how can you tell(estimate...) your EA will work in the future and for how long ??
That depend on the strategy wich you have code, what kind of strategie that is. if you backtest a strategie on EURUSD or simular Major Pair from 1999 till end of 2006 with a scalping strategie then you get most time a consistent gain equity shown. But because of spread and other small thinks like slippage in real the strategie does not work.
But if your strategie has a technical good way, then it is good possible that it work also in future.
And you must be carefull with backtest that are make with kontrollpoints, i have see that this backtet look also most time to good and with every tick much different. The best backtest is so with every tick.
That depend on the strategy wich you have code, what kind of strategie that is. if you backtest a strategie on EURUSD or simular Major Pair from 1999 till end of 2006 with a scalping strategie then you get most time a consistent gain equity shown. But because of spread and other small thinks like slippage in real the strategie does not work.
But if your strategie has a technical good way, then it is good possible that it work also in future.
And you must be carefull with backtest that are make with kontrollpoints, i have see that this backtet look also most time to good and with every tick much different. The best backtest is so with every tick.
I used 90% modelling quality for the test
and emulated a spread of 4 pips (which is quite high for EUR/USD pair)
I've written some EA which I backtested on 10 years of 15min EUR/USD data
If it shows consistent profit through all that period will it be normal to assume it will work for the next 10 years ?
As you may have guessed, it simply is not enough to have a strategy that generates profits under backtests. When it comes to using backtest results to speak to the likelihood (statistics) of future results it is more important that you properly compute the variance (standard deviations and all that) in the rate of return of your strategy over time in the backtest.
Knowing how to correctly calculate your expected "risk of ruin", doing that calculation, and understanding the results of the calculation are vital to your ability to understand the likelihood of your backtest results being indicative (or not) of future results. A good reference article on risk of ruin is this one: http://www.futuresmag.com/Issues/2009/August2009/Pages/Minimizing-your-risk-of-ruin.aspx
Do not be surprised if you come to find out that while your backtest results show you to be making money "in the long run", the statistics may well predict high likelihood of catastrophic drawdown and considerable losses are in store in the future.
Once you have your risk of ruin computed, then you can use it to answer your question with mathematical certainty. Risk of ruin can be used to answer the question "what is the likelihood that I will lose money in the long run over the next 10 yrs?" as well as computing the probability of losing X% or profiting Y%.
thanks for the info !
I had a feeling it can't be that easy
If you are not scalping but are using fixed pip targets, that will usually not last across the larger changes in market pattern
Backtesting is relatively 'kind' to an EA as spread is fixed and price action somewhat smoothed
Purely indicator driven EA's suffer worst from this effect and will be most misled in backtesting/optimization
FWIW
-BB-
My method is right alongside how I prepare my list of unexpected problems that I'll encounter.
If you have prepared a list of "unexpected" problems...then how do they continue to remain unexpected?
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
I've written some EA which I backtested on 10 years of 15min EUR/USD data
If it shows consistent profit through all that period will it be normal to assume it will work for the next 10 years ?