I’m doing some testing on Woohoo that I think other system developers will find interesting. There are a few things I’ve been wanting to test out, and they are:

Day of Week Profitability (ex. for a trade I might find that Monday trades lose money in the long run, while Wednesday trades produce the most profit.) (Why: Woohoo only holds 1-day trades.)

Close Trades at 7:30am cst (Just before NY opens.)

Remove Holidays - don’t trade holidays or the business-day after. (Low volatility caused by lack of traders, not bull/bear struggle.)

For the Day of Week, I’m removing trades where the day in question is generally unprofitable. If a particular day is very profitable, I double the position size for that day. I’m running this on a trade-by-trade basis, and so far each trade is proving 2-3x more profitable.

Any thoughts? Do you have similar ways that you optimize? What are some filters you like to use?

Another example of a filter I like is to only buy when the bid is above a certain moving average, and vice versa for selling. When developing, I throw this in and see what happens, and test a few different moving averages.

(Note: Woohoo isn’t currently changing, I’m just testing and looking.)

What software do you use for backtesting?

Currently Metatrader4.

What language is the system written in?

Your theory about profitable and unprofitable days has merit based on my research with futures. One has to be diligent not to curve fit day based strategies and also be prepared to have unlucky days when all the instruments will be unprofitable, creating a drawdown, that cosmetically, isn’t pretty. Conversely, winning streaks also occur that look fine and make you feel as if you’re smart.

What role does luck play into the equation? More than most developers would care to admit but that isn’t important if the statistical probabilities are in your favor.

It’s important for both punters and developer that this is a long term proposition, unlike going to a casino and hoping for a lucky streak while you’re on vacation.