Community

The idea behind an optimization is simple. First you must have a trading system, this may be a simple moving average crossover for example. In almost every system there are some parameters (indicator periods, comparative constants, etc.) that decide how given system behave. The optimization means to test the system with different parameter values to […]

This is another type of Monte Carlo simulations, in this case it simulates random changes in properties that require the strategy to be retested – such as changes in spread, slippage, strategy parameters, or history data. Because every simulation requires a complete backtest this cross check could take long time. It the backtest on main […]

This test for robustness is quite though – it means testing the same strategy on different markets – it means different bol(s) and/or another timeframe(s). Robust strategy should ideally work on multiple symbols/timeframes. In reality, because each market has its own characteristics, daily volatility, etc., it will be not easy to find a strategy that […]

This cross check run simulations where in each simulation it manipulates the existing trades – shuffles them, misses some and so on. It is very quick, because it doesn’t require running backtests, it works on already existing trades from main backtest. The idea behind this is to verify how much the strategy equity curve depends […]

This test is simple – it backtests the strategy again on the same data, but with higher precision. It is usually best to make the main test on the fastest Selected timeframe precision, because it can very quickly filter out bad strategies – these that produce no trades or whose Net profit is negative. Once […]

StrategyQuant X allows you to use cross checks (robustness tests) during the strategy build, or when retesting the strategies. There is a number of cross checks that can be used, ranging from simple ones to very complex ones, and they are simple to choose – you can just move the slider in the simple settings: […]

Cross checks – robustness tests and analysis Overfitting or curve-fitting strategy to the historical data on which it was build is the biggest danger of strategies generated using any machine learning process. During or after developing new strategy you should make sure your strategy is robust – which should increase the probability that it will […]

when you set Build mode to Genetic evolution in What to build -> Build Mode, a new tab Genetic options will appear in the settings. There you can set up various properties of genetic evolution. Genetic options Max # of Generations number of generations for which the population will be evolved. Recommended value from 5 […]

When the strategies are generated, every new strategy is backtested on a history data and the results of the backtest are then used to compute the Fitness (rank) of the strategy. Fitness is number from 0 to 1 and it should reflect the “quality” of the strategy according to the given criteria. In this screen […]

Cross check tests are additional tests that can be applied to your strategy. The idea behind robustness testing is to verify how well the strategy will behave when there are small changes in inputs, history data or other components of the strategy. This section is described in more detail in the Advanced functionality and Cross […]