Why Backtesting does not work: An Overview
LHF Sim
Thats why one must use a simulator; at least I find that a lot more useful than a demo account for testing manual set-ups.
mike
Whilst I agree backtesting doesn't "work", as a standalone tool, it doesn't mean it's not helpful.
I didn't try ea or backtesting
but I doubt that it work, after I see FOREX (up or down or lingering) is pretty much unlike Medic MCQ (true (up ) or false (down) or don't know)
because the volatility is like guessing a humping rabbits
thanks for the insight on backtesting -- I think the keyword is steady incremental (not the case , in forex)
The premise in the opening post of this thread seems to be that because historical data is used in testing that it cannot be useful, however if the advisor is operating purely on historical information, then the premise is correct.
However, I do not see how this can be, since it is pointless to create something that attempts to work off historical data only. If the strategy is reacting to price movement rather than historical data then the data is only supplying the market movement necessary to test if the strategy is viable. In this respect testing can be very useful, for one thing is certain, if a strategy is not profitable during testing, it is highly unlikely that it will be profitable in a live situation.
Surely that is the bottom line, does it work! I am aware that many developers like to sit back and look at nice sets of numbers and pretty straight line upward graphs, when great backtesting results become the goal to be achieved, then backtesting is not useful, but in the end, it not exclusively about successful backtesting results, but rather it is all about does it work and is it profitable in real life.
Anything that assists in that goal is useful, including testing. However, testing results should not become the focus of the goal to be achieved, success in real life should be and the main overriding parameter should be, does it work in real life?.
A distinction needs to be made between curve-fitting and natural backtesting.
Not sure
I'm not sure I agree with the back test theory.
There are so many factors involved that it may not be possible to get an accurate back test but if you use the statistics as a filter then it should be of help.
Filters are the one tool to assess whether a system is safe to use.
Whilst I used my own 'purpose built' system to trade before my coder took it over for sale, I can honestly say that systematic back testing has paid a lot of dividends.
I consider my system was by far the best but there will be other systems that will give good results.
If you want to see the results of mega back testing go to his site
powertraderpro and see for yourself.
Regards
TEAMTRADER
Trade what you see and not what you hear or hope
when do we use CURVE-FITTING ??
for what type of indicators?
Backtest is very helpfool
1. If strategy doesn't perform on backtest it most probably won't perform in real trading;
2. Past doesn't repeat itself as much as somebody would like it. Because of that backtest is even better as Demo trading. You can't loose or win, but you can at least check pount 1. Weel, back test is even less emotional as Demo, but you can backtest years in seconds.
3. Backtest is very helpfool exploring strategies, how do they perform in different conditions, what is worst and best results etc.
4. Can you use bactest as proof for future wins? No. Backtest can just give you idea of probabilities. Thats it.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use
Why Backtesting does not work: An Overview
Article written by http://www.forextraders.com
Backtesting is the application of a trading strategy to historical data with the purpose of establishing its profitability. The assumption behind this approach is that if a method works in the past, it will also work in the future. Indeed, one of the fundamental tenets of technical analysis claims that “history repeats itself”, or, in other words, that price trends and patterns are repeated constantly in spite of the apparent randomness observed in the price action. If something worked in the past, it’s likely to work in the future as well.
But while this assumption may sound fair enough on first consideration, it is in fact totally out of basis in trading. It has been shown, time and again, that, as a chaotic process, (further examples of which include the Brownian motion of molecules, patterns of earthquakes, and weather changes), market action is in fact very resistant to any kind of mathematical approach that depends on past data for the establishment of future developments. Analysis of the historic record of prices is almost completely irrelevant to future market action.
Can we then expect to apply backtesting methods to our strategies and expect to be profitable?
What is the Purpose of Technical Analysis?
Technical analysis serves the purpose of determining entry or exit points of trades. Technical strategies cannot predict market events in advance since the chaotic nature of the markets means that any technical theory valid today may lose its significance in the next instant.
Technical analysis is useful mostly as a way of structuring and ordering our trading decisions, and preventing arbitrariness in decisions from causing a rapid evaporation of the account. The relationship between gambling and trading is neither controversial nor a secret. A large number of successful traders, past and present, were also successful gamblers, attesting to the strong relationship between trading and the management of risk and probabilities. Technical tools help this goal by allowing traders to analyze the risk and profit potential of their trades, and to avoid excessive risk taking. It should be clear however, that the highly subjective nature of the predictions made with technical tools easily precludes its use as a tool for knowing, in advance, where the price will go. It is one of the most common occurrences in trading that you get two entirely divergent viewpoints from two experienced analysts examining the same chart pattern. In other words, technical analysis should help us know in advance what we will do in response to a particular price development, but it never allows us to predict the response of the market to any development.