process as follows :

ListPlot[prices=Accumulate[sampleData]]

An Empirical Distribution

Lets take a look at a real price series, comprising 1 – minute bar data in the June ‘ 14 E – Mini futures contract.

As with our simulated price process, it is clear that the real price process for Emini futures is also non – stationary.

What about the returns process?

ListPlot[returnsES]

Notice the banding effect in returns, which results from having a fixed, minimum price move of $12 .50, rather than a continuous scale.

The empirical returns distribution doesn’ t appear to be Gaussian – the distribution is much more peaked than a standard Normal distribution with the same mean and standard deviation. And the higher moments don’t fit the Normal model either – the empirical distribution has positive skew and a kurtosis that is almost 9x greater than a Gaussian distribution. The latter signifies what is often referred to as “fat tails”: the distribution has much greater weight in the tails than a standard Normal distribution, indicating a much greater likelihood of an extreme value than a Normal distribution would predict.

Non-Stationarity: Two States

Non – stationarity arises when one or more of the moments of a distribution vary over time. Let’s take a look at how that can arise, and its effects.Suppose we have a Gaussian returns process for which the mean, or drift, or trend, fluctuates over time.

Let’s consider a simple example where the process drift is μ1 and volatility σ1 for most of the time and then for some proportion of time k, we get addition drift μ2 and volatility σ2. In other words we have:

This simple model fits a scenario in which we suppose that the returns process spends most of its time in State 1, in which is Normally distributed with drift is μ1 and volatility σ1, and suffers from the occasional “shock” which propels the systems into a second State 2, in which its distribution is a combination of its original distribution and a new Gaussian distribution with different mean and volatility.

Let’ s suppose that we sample the combined process y = Φ1 + k Φ2. What distribution would it have? We can represent this is follows :

The result is just another Normal distribution. Depending on the incidence k, y will follow a Gaussian distribution whose mean and variance depend on the mean and variance of the two Normal distributions being mixed. The resulting distribution in State 2 may have higher or lower drift and volatility, but it is still Gaussian, with constant kurtosis of 3.

In other words, the system y will be non-stationary, because the first and second moments change over time, depending on what state it is in. But the form of the distribution is unchanged – it is still Gaussian. There are no fat-tails.

Non – Stationarity : Random States

In the above example the system moved between states in a known, predictable way. The “shocks” to the system were not really shocks, but transitions. But that’s not how financial markets behave: markets move from one state to another in an unpredictable way, with the arrival of news.

We can simulate this situation as follows. Using the former model as a starting point, lets now relax the assumption that the incidence of the second state, k, is a constant. Instead, let’ s assume that k is itself a random variable. In other words we are going to now assume that our system changes state in a random way. How does this alter the distribution?

An appropriate model for λ might be a Poisson process, which is often used as a model for unpredictable, discrete events, ranging from bus arrivals to earthquakes. PDFs of Poisson distributions with means λ=5, 10 and 20 are shown in the chart below. These represent probability distributions for processes that have mean arrivals of 5, 10 or 20 events.

The first two moments of the distribution are as follows :

Through[{Mean,StandardDeviation}[y]]

As before, the mean and standard deviation of the distribution are going to vary, depending on the state of the system, and the mean arrival rate of shocks, . But what about kurtosis? Is it still constant?

Kurtosis[y]

Emphatically not! The fourth moment of the distribution is now dependent on the drift in the second state, the volatilities of both states and the mean arrival rate of shocks, λ.

Let’ s look at a specific example. Assume that in State 1 the process has volatility of 7.5 %, with zero drift, and that the shock distribution also has zero drift with volatility of 65 %. If the mean incidence rate of shocks λ = 10 %, the distribution kurtosis is close to that seen in the empirical distribution for the E-Mini.

Kurtosis[y] /.{σ1→0.075,μ2→0,σ2→0.65,λ→0.1}

{35.3551}

More generally :

ListLinePlot[Flatten[Kurtosis[y]/.Table[{σ1→0.075,μ2→0,σ2→0.65,λ→i/20},{i,1,20}]],PlotLabel→Style["Kurtosis vs Mean Shock Arrival Rate", FontSize→18],AxesLabel->{“Incidence Rate (%)”, “Kurtosis”},Filling→Axis, ImageSize→Large]

Thus we can see how, even if the underlying returns distribution is Gaussian in form, the random arrival of news “shocks” to the system can induce non – stationarity in overall drift and volatility. It can also result in fat tails. More specifically, if the arrival of news is stochastic in nature, rather than deterministic, the process may exhibit far higher levels of kurtosis than in its original Gaussian state, in which the fourth moment was a constant level of 3.

