Create and Test Forex Strategies
This is an old revision of the document!
The acceptance criteria are widely used in EA Studio (in the Generator, Reactor, Validator, Optimizer). The acceptance criteria serve as a filter when a new strategy is generated or enters EA Studio. If the strategy passes it will be pushed to the respective collection. For example if acceptance criteria are enabled in the Optimizer it will not display strategies that do not pass the acceptance criteria.
The acceptance criteria are applied to the backtest result stats. There are three zones of the backtest the acceptance criteria applies to (see the picture below)
If we have Out of Sample enabled all three zones will be used.
If Out of Sample is disabled only the first (Complete backtest) part will be used.
Acceptance criteria work based on the strategy backtest. The strategy backtest results can be seen in Strategy > OOS Monitor tab or Strategy > Report tab. OOS Monitor is better since you can see the distinct zones of in sample and out of sample testing plus the complete backtest of the strategy.
If you find a strategy with the Generator that you like you can set the acceptance criteria so the generator will find similar strategies. Just go to the Strategy > OOS Monitor and see the results of the backtest. Then enter them in acceptance criteria settings. For example if you liked a strategy that has a drawdawn of 8% in the acceptance criteria settings enter a number that is a little biter more loose - 10% for Maximum drawdawn etc. From now on the generator will only give you strategies that fit these criteria.
Add validation criteria allows you to add another acceptance criterion to the list. A strategy will pass the acceptance criteria only if it passes all tests in the list.
To remove a test just click the X button to the right of it.
Reset - resets the acceptance criteria for this zone of the backtest to it's default values.
EA Studio backtesting engine is very fast because it operates with bar data. Very rarely however this leads to ambiguities in trading decisions. For example if there are protections like Stop Loss or Take Profit it might be important if the price went to bar's high or low price first. For such bars the backtest engine cannot what happened first so it marks the bar as “ambiguous”. This happens rarely but this criterion allows you to require a high percentage of the bars to be unambiguous to consider the strategy good enough.
The Reference line is an ideal line, which represents how the strategy's balance should move up in ideal conditions. This is of course impossible. The balance line deviates from the Reference line. The maximum balance deviation represents how far away the real balance is “allowed” to move away from the Reference line. For example, if the maximum balance deviation has a value of 20%, each of the points of the balance line, should not deviate more than 20% (up or down) from the value of the reference line for that bar. If the strategy's balance line moves further than 20% of the reference line, the strategy does not fulfill this criteria.
How many consecutive losing trades can the strategy have.
Set the maximal number of trades to make sure the strategy is not over-optimized. Also trade less often to avoid losing money on the spread.
The account equity should not go below the given percent value.
Set the maximum amount of consecutive days in which the strategy is not making profit.
Set the minimum average Holding Period Return in percent.
Defaults to 100. At least that much trades should happen in the backtest. This is important for the Generator and Optimizer in order for the backtest result to be trustworthy. If there are too few trades, the parameters below might have good-looking values, but in the same time harm your profits. For example if you have a win/loss ratio of 1.0, it is great, but if the trade count is only one, this strategy not one you can rely on.
The strategy should make at least the set amount in your account's currency. Useful for the whole backtest as well as the Out of Sample part.
Set the minimum profit factor i.e. gross profits divided by gross losses.
The strategy should make at least this amount per day in your account's currencly.
The strategy should have at least this this profit/drawdawn ratio.
The strategy should have this or a higher Sharpe ratio.
The strategy should have this or a higher System Quality Number.
The win/loss ratio is:
( (number of profitable trades) / (number of profitable trades + number of losing trades) )
Its value might vary between zero and one. We advise that, if you are using this requirement, keep the “Minimum number of trades” and “Minimum profit per day” requirements enabled too.
Shows how stable the profits are over time. Imagine we have a strategy with minimum months of profit equal to 60%. This means that if the strategy backtest was for 10 months of trading we can see that the strategy was profitable for 6 months and in the other 4 months it didn't make any profits.