Let's say you're a swing trader with a max holding period of about a month. Trading stocks only,not options,stocks themselves only. How many trades is it going to take to pronounce a strategy genuinely effective? Mostly just semi-formal backtesting. I know the more the better but there comes a point when it kinda just becomes overkill and wasting time. 300? 700? 1000?
Adequate sample size?
- Thread starter tomkat22
- Start date
well more is better - but here is the thing - it's hard to build a model that will encompass all the different traits of the market.
also when optimizing to have a robust system you need to be able to trade with less than optimal parameters so you should have several settings that make money otherwise it's over fitted and will not hold up.
any more confusion i can add? lol
also when optimizing to have a robust system you need to be able to trade with less than optimal parameters so you should have several settings that make money otherwise it's over fitted and will not hold up.
any more confusion i can add? lol
Years ago I traded as CTA with computerized systems. I had run hundreds of thousands of tests, & traded real-time about 30 different systems on futures (excellent results), & then indiv stocks.
The numbers are not the important factors. Methodology is the issue; some questions:
1. Are your trading methods based on proper principles?
2. When you review the individual trades do they all make sense, regardless of P/L?
3. Do your overall statistics make sense? For instance, if you win on more than 70-75% of your trades, most likely it is the result of pure data fitting. Similarly, if your total profit, on say your 1,000 trades, are due to just a few indiv trades, then also, probably due to data fitting.
4. Do your systems use approx same variables, in all stocks, futures, etc?
5. Do your systems go short as well as long? With same variables?
6. Do neighboring variables also yield about same results? For instance, although I never used any moving averages, if your system works with a 27 day moving average, it should also work with say 21 - 33 moving averages.
7. Do you always have some stop-loss point for every position?
I think above are key points. But as implied below, it has been a very unusual period in the markets the past decade, mostly due to Central Banks printing money, & maintaining ZIRP. The old adage that 20 years experience (trading or system test results) is actually just 1-year experience, repeated 20 times.
I've heard it said that 10,000 hours of screen time is required to become profitable as a trader and was curious if there was a "magic number" of trades for someone swing-trading stocks. I would think after 4 or 5 hundred trades the market has thrown just about everything at a swing trader it's going to.
Perfect practice makes perfect.
10k hours spamming the wrong stuff is just running on a wheel, doesn't mean you can read a map afterwards.
Google sample size t-tests and type 2 error