I have two definitions of trend I wish to test to see if trend qualifiers increase accuracy and two different types of setups that I will use (same setups for each trend criteria). I also wish to incorporate 3 different money management strategies to see what would have performed better. all in all out (scaling in only with really strong trend and another signal), scaling out, and both scaling out and scaling back in with new signals (if it happens). The money management strategies stats, however, are not as important to me as the other stats. My time frame is a 0-a few days if stock doesnt act like I want and possible 2-4 weeks on average if it does. I am a swing trader in stocks and I trade with the trend and do better in a trending environment. I have no backtesting experience, I will be backtesting by hand. How many stocks do I need in my analysis, how many months back, how many total trades to give me at least somewhat an idea of how I can expect my system to perform. And anything else? I figure I can crunch about 100-200 setups in a weekend without gouging my eyes out. So in a month I could have a sample size of 1000 trades. Thanks guys.
When it comes to data more is better UNTIL you get to a point where the market dynamics have changed. As a rule of thumb, I personally think 500 points of data is what you need. This means if you are trading weekly thats 50 bars so 500 bars is 10 years. If you are trading using 15minute bars, then a week is already 120 bars, so a month is sufficient. In my opinion.
I don't trade off the weekly charts. I was just talking about average trade duration. Or am I misunderstanding your post? If market conditions are right I end up with 3-10 setups a week. This is 30 a month/360 a year. So I would only need about 2 years worth of data it appears. I am not selective with regards to what stocks I trade. I take all signals that provide the proper r/w and other criteria. Should I test multiple stocks?
2 years is plenty. What bars do you use to trade off? I think you are making it very hard for yourself and will unneccesarily waste alot of time if you try to do this without getting software. Any serious trader has their system coded into a software where it can be easily tested. Well firstly by trading by hand, it will take you ages to do things like: *Optimise parameters *Test different universes *Experiment with different money management strategies. Whereas all the above can be done in a matter of minutes with good software. And it will introduce biases into your trading because you will see the right hand side of the chart. Do you have enough money to trade all signals? Will you pyramid trades? How about re-investment of profits. Most importantly, dont forget to leave some data for out of sample testing (maybe the last few months) so you can validate your results. My blog has alot of good information on systems trading that may interest you. thetrendfollower.blogspot.com Best regards, Nizar.