How many transactions should there be to check the stability of the system
Statistical data does not reflect the full picture of the effectiveness of the Forex trading system without information on the number of transactions. The relevant sample should contain such a number of transactions, which will be sufficient to determine the stability of the strategy, its quick adaptation to various kinds of changes, etc. Moreover, not only the number of transactions determines the effectiveness of the system, but also the time period. For example, an intraday strategy tested on a yearly interval with 120 trades will more accurately reflect the actual results than a scalping tactic with 300 trades on a 1 month interval.
How to evaluate the relevance of a sample for the Forex trading system
The answer: “The more the better” – is not accurate and correct from the point of view of mathematical statistics. For long-term strategies on weekly charts, testing on a 5-6 year history with entering the market 30-50 times is sufficient; for scalping, pipsing or averaging, you can take a shorter time period, but with a large number of transactions. The sampling accuracy can be improved by multicurrency testing, but:
- the same strategy can give different results on instruments with different liquidity;
- most platforms do not have multi-currency testing functionality.
This figure shows an example of the dependence of the standard statistical error on the volume of open positions. The chart shows that after the 100-200 trades segment, saturation occurs. If before this section, to reduce the error by 5%, it was enough to increase the sample by 20-50 transactions, then after this section, to reduce the error by another 5%, the sample should be increased by 200-300 transactions.
Another example of the need to expand the sample is well illustrated by the following figure:
It shows that with an equal number of open positions in the first case, both systems seem to be stable, but with an increase in the number of transactions, the second Forex trading system has remained practically unchanged, and the first shows instability.
Output… The optimal testing period for the “time” parameter is from 1 year (3-5 years), for the “number of transactions” – over 150-200 transactions. The most effective way to test a Forex trading system is multicurrency testing. If the equity does not change much with a slight adjustment of the indicator settings, the strategy can be called stable. Equity assessment and optimization of indicator settings are carried out only in conjunction with the analysis of parameters for evaluating trading systems. If you still have questions, ask our analysts in the comments after the article!