Understanding SQ’s strategy development process

Mr. Fric kindly posted a process how to develop a strategy that fullfills a lot of robustness requiredments. It can be found here. I’ve used the example file Mr. Fric posted. The VPS ran fort wo days to complete the process. I’ll now go on in the process:

• First Filter – Out of sample (OOS) check

All strategies with a OOS profit smaller than $500 are deleted. 175 deleted – 1825 left.

• Second filter - Retesting and second OOS check

The strategies are tested with higher precision on additional datasets and time periods. All strategies with a OOS profit smaller than $500 are deleted. 430 are left.

• Third filter - Gbpusd check

The strategy should be profitable on Gbpusd too. 10 strategies are left.

• Fourth filter - Robustness tests

Now we use a fixed amount moneymanagement method and do at least 20 monte carlo simulations for each strategy. We skip every strategy that shows a much higher drawdown in the 95% confidence level. I've selected five strategies and took a look on the trading rules. Mr. Fric gives the advice that they should make some sense at this stage.

• Fith filter - Walk-Forward Matrix test

We wanna see if the strategy is adaptable to a wide range of market conditions. In the WF Matrix we are looking for nearby cells that have passed the test. Unfortunately I chose a strategy example that did not pass this test.

I will now do this process once again for the daily timeframe.