Jump Diffusion

Nobel – prize winning economist Robert Merton extended this basic concept to the realm of stochastic calculus.

In Merton’s jump diffusion model, the stock price follows the random process

∂S_{t} / S_{t} =μdt + σdW_{t}+(J-1)dN_{t}

The first two terms are familiar from the Black–Scholes model : drift rate μ, volatility σ, and random walk W_{t} (Wiener process).The last term represents the jumps :J is the jump size as a multiple of stock price, while N_{t} is the number of jump events that have occurred up to time t.is assumed to follow the Poisson process.

PDF[PoissonDistribution[λt]]

where λ is the average frequency with which jumps occur.

The jump size J follows a log – normal distribution

PDF[LogNormalDistribution[m, ν], s]

where m is the average jump size and vis the volatility of the jump size.

In the jump diffusion model, the stock price S_{t} follows the random process dS_{t}/S_{t}=μ dt+σ dW_{t}+(J-1) dN(t), which comprises, in order, drift, diffusive, and jump components. The jumps occur according to a Poisson distribution and their size follows a log-normal distribution. The model is characterized by the diffusive volatility σ, the average jump size J (expressed as a fraction of S_{t}), the frequency of jumps λ, and the volatility of jump size ν.

The Volatility Smile

The “implied volatility” corresponding to an option price is the value of the volatility parameter for which the Black-Scholes model gives the same price. A well-known phenomenon in market option prices is the “volatility smile”, in which the implied volatility increases for strike values away from the spot price. The jump diffusion model is a generalization of Black–Scholes in which the stock price has randomly occurring jumps in addition to the random walk behavior. One of the interesting properties of this model is that it displays the volatility smile effect. In this Demonstration, we explore the Black–Scholes implied volatility of option prices (equal for both put and call options) in the jump diffusion model. The implied volatility is modeled as a function of the ratio of option strike price to spot price.

NOTE: if you are unable to see the Mathematica models below, you can download the free Wolfram CDF player and you may also need this plug-in.

You can also download the complete Mathematica CDF file here.

In this post I want to explore aspects of scalping, a type of strategy widely utilized by high frequency trading firms.

I will define a scalping strategy as one in which we seek to take small profits by posting limit orders on alternate side of the book. Scalping, as I define it, is a strategy rather like market making, except that we “lean” on one side of the book. So, at any given time, we may have a long bias and so look to enter with a limit buy order. If this is filled, we will then look to exit with a subsequent limit sell order, taking a profit of a few ticks. Conversely, we may enter with a limit sell order and look to exit with a limit buy order.
The strategy relies on two critical factors:

(i) the alpha signal which tells us from moment to moment whether we should prefer to be long or short
(ii) the execution strategy, or “trade expression”

In this article I want to focus on the latter, making the assumption that we have some kind of alpha generation model already in place (more about this in later posts).

There are several means that a trader can use to enter a position. The simplest approach, the one we will be considering here, is simply to place a single limit order at or just outside the inside bid/ask prices – so in other words we will be looking to buy on the bid and sell on the ask (and hoping to earn the bid-ask spread, at least).

One of the problems with this approach is that it is highly latency sensitive. Limit orders join the limit order book at the back of the queue and slowly works their way towards the front, as earlier orders get filled. Buy the time the market gets around to your limit buy order, there may be no more sellers at that price. In that case the market trades away, a higher bid comes in and supersedes your order, and you don’t get filled. Conversely, yours may be one of the last orders to get filled, after which the market trades down to a lower bid and your position is immediately under water.

This simplistic model explains why latency is such a concern – you want to get as near to the front of the queue as you can, as quickly as possible. You do this by minimizing the time it takes to issue and order and get it into the limit order book. That entails both hardware (co-located servers, fiber-optic connections) and software optimization and typically also involves the use of Immediate or Cancel (IOC) orders. The use of IOC orders by HFT firms to gain order priority is highly controversial and is seen as gaming the system by traditional investors, who may end up paying higher prices as a result.

