We can discuss pros and contras of system trading for a long time. But it is obvious that there is a significant number of traders for which the only way to cope with their emotions and, therefore, survive on the market is to use formalized conditions of opening/closing positions. However, the effective use of any trading approach implies a distinct realization of both its advantages and limitations.

The main limitation for automated trading systems is the fact that they are only efficient under certain conditions on the markets. If you trade a trend-following system, it will be helpless on trendless markets; if you are armed with counter-trend systems, you will suffer from very essential losses on the markets characterized by strong trending. Even the most beautiful woman of France cannot give more than she can give. The same is with systems - trading them, you cannot earn more than allowed by the market as of the present time.

This is why one of important elements of system trading is the choice of markets to trade on - their nature must match your trading systems. However, nobody would ever guarantee that the nature of the market traded will not change later. E.g., you are trading trend-following systems, and your market has entered a long-termed sideways trend. If, in these changed market conditions, you will continue following the signals of your trading systems time by time, once having got into a "ripsaw", you will obtain one loss by another. As soon as the market stops matching your trading system, trading under this system will result in your systematic losing your capital.

The deposit burns away like a candle blown by the wind, and the trader's only remaining source is to make a helpless gesture perplexedly and blame the "bad" system for everything. The thing that hurts most of all is that this "bad" system comes to life after a certain period of time and starts bringing good profits - but not to you.

It would be quite reasonable to trade a system when the market corresponds with it, and to stop it or reduce the volumes of positions opened in the periods of time, when the market discontinues corresponding with your system.

However, how can we know whether the market corresponds with our trading system or not? The analysis of capital curve can give us an answer. There is a whole group of methods that analyze the efficiency of a trading system at latest using the capital curve changes and allow us on the basis of this analysis to filter some trades, the risk for which exceeds the limits determined by the trader.

Applying filters on the basis of capital curve analysis allows us to reduce the trading risk. However, the reverse of this medal is decreasing in return of the system, because its trades have, all in all, a positive expected payoff, and the group of riskier trades selected using filters usually has a positive expected payoff, as well. This is why the filtering of trades must be aimed at achieving an asymmetric decrease in risk and in return, the risk being reduced more essentially than the return. This asymmetric risk reduction improves the total efficiency of the system by lowering the return dispersion and the depth of maximal drawdowns1 related to this parameter.

As a result of correct filtering of riskier trades, a new trading system comes into being, in which the return decreases to the smaller degree than the values of drawdowns. Such system provides a trader with several choices, one of which proposes him or her to consider the decreased return to be a reasonable rate for the decreased depth of drawdowns. Another choice would be the use an increased leverage in the new system, which would allow us to get essentially higher return at the same expected depth of drawdowns as in the initial system without trades filtering.

We would like to offer our readers a method that allows you to numerically assess the correspondence of a trading system with the market conditions. The method is based on finding the trading system/market conformity coefficient. However, let me write a few words about the theory underlying this method first.

Zero Return Curve

A trading system that produces neither profits nor losses complies with the following requirement (considering the commission fee):

%Win * Avg.Win = %Loss * Avg.Loss

where:

• %Win - percentage of profitable trades;
• Avg.Win - the value of an average profitable trade;
• %Loss - percentage of losing trades;
• Avg. Loss - the value of an average losing trade.

Since %Loss = 100% - %Win, the expression can be rewritten as follows (Formula 1):

Avg.Win / Avg.Loss = (100% - %Win) / %Win

Having solved this equation for the different values of %Win and plotted this solution onto the chart, we will obtain the zero return curve. The area below this curve will correspond with the set of losing trading strategies, whereas the area above the curve will do with the set of profitable trading strategies (Fig. 1).

Fig.1. Zero Return Curve.

Two considerations are obvious here: first, the higher the trading system test results are located above the curve, the better. Second, if we take the results of the trading system in some moving time window, they will "migrate" moving now to one direction then to another one, sometimes even falling below the zero return curve. This "migration" reflects the degree of system/market concordance changes - in the periods of good correspondence with the market, the results will be located highly above the curve, whereas they will surely "dive" below the curve when the system "breaks down".

When choosing a trading system to be used in real trading, we usually want its parameters (Avg.Win/Avg.Loss and %Win) to be at a certain safe distance from the zero return curve, so that chance fluctuations of return don't put these parameters into the loss-making area. However, what distance can be considered as relatively safe? Generally, the choice is of very subjective nature. We are imposed upon the approached described by Ryan Jones in his book The Trading Game: "The standard for a very good system is to generally take a 10 percent lower winning percentage while maintaining a 1.0 better win/loss ratio over the breakeven point. If this combination exists, you are about as close to a Holy Grail strategy as you are going to get."

Let's express this safety criterion in a formula:

Avg.Win/Avg.Loss = (100% -( %Win-10%)) / (%Win-10%) + 1

Now let's solve this equation and plot the values of the obtained safety curve on the chart (Fig. 2):

Fig. 2. Zero Return Curve and Safety Curve.

In Fig. 2, two curves create, according to the accepted criterion, three areas: the relatively safe trading area located above the safety curve, the profitable, but potentially risky trading area located between the curves, and the losing trading area located below the zero return curve. Let's consider how we can use the areas detected in this manner to apply trading strategies in the most optimal way.

