How do you know when successfully backtested strategy will succeed in real trading?

[comagnum]

Having a entry/exit signal that gives you a level of confidence is good.
Having a good risk mgmt strategy is critical to stay in the game. Most veterans consider this far more important than the entry/exit signals.

Your success will ultimately depend on how disciplined you are & how sound your risk mgmt rules are. You will have to overcome your emotions which is the hardest part of trading. You will also need extraordinary discretion; when & how to tweak it, drawdowns -when to abort it, when to add new systems, etc.

Richard Dennis - the famous trader that started the Turtles had said something like 'I could print my trading system in the New York Times, few if any would ever be able to follow it' . The same applies to automated traders that create their own program trading - creating it is the easy part - following it is the hard part.

 

[JSOP]

The only things that I can think of that backtests cannot test for is execution quality of your trades. In backtests the execution quality of the trades is always 100% instantaneous; there is never any lags, delays, liquidity problems and etc. whereas in real trading, you are going to face all of these problems and execution quality is never 100% as reflected in backtests. If the success of your trading system is heavily relied upon execution quality then your backtest results is going to differ from real trading unless the execution quality of your platform and your broker is extremely close or exactly as how it's in backtests.

 

[Mysteron]

The only things that I can think of that backtests cannot test for is execution quality of your trades. In backtests the execution quality of the trades is always 100% instantaneous; there is never any lags, delays, liquidity problems and etc. whereas in real trading, you are going to face all of these problems and execution quality is never 100% as reflected in backtests. If the success of your trading system is heavily relied upon execution quality then your backtest results is going to differ from real trading unless the execution quality of your platform and your broker is extremely close or exactly as how it's in backtests.

All the more reason not to use broker tools and instead develop your own model which includes execution innefficiences. Backtesting by itself though is pointless, its essential that there is the means to run a trading method forward on new data.

 

[MattZ]

Still a beginner here. Back testing must have some validity.

I've seen in a few places that strategies that work great in backtesting then fail in real trading. I don't see why, the investing universe can't have gone through a wormhole just when you went live. But anyway, how do you know if your successfully backtested strategy is going to work in real time?

All strategies that went from paper to real time trading needed an adjustment. Some strategies do not take cost into consideration, some may assume you get filled on all limit order of entry and profit, etc. Only live trading will point out the wrong assumptions you had in your theories. Some could be fixed and some may not due to certain limitations.

 

[VincentvanG]

Does history repeat itself? What is the likelihood that the historical data that you back tested will repeat in the future? If it does then go live.

 

[murray t turtle]

Having a entry/exit signal that gives you a level of confidence is good.
Having a good risk mgmt strategy is critical to stay in the game. Most veterans consider this far more important than the entry/exit signals.

Your success will ultimately depend on how disciplined you are & how sound your risk mgmt rules are. You will have to overcome your emotions which is the hardest part of trading. You will also need extraordinary discretion; when & how to tweak it, drawdowns -when to abort it, when to add new systems, etc.

Richard Dennis - the famous trader that started the Turtles had said something like 'I could print my trading system in the New York Times, few if any would ever be able to follow it' . The same applies to automated traders that create their own program trading - creating it is the easy part - following it is the hard part.

%%
Good points.
I found out later+ its a matter of public knowledge; Rich Dennis used Wall Street Journal [WSJ] for furniture/ seats @ his home. Never met him but i enjoy what he wrote/ Jack Schwager . I prefer IBD but it takes all kind$ to make a market.........................................................

 

[Stratos Capital]

I would like to take a statistical perspective.
1) The more trades your back-test executed, the more realiable is your backtested results.
2) The larger your edge, the more likely its really going to work.
3) The longer the time frame back-tested, generally, the more reliable the results are.

 

[2weels]

I'd like to thank you all for responses in this thread. It is extremely informative for me. Thanks

 

[sle]

My personal "parameters" in order of decreasing importance
- the trade/strategy has structural reason to exist
- the trade/strategy has theoretical edge
- the strategy has few free variables that can result in a curvefit
- the strategy back test spans a period that includes different regimes

 

[wrbtrader]

Still a beginner here. Back testing must have some validity.

I've seen in a few places that strategies that work great in backtesting then fail in real trading. I don't see why, the investing universe can't have gone through a wormhole just when you went live. But anyway, how do you know if your successfully backtested strategy is going to work in real time?

You'll be surprise by how many traders used the work "back tested" when in reality it was not properly backtested. Also, real money trading involves some elements that can not be duplicated in a back test.

This forum ET is litter with trade journals that have poor real money trading results in comparison to very profitable back test results. Some where in fact green (profitable) in back test but red (losing) in real money trading.

How do you know its going to work in real time ?

There's no guarantee and you must start with small account or small trades to see how the method performs with real money. Also, best to include simulator trading between the back test and real money trading just to become familiar with the trade execution platform.

In contrast, if your trade method is 100% automated...the results of your back test will be more realistic about what you expected in comparison to a trade method that's not automated.