Another approach is to layer limit orders at price points up and down the order book, establishing priority long before the market trades there. Order layering is a highly complex execution strategy that brings addition complications.

Let’s confine ourselves to considering the single limit order, the type of order available to any trader using a standard retail platform.

As I have explained, we are assuming here that, at any point in time, you know whether you prefer to be long or short, and therefore whether you want to place a bid or an offer. The issue is, at what price do you place your order, and what do you do about limiting your risk? In other words, we are discussing profit targets and stop losses, which, of course, are all about risk and return.

Risk and Return in Scalping

Lets start by considering risk. The biggest risk to a scalper is that, once filled, the market goes against his position until he is obliged to trigger his stop loss. If he sets his stop loss too tight, he may be forced to exit positions that are initially unprofitable, but which would have recovered and shown a profit if he had not exited. Conversely, if he sets the stop loss too loose, the risk reward ratio is very low – a single loss-making trade could eradicate the profit from a large number of smaller, profitable trades.

Now lets think about reward. If the trader is too ambitious in setting his profit target he may never get to realize the gains his position is showing – the market could reverse, leaving him with a loss on a position that was, initially, profitable. Conversely, if he sets the target too tight, the trader may give up too much potential in a winning trade to overcome the effects of the occasional, large loss.

It’s clear that these are critical concerns for a scalper: indeed the trade exit rules are just as important, or even more important, than the entry rules. So how should he proceed?

Theoretical Framework for Scalping

Let’s make the rather heroic assumption that market returns are Normally distributed (in fact, we know from empirical research that they are not – but this is a starting point, at least). And let’s assume for the moment that our trader has been filled on a limit buy order and is looking to decide where to place his profit target and stop loss limit orders. Given a current price of the underlying security of X, the scalper is seeking to determine the profit target of p ticks and the stop loss level of q ticks that will determine the prices at which he should post his limit orders to exit the trade. We can translate these into returns, as follows:

to the upside: Ru = Ln[X+p] – Ln[X]

and to the downside: Rd = Ln[X-q] – Ln[X]

This situation is illustrated in the chart below.

The profitable area is the shaded region on the RHS of the distribution. If the market trades at this price or higher, we will make money: p ticks, less trading fees and commissions, to be precise. Conversely we lose q ticks (plus commissions) if the market trades in the region shaded on the LHS of the distribution.

Under our assumptions, the probability of ending up in the RHS shaded region is:

probWin = 1 – NormalCDF(Ru, mu, sigma),

where mu and sigma are the mean and standard deviation of the distribution.

The probability of losing money, i.e. the shaded area in the LHS of the distribution, is given by:

probLoss = NormalCDF(Rd, mu, sigma),

where NormalCDF is the cumulative distribution function of the Gaussian distribution.

The expected profit from the trade is therefore:

Expected profit = p * probWin – q * probLoss

And the expected win rate, the proportion of profitable trades, is given by:

WinRate = probWin / (probWin + probLoss)

If we set a stretch profit target, then p will be large, and probWin, the shaded region on the RHS of the distribution, will be small, so our winRate will be low. Under this scenario we would have a low probability of a large gain. Conversely, if we set p to, say, 1 tick, and our stop loss q to, say, 20 ticks, the shaded region on the RHS will represent close to half of the probability density, while the shaded LHS will encompass only around 5%. Our win rate in that case would be of the order of 91%:

WinRate = 50% / (50% + 5%) = 91%

Under this scenario, we make frequent, small profits and suffer the occasional large loss.

So the critical question is: how do we pick p and q, our profit target and stop loss? Does it matter? What should the decision depend on?

Modeling Scalping Strategies

We can begin to address these questions by noticing, as we have already seen, that there is a trade-off between the size of profit we are hoping to make, and the size of loss we are willing to tolerate, and the probability of that gain or loss arising. Those probabilities in turn depend on the underlying probability distribution, assumed here to be Gaussian.

Now, the Normal or Gaussian distribution which determines the probabilities of winning or losing at different price levels has two parameters – the mean, mu, or drift of the returns process and sigma, its volatility.