Trading System Safety Factor (TSSF)

To numerically assess to what degree a trading system is safe in the current market conditions, we will use a time window N trades long and find the values of Avg.Win/Avg.Loss and %Win inside this window. In this present article, this window is determined as 20 trades long, so, when speaking about the current results of the trading system, we will imply Avg.Win/Avg.Loss and %Win determined by the results of those last 20 trades. However, the size of your time window may differ from this randomly taken number.

It is better to choose the time window on the basis of the best/worst historical sequence of trades - say, if the Most consecutive wins = 9, Most consecutive losses = 4, %%Wins (in Tester) = 60%, then there is no reason to use a window that is smaller than 23 = 1.5*9/0.6 (or 30=2*9/0.6). (A remark by Sergey Fishchenko (aka FSV)).

Knowing the value of %Win for the last 20 trades, we can use the above formula to find the value of Avg.Win/Avg.Loss that would correspond with it on the safety curve. We will determine the ratio between the real value of Avg.Win/Avg.Loss for the last 20 trades and this "safe" Avg.Win/Avg.Loss value as trading system safety factor (TSSF).

To calculate the TSSF, we can use the following formula:

Avg.Win / Avg.Loss ((110% - %Win) / (%Win-10%) + 1)

It is clear that the higher the value of TSSF is, the more profitable the system was recently and, therefore, the more it corresponded to the market conditions. If the TSSF falls below 1, it means that it has gone to the increased trading risk area. If the values are even lower, then it means that the system has gone to the losing trading area. However, we are interested in only one thing: Is the trading system located in the relatively safe trading area or not? Depending on this, we can use several trading strategies. Let's consider some possible alternatives.

Theoretically, when finding the current results of a trading system in each area, we can determine the volume of a position to be opened as follows:

1. to open a position with the greatest possible leverage.
2. to open a position with the entire capital available without leverage.
3. to open a position in proportion to TSSF.
4. not to open a position at all.

There are other alternatives possible, but we will consider here only the above listed ones. Recall that we are only interested in whether we are in the relatively safe trading area or not. Then, having discarded some alternatives as unreasonable (for example, to open a position using the maximal credit being below the safety curve, or not to open a position being above), we will obtain the following possible combinations:

Two of the listed strategies don't use the safety curve and, therefore, cannot be considered as a control group. This is strategy 1 opening a position, in any case, for the entire capital available and with the maximum use of leverage, and strategy 5 opening a position for the entire capital available, in any case.

Let's consider how these strategies work using a real example. I use a system based on breaking through resistance lines on the H1 timeframe. This system is intended for long positions only and is traded with identical parameters on a portfolio of 8 most marketable Russian securities circulating on the MICEX (RAO UES of Russia, Sberbank, Rostelecom, NorNickel GMK, LUKoil, Surgutneftegaz, Yukos, and Tatneft). In Fig. 3 and 4, the values of %Win and Avg.Win/Avg.Loss are given for the period from the beginning of 1998 to mid-June 2002, respectively, detected in the time window 20 trades long.

Fig. 3. Moving value of %Win.


Fig. 4. Moving value of Avg.Win/Avg.Loss.

Using the above formula, let's calculate the TSSF - see Fig.5:

Fig. 5. The Values of TSSF.

It should be recalled that, if the value of TSSF is less than 1, the system leaves the relatively safe trading area. The figure obviously demonstrates the fact mentioned above: You cannot gain profits continuously on the market. The safety factor of this system was less than 1 during quite long periods of time, and those periods were unambiguously related to the drawdown periods of the account (Fig. 6):

Fig.6. Drawdowns of the Trading System.

Now let's try to check whether it is possible to achieve the asymmetric reduction of the trading system risk using the above strategies. As an example of how the analysis should be performed, let's compare strategies #5 and #7. The former strategy (#5) implies buying by the trading system's signal for the entire capital available (since there are 8 securities in the portfolio, buying for 1/8 of the capital). The latter trading strategy (#7) implies that we buy for the entire capital available if TSSF > 1, and don't buy at all at TSSF<=1.

The trading system statistics when using these strategies is given in the table below:

Thus, strategy 7 results in decreasing both the return and the risk, the desired asymmetry having been achieved: The risk decreased to a larger degree than the return, which follows from the increased return/maximal DD ratio.

Conservative traders can be satisfied with this essential risk reduction. For those being in more aggressive mood, there is a solution in form of increasing the leverage at trading using strategy 7. The results of trading with a leverage are given in the table below:

Trading strategy 7 with the leverage of 1.5 allows us to have the same returns as strategy 5, but with a lower risk. The leverage of 2 provides, with the risk being equal to the initial one, the returns 50% larger than in initial strategy 5.

Thus, we have considered only one alternative of the strategy using TSSF, which filters out trades with TSSF < =1. As a result, we obtained three versions of MM strategies:

1. reducing both the risk (to larger extend) and the returns.
2. reducing the risk at the same returns.
3. increasing returns at the same risk.

A trader can choose among these versions the one that meets his or her preferences most of all. A complete analysis of all strategies with their special features can take a long time. However, the information resulting from such analysis may be very important and useful. You can practically always find such a version of your MM strategy that will essentially improve the initial version. You should only clearly realize that a trader must not trade in the same way all the time, independently on the market conditions. Quite the opposite, taking into consideration the market phase and the relationships between the trading methods used supplies your trading with additional flexibility and allows you to get higher returns and to avoid undue risks.