Greetings, I've written a small book for the enjoyment of system traders and developers if they are interested. It is in the Free Book section of TradingSystemDesign.com. Good trading.
Trading Book
- Thread starter EliteThink
- Start date
You are welcome. Actually, I simply wrote it in Word and had it dumped into pdf for download. I've seen your work and you are a pretty advanced developer so at best it would be a quick read. I hope you like it though.
EliteThink, great little book! We need more contributors like you and for example, acrary!
A few comments on the book:
About walk-forward testing, I completely agree; I've thought about it before. We have two periods A and B; when added together they make a bigger C period. What happens in back-testing: we test until we find something that looks good on C. What happens in walk-forward: we test until we find something that looks good on A *and* B. Isn't that the same thing?
In either case it seems we're simply throwing ideas at the data until the equity curve looks nice. In either case there's survivorship bias in the system, meaning only systems that look good where we want them to look good are considered. So to say that somehow walk-forward testing reduces curve-fitting is kind of strange. I don't see it; maybe it's true in Neural Networks but in regular systems I don't see it.
FFP is Ryan Jones' method right? I don't deal with MM enough to memorize the abbreviations yet
Still on the please-o-please-let-me-find-positive-expectancy part
Do you think it's possible to create an intra-day system in the SP that has a very high of profitable weeks (like 80%++)? I know there's a lot of discretionary traders who have maybe 1-2 losing days in a week. I'd like to be more like them in terms of equity curve smoothness
A few comments on the book:
About walk-forward testing, I completely agree; I've thought about it before. We have two periods A and B; when added together they make a bigger C period. What happens in back-testing: we test until we find something that looks good on C. What happens in walk-forward: we test until we find something that looks good on A *and* B. Isn't that the same thing?
In either case it seems we're simply throwing ideas at the data until the equity curve looks nice. In either case there's survivorship bias in the system, meaning only systems that look good where we want them to look good are considered. So to say that somehow walk-forward testing reduces curve-fitting is kind of strange. I don't see it; maybe it's true in Neural Networks but in regular systems I don't see it.FFP is Ryan Jones' method right? I don't deal with MM enough to memorize the abbreviations yet
Still on the please-o-please-let-me-find-positive-expectancy partDo you think it's possible to create an intra-day system in the SP that has a very high of profitable weeks (like 80%++)? I know there's a lot of discretionary traders who have maybe 1-2 losing days in a week. I'd like to be more like them in terms of equity curve smoothness
Glad you liked it. FFP is fixed fraction position sizing as you probably surmised. It works for me but others have different techniques.Quote from traderkay:
EliteThink, great little book! We need more contributors like you and for example, acrary!
A few comments on the book:
About walk-forward testing, I completely agree; I've thought about it before. We have two periods A and B; when added together they make a bigger C period. What happens in back-testing: we test until we find something that looks good on C. What happens in walk-forward: we test until we find something that looks good on A *and* B. Isn't that the same thing?
FFP is Ryan Jones' method right? I don't deal with MM enough to memorize the abbreviations yet
Do you think it's possible to create an intra-day system in the SP that has a very high of profitable weeks (like 80%++)? I know there's a lot of discretionary traders who have maybe 1-2 losing days in a week. I'd like to be more like them in terms of equity curve smoothness
As far as an intra-day system that wins 80% of all weeks, possible but tough to do. I feel it may be better to focus on win% of trades and the profit factor among others things. If you can get a high enough win % and pf then the weeks should fall into place. Also, a low average between peaks a good way for to target a smooth equity curve and high sharpe ratio. Best to you.
I've been told fixed ratio position sizing is better than fixed fractional. I'd have to say fixed ratio is only prudent if the account is for an individual and not a fund of some kind, but it is up to the trader.
Glad you are enjoying it as it was fun to write. Are any of you trading your systems now?