Over short time intervals the effect of volatility outweigh any impact from drift by orders of magnitude. The reason for this is simple: volatility scales with the square root of time, while the drift scales linearly. Over small time intervals, the drift becomes un-noticeably small, compared to the process volatility. Hence we can assume that mu, the process mean is zero, without concern, and focus exclusively on sigma, the volatility.

What other factors do we need to consider? Well there is a minimum price move, which might be 1 tick, and the dollar value of that tick, from which we can derive our upside and downside returns, Ru and Rd. And, finally, we need to factor in commissions and exchange fees into our net trade P&L.

Here’s a simple formulation of the model, in which I am using the E-mini futures contract as an exemplar.

For the ES contract we have a min price move of 0.25 and the tick value is $12.50. Notice that we scale annual volatility to the size of the period we are trading (15 minute bars, in the following example).

Scenario Analysis

Let’s take a look at how the expected profit and win rate vary with the profit target and stop loss limits we set. In the following interactive graphics, we can assess the impact of different levels of volatility on the outcome.

Expected Profit by Bar Size and Volatility

Expected Win Rate by Volatility

Notice to begin with that the win rate (and expected profit) are very far from being Normally distributed – not least because they change radically with volatility, which is itself time-varying.

For very low levels of volatility, around 5%, we appear to do best in terms of maximizing our expected P&L by setting a tight profit target of a couple of ticks, and a stop loss of around 10 ticks. Our win rate is very high at these levels - around 90% or more. In other words, at low levels of volatility, our aim should be to try to make a large number of small gains.

But as volatility increases to around 15%, it becomes evident that we need to increase our profit target, to around 10 or 11 ticks. The distribution of the expected P&L suggests we have a couple of different strategy options: either we can set a larger stop loss, of around 30 ticks, or we can head in the other direction, and set a very low stop loss of perhaps just 1-2 ticks. This later strategy is, in fact, the mirror image of our low-volatility strategy: at higher levels of volatility, we are aiming to make occasional, large gains and we are willing to pay the price of sustaining repeated small stop-losses. Our win rate, although still well above 50%, naturally declines.

As volatility rises still further, to 20% or 30%, or more, it becomes apparent that we really have no alternative but to aim for occasional large gains, by increasing our profit target and tightening stop loss limits. Our win rate under this strategy scenario will be much lower – around 30% or less.

Non – Gaussian Model

Now let’s address the concern that asset returns are not typically distributed Normally. In particular, the empirical distribution of returns tends to have “fat tails”, i.e. the probability of an extreme event is much higher than in an equivalent Normal distribution.

A widely used model for fat-tailed distributions in the Extreme Value Distribution. This has pdf:

In order to set the parameters of the EVD, we need to arrange them so that the mean and variance match those of the equivalent Gaussian distribution with mean = 0 and standard deviation . hence:

The code for a version of the model using the GED is given as follows

We can now produce the same plots for the EVD version of the model that we plotted for the Gaussian versions :

Expected Profit by Bar Size and Volatility – Extreme Value Distribution

Expected Win Rate by Volatility – Extreme Value Distribution

Next we compare the Gaussian and EVD versions of the model, to gain an understanding of how the differing assumptions impact the expected Win Rate.

Expected Win Rate by Stop Loss and Profit Target

As you can see, for moderate levels of volatility, up to around 18 % annually, the expected Win Rate is actually higher if we assume an Extreme Value distribution of returns, rather than a Normal distribution.If we use a Normal distribution we will actually underestimate the Win Rate, if the actual return distribution is closer to Extreme Value.In other words, the assumption of a Gaussian distribution for returns is actually conservative.

Now, on the other hand, it is also the case that at higher levels of volatility the assumption of Normality will tend to over – estimate the expected Win Rate, if returns actually follow an extreme value distribution. But, as indicated before, for high levels of volatility we need to consider amending the scalping strategy very substantially. Either we need to reverse it, setting larger Profit Targets and tighter Stops, or we need to stop trading altogether, until volatility declines to normal levels.Many scalpers would prefer the second option, as the first alternative doesn’t strike them as being close enough to scalping to justify the name.If you take that approach, i.e.stop trying to scalp in periods when volatility is elevated, then the differences in estimated Win Rate resulting from alternative assumptions of return distribution are irrelevant.

