Trading system design and optimization

Usually, the formation of a trading system needs to go through three stages. The first is the generation of a trading idea, which can be a complex theory, a simple technique, or any other idea that you think can be profitable. The second is to transform the concept into a corresponding strategy. Specifically, it is necessary to clarify the core operation arrangements such as entry, stop loss, and take profit that are compatible with the strategy. Finally, a complete system is formed, that is, on the basis of the above, the various elements are solidified into clear trading rules to form a reproducible trading system. The so-called reproducible means that anyone who operates according to the signals of this system has no difference in performance .

The most critical sign of a successful trading system is that its performance generally shows large profits and small losses for a long period of time, at least it must be able to pass the test of historical market conditions. However, the author believes that this is far from enough, and further important analysis of the four dimensions of the trading system is needed to determine the profitability of the trading system and determine the direction of future improvement.

The first dimension is the winning rate of the trading system, that is, the respective proportions of the number of profitable transactions and the number of losing transactions in the total number of transactions.

The second dimension is the profit-loss ratio of the system, which is the ratio of the average amount of money won each time to the average amount of money lost each time.

The third dimension is the friction rate of the system. Everyone knows that futures is a zero-sum transaction. Every time a transaction is completed, various fees including handling fees must be paid. No matter whether the transaction is profit or loss, it cannot be avoided. This fee rate That is the friction rate.

The fourth dimension is frequency, which refers to the average number of times the system sends out trading signals within a period of time, such as a year.

Now we can analyze the correlation of the above four dimensions with the profitability of the trading system.

First of all, is a high winning rate guaranteed to be profitable? Not necessarily, in extreme cases, such as 99 consecutive 100% profits, as long as there is one 100% loss, even if there is a 99% winning rate, the final result will still be zero.

Secondly, is a high profit-loss ratio guaranteed to make a profit? Not necessarily, for example, an average of 3 wins and 2 losses, but if an average of 1 win is accompanied by 2 losses, then it will still be a loss after offsetting, so the winning rate must be combined with the profit-loss ratio to form the most common mathematical expression of the trading system , that is: total profit = profit times × average profit amount - loss times × average loss amount.

The third is the friction rate, that is, the amount of money consumed by each transaction. This is an absolute consumption, so of course the less the better.

The last is the frequency, that is, how often is a transaction performed on average. If it takes one to two years for a profitable trading system to capture a trading opportunity, this is by no means an excellent trading system.

Next, let us combine the four-dimensional evaluation method to analyze the two basic types of trading systems.

The author believes that there are two basic types of trading systems, one is the oscillating trading system based on the oscillating market, and the other is the trend trading system based on the trending market.

The oscillating trading system strives to maximize profits in the oscillating market, while trying to avoid possible losses when the trend comes, while trend trading is just the opposite.

From the comparison of the above four dimensions, we will find that compared with the trend-based trading system, the oscillating trading system usually requires a higher winning rate, generally greater than 50%; a relatively loose profit-loss ratio, which can be below 1; due to the trading frequency Higher, requires a very low friction rate, that is, this type of system needs to accumulate profits through a very large number of transactions, extreme such as intraday speculation, it is necessary to minimize friction loss; correspondingly, the oscillation type The capital curve of the trading system is relatively smooth, with less retracement. Trend trading usually only requires a low winning rate, which can be less than 50% or even lower, but must have a higher profit-loss ratio, such as advance 3 retreat 2, or even advance 2 retreat 1 or higher, and a relatively loose friction rate . Because the transaction frequency is relatively low, higher transaction costs can also be tolerated, and its capital curve may have a certain percentage of large retracement. If someone wants to ask whether it is possible to design a trading system that combines the two, I think it can be realized, but the composition is more complicated.

We know that parameter optimization and capital management are two important ways to improve the performance of trading systems. Next, we will discuss the focus of performance improvement by combining the characteristics of the two basic types of trading systems.

The first is the relationship between system profitability and parameter optimization. Even if a system can be profitable through the historical market test under certain parameter conditions, it may not necessarily be profitable in reality. The problem is most likely to lie in two links. One is that it cannot be realized a priori, which means that the idealized degree of the test transaction cannot be fully realized in reality. For example, the liquidation in the test may be a real stop, or it may be specifically related to the market selected during the test period. The second is that the posterior cannot be realized, which means that the transaction in the test requires very precise operating discipline, but in the actual implementation, due to the operator's reasons, the system cannot be fully mechanized, resulting in a profitable system turning into a loss. The first case can be optimized mainly by adjusting parameters, while the second case needs to adjust the system so that people and the system can achieve harmony and unity. For the same strategy, different parameters may determine the profit and loss of the same system, and parameter optimization plays a very important role.

The second is the relationship between system profitability and fund management, where fund management mainly refers to position control. Because the oscillating trading method mainly accumulates profits through multiple small profits, it requires a relatively large position at the beginning in most cases, and requires very strict stop loss control when the trend may come, so for oscillating trading systems Focus on parameter optimization. Trend trading can tolerate more trial and error, the purpose is to let the profits run, so relatively speaking, fund management or position management has more room to play in the trend trading system.

The author believes that for the trend trading system, on the one hand, it is necessary to improve the winning rate of the transaction through parameter optimization, and then improve the profit rate of the system, especially each product may have different optimal parameters, and sometimes the market within a period of time will also have special On the other hand, it is also necessary to improve the profit margin of the system by increasing the profit-loss ratio. Sometimes the role of fund management, that is, position management, is even more important than parameter optimization. The key lies in whether the operator can distinguish between Under which conditions in the trend trading system are more likely to have a strong trend, and which transactions will be stopped by the system. Only by deeply understanding one's own trading system and making corresponding improvements can the performance of the trading system be continuously improved.