Menu
Brokers
MT4 Forex Brokers
MT5 Forex brokers
PayPal Brokers
Skrill Brokers
Oil Trading Brokers
Gold Trading Brokers
Web Browser Platform
Brokers with CFD Trading
ECN Brokers
Bitcoin FX Brokers
PAMM Forex Brokers
With Cent Accounts
With High Leverage
Cryptocurrency Brokers
Forums
All threads
New threads
New posts
Trending
Search forums
What's new
New threads
New posts
Latest activity
Log in
Register
Search
Search titles only
By:
Search titles only
By:
Menu
Install the app
Install
Reply to thread
Forums
Forex Discussions
General Forex Discussion
Why backtests are useless, EAs are flawed and their parameters are bad [DISCUSS]
JavaScript is disabled. For a better experience, please enable JavaScript in your browser before proceeding.
You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an
alternative browser
.
Message
[QUOTE="Darwin, post: 55170, member: 28524"] First of all, curve fitting is not just a problem of fitting too much. You could take a random trading strategy, evaluate it on, lets say 2010-2012 and then most of the profitable ones would be "overfitted", meaning they would not make profit on 2013. (I once tested this using a few hundred algorithmic generated strategies) Yes, and we WANT to fit the input parameters to the latest available curve, as it is the best bet that the characteristics of this curve will go on in the future, and its also the only dataset we can work with. (best bet != certain) And finally thats EXACTLY what a WalkForwardAnalysis is there for, to measure if and to what extent a given strategy, fitted on the latest curve, is representative of the future market. And it does so by making ~100-150 of these "adapt on the latest curve, test on future market" tests. Hope that makes it clear :) Tough, WFA is no holy grail, and there is no guarantee that the traded inefficiency will hold in the future, but thats a different problem, eliminating not-working strategies. But to test a strategy's ability to model meaningfull inefficiencies in the future markets based on the last available data, it is at least a lot better than a backtest. -Darwin [/QUOTE]
Insert quotes…
Verification
Post reply
Top
Bottom
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.
Accept
Learn more…