If you only try to scalp when volatility is under, say, 20 % and you use a Gaussian distribution in your scalping model, you will only ever typically under – estimate your actual expected Win Rate.In other words, the assumption of Normality helps, not hurts, your strategy, by being conservative in its estimate of the expected Win Rate.

If, in the alternative, you want to trade the strategy regardless of the level of volatility, then by all means use something like an Extreme Value distribution in your model, as I have done here.That changes the estimates of expected Win Rate that the model produces, but it in no way changes the structure of the model, or invalidates it.It’ s just a different, arguably more realistic set of assumptions pertaining to situations of elevated volatility.

Monte-Carlo Simulation Analysis

Let’ s move on to do some simulation analysis so we can get an understanding of the distribution of the expected Win Rate and Avg Trade PL for our two alternative models. We begin by coding a generator that produces a sample of 1,000 trades and calculates the Avg Trade PL and Win Rate.

Now we can generate a random sample of 10, 000 simulation runs and plot a histogram of the Win Rates, using, for example, ES on 5-min bars, with a PT of 2 ticks and SL of – 20 ticks, assuming annual volatility of 15 %.

The setting of optimal profit targets are stop loss limits depend critically on the volatility of the underlying, and needs to be handled dynamically, depending on current levels of market volatility

At low levels of volatility we should set tight profit targets and wide stop loss limits, looking to make a high percentage of small gains, of perhaps 2-3 ticks.

As volatility rises, we need to reverse that position, setting more ambitious profit targets and tight stops, aiming for the occasional big win.

One of the issues that comes up regularly is how, as an investor or other interested party, one can protect oneself from unscrupulous scam artists posing as professional traders or money managers. This is a particular problem on web sites featuring trader forums, where individuals with unverified track records claiming stellar trading histories use their purported trading “prowess” to try to impress and intimidate other participants, usually impressionable newbies. The purpose of this post is to provide some guidance to help investors, traders and other fellow travelers sort the wheat from the chaff. We’ll be doing some forensic analysis on the track record for a strategy in NG futures that one such character recently posted in one of these forums, as a classic example of the kind of fakery I am describing.

One thing you should understand about scam artists operating on forums, is that they don’t work alone: usually they have a bunch of groupies who will shill for them at every opportunity and who will try to shout down any investigative questioning. Don’t be deterred. These know-it-alls are usually just ignorant dupes, who understand no more about trading than the scam artist. They may just as easily be fellow-scam artists themselves.

THE FIRST BIG RED FLAG: UNWILLINGNESS TO PRODUCE A TRACK RECORD
Anyone claiming to be a CTA or professional money manager (or whose shills claim he is one) has to have a track record that is freely available in the public domain. So how does a scam artist overcome a challenge to produce it? He will claim that he “can’t advertise”, or make some other, similar excuse. Don’t accept that at face value. Ask him to PM it to you. If he won’t, there’s already a high probability he’s a con artist.

THE SECOND BIG RED FLAG: CURVE FITTING
Let’s say our suspect meets the challenge and produces a track record. Ideally this will be an audited P&L statement, but let’s assume for the purposes of this discussion that he produces something along the lines of the Performance Reports produced by a product like Tradestation or MultiCharts, i.e. we are dealing with a simulated back-test.

If your suspect produces a back-test, you can be pretty sure it’s going to look good – otherwise he wouldn’t produce it. The task now is to dig into those reports to spot the red flags that give clues as to whether it might be fake.
Now of course any trading system is going to make assumptions – about fill rates, slippage, commissions, capacity etc. All that is fine, as long as the assumptions are clearly stated. You might want to challenge any or all of the assumptions, and the trader may disagree with you about some or all of them. That’s perfectly ok – it’s an honest, open discussion about a set of investment assumptions that have been revealed at the outset.

But here is what is NOT ok: any opacity about which data was used to build the trading model and which data was used to test it. The former, the in-sample (IS) data set, used to construct the model, must be entirely separate and distinct from the out-of-sample (OOS) data set. It is trivially easy using a tool like Tradestation to produce a trading system that shows stellar results in-sample, but which will immediately crash and burn when it is used in live trading. This is known as curve-fitting. And it’s by far the most common method by which scam artists try to dupe investors.

In order to demonstrate the robustness of the system prior to risking real money, a genuine trader will test his system OOS and show you the results. What you are looking for ideally is congruity between the IS and OOS results. Now by congruity, I don’t mean that they should be identical. Far from it – markets evolve and strategy performance will vary over time. But what you are hoping is that the key performance metrics in the OOS and IS periods, such as annual returns, Sharpe ration, PNL per contract, profit ratio and win rate, will be comparable. At the very least, you would like to be able to identify some portion of the IS data set for which the strategy performance characteristics are similar to those in the OOS period.

Any – I mean ANY – ambiguity or lack of clarity about which data was used to build the model and which was used for OOS testing is a HUGE red flag. Chances are, your scam artist is already trying to fudge the issue that he curve-fitted the system.
This was the case in the recent forum post we are using as a test case. The trader made no attempt whatsoever to clarify which data was used for model development and which for testing. Immediately, I was suspicious and began looking for other evidence of curve fitting. It didn’t take me long to find it.

THE THIRD BIG RED FLAG: THE EQUITY CURVE
The first item I turned to in the performance reports was the equity curve and I immediately spotted two rather large clues that I was dealing with a fake.

The first clue was the large sign on the chart labelled “live start date”. What does this mean? This is a back-test, so all of the results are theoretical, including those after the supposed “live start date” sometime in 2013. What the faker is trying to do is imply the part of the equity curve shown after that date indicate actual performance results. He doesn’t actually claim this, so he has plausible deniability if you call him on it (“I said it was just a back test”). But he hopes that you won’t, and that, by default, you’ll accept these results are real. But they aren’t.

The second clue of fakery is much more important: the equity curve itself. When someone shows you and equity curve like the one reported by this trader, rising in a straight line from the lower left to upper right quadrants, you can be 99% confident that you are dealing with a fake.
You see, in finance there are almost never any straight lines. They are as rare as unicorns. Especially when it comes to strategy performance. They only time you will EVER see an equity curve like this is when you are looking at the equity curve of (i) a high frequency market making trading system or (ii) a fake, produced by curve fitting a strategy to the ENTIRE data set.
And this strategy was not high frequency – as we shall see, it operated on 15 minute bars, holding positions overnight.

THE FOURTH BIG RED FLAG: GOD’s EQUITY CURVE
I said that straight line equity curve were extremely rare. In fact, even God’s equity curve isn’t often a straight line. What does that mean?

Suppose you had a strategy that could predict with 100% accuracy whether the market would go up or down over the next bar (whether you are using daily bars, or 15 minute bars, as in our example). The system would buy (or hold) when the market was forecast to rise, and sell when the market was predicted to fall. What would the performance of such a perfect system look like? Pretty stellar, obviously. And most people would guess that the system’s equity curve would be a straight line, or maybe even exponential in shape. In fact that’s typically not the case. God’s equity curve will be sloped and kinked, just like any other equity curve. And if your suspect’s equity curve is real, it should show some commonality with God’s equity curve, by which I mean it should show changes in slope and level that reflect those seen in the perfect equity curve.

What does God’s Equity Curve look like in NG futures?

As you can see it’s not straight. In fact it’s concave. So a REAL equity curve should have similar characteristics, like this one, for example:

As you can see, the equity curve of the real trading system track’s God’s Equity Curve, albeit at a much lower level. It’s concave, with an upswing during the final few months of trading, just like God’s. That’s a good sign that the strategy back-test is very likely genuine (which it is – I produced it).

Why is Gods’ Equity Curve the shape it is? The answer will vary from market to market. In the case of NG, the suggestion is that the market is becoming more efficient: simple trading strategies based on technical indicators work less well than they did five years ago. We have seen something very similar in F/X markets. During the 1970’s and 1980’s when Soros was active in the field, simple strategies like moving average crossovers made great returns, but these entirely dissipated in the 1990’s, with the advent of widely available computing power.

THE FIFTH BIG RED FLAG: THE SHILL SHOUTDOWN
When I posted my analysis, which clearly indicated fakery by this well known forum participant, I was immediately flamed by one of his supporters who shouted something to the effect that (i) everyone knows that the downward slope of God’s Equity Curve was caused by volatility and (ii) the star trader, unlike God, or me, knows about position sizing.

This attempt at misdirection in the face of awkward facts is a classic sign of fakery. What distinguishes the shill post is:

(i) Immediacy – clearly no attempt has been made to evaluate the argument or analysis. The shill simply attempts to drown out the critic with a lot of noise, as quickly as possible.

(ii) Plausibility – shills will throw around terms that lend plausibility to their objection, but which after a moment’s reflection are entirely irrelevant or, as in this case, detrimental to their own cause.

(iii) Invective – the more intemperate the post, the more likely the shill is simply trying to provide cover for the faker.

So let’s take a moment to dispose of the plausible sounding objections posted by the shill in this example.
I am going to take it as read that everyone understands that trading profitability is positively correlated with volatility. There is a huge amount of empirical research supporting that finding, but to keep it simple we can appeal to one of the cornerstones of modern finance: risk and return. The higher the volatility, i.e. the greater the risk, the greater the return traders and investors in the markets will require on their capital. This is a principle of modern financial theory that even a graduate of the Scranton college of fine art should be expected to appreciate.

So what’s the story with NG volatility? You can see the time series of NG volatility in the chart below. One feature stands out above all others: the upward slope of the curve. NG volatility has RISEN over the sample period from 2008 to 2014. Consequently, returns from trading NG futures should also have RISEN rather than fallen. One thing we can say for sure, whatever caused the concave shape in God’s Equity Curve in NG futures, it was NOT volatility!

Turning to the shill’s next, plausible sounding, but dubious “explanation”, position sizing: this really is completely irrelevant. Because, as we shall see from an examination of the performance report, the track record was created by trading a constant one-lot! So this was just an attempt to sound “sophisticated” by someone trying to misdirect the reader away from the increasingly obvious evidence of fakery.

THE SIXTH BIG RED FLAG: LOW DRAWDOWNS AND OVERNIGHT GAP RISK
One of the highly unusual features of our faker’s equity curve is it’s exceptional smoothness. Low volatility in the equity curve is, in and of itself, an indicator the track record results from curve fitting. But we can get even more insight by digging into the performance report, shown below.

As you can see from the second page of the report, the strategy holds positions for an average of 57 15-minute bars, equivalent to slightly over 14 hours. So this is a low frequency strategy that takes overnight risk. Now, as any trader will know, overnight gap risk in a product like NG can be very significant and likely to be produce much larger drawdowns over a 5 year period than the $8,470 reported here.

The only other possible explanation is that the strategy is traded continuously through both day and night sessions. But this is not only itself improbable, it gives rise to another implausibility: liquidity in the overnight session is so poor that the strategy is unlikely to be able to trade more than 1-2 contracts, at most. This would be of little value to a CTA, or its customers, whatever the star trader’s protestations that his “clients are happy”.

There is no plausible way to resolve the disconnection between the low drawdown, overnight gap risk and market illiquidity. The most plausible explanation: the back-test is a curve fitting exercise.

THE SEVENTH AN FINAL BIG RED FLAG: INCONSISTENCY BETWEEN PERFORMANCE METRICS
As any experienced strategy developer knows, you can get some of the things you want, but you can never achieve all of them. Amongst the desirable features to be maximized are
• Profit factor
• Average PNL per contract
• Percentage win rate

There is a trade-off between the features. A high PNL per contract typically means you are trading less frequently, with longer hold periods, and consequently the percentage win rate tends to be lower. Alternatively, you can increase the win rate, at the cost of lowering the average PNL per contract and/or the profit factor. And so on.

This strategy purports to have it all: a high average PNL per contract resulting from low frequency trading, coupled with good percentage win rate of over 50% and profit factor. A win rate of much over 40% is highly unusual for a momentum strategy entering and exiting with market or stop orders – and its almost inconceivable for a strategy with a PNL per contract and profit factor as large as suggested here.

CONCLUSION
This back-test fails the sniff test on so many levels, I would rate the chance of it being real as less than 1 in 1000.
The final, conclusive proof of fakery is that the “star trader” responsible for producing the report was unable and/or unwilling to attempt to answer even a single one of the criticisms.

So, be warned. If you see forum members banding about track records like this one, you can be sure that they and their strategies are likely to be fake, and not to be trusted.