Quote:
Originally Posted by bossxero
Of course backtesting is not reliable, there are too many variable that are in effect during live market action such as broker spreads, news etc. Backtesting can just give an idea of performance. Of course, we should be using a 90% modelling quality to get better backtest results which are closer to real market action.
|
I dont mean to be rude but I'm so tired of hearing this mantra. "bactesting is not reliable" Bactesting is as reliable as you want to make it. I have been watching charts for years in live markets and I keep records, I also compare multiple brokers. The bottom line is that with minor variations (1 to 3 pips) all candles (especially on 1 hour and above) open and close at the same time.
I have forward tested this data (live traded) in my many years of trading, why then would I (or anyone else) assume that the data suddenly changes when I look back? Are you suggesting that the open and close of candles change after they are formed? When? after an hour or do all of last weeks candles suddenly change for no reason after a month has passed?
The notion is ludicrous. You just have to use some common sense.
If the system you are backtesting requires a 35 pip stop then ignore borderline trades. For example trades that come within one pip of the stop will be counted as a loss. This will still work because I can assure you, after trading on 3 or more brokers at one time, that a trade can be stopped out on one, but not the other two. Once again just use some common sense.
All systems should be tested on the open of the next candle after the signal anyway. And if the variance on that close is 1 or 2 pips then how can you not build a good picture of possible results. (possible results is the intention of backtesting anyway)
The real issue here is not backtesting being unreliable in its function but the fact that the market can change substantially if a new economic environment occurs (sub prime anyone?). A 35 pip stop 3 months ago may need to be a 45 pip stop today. For example I always increase my stops in August. Why? The market tends to overshoot with less liquidity. The smaller traders get more control while the big boys are all on vacation.
So if you are thinking of backtesting, use some common sense. Only take results on the close of the next candle, after the signal and do it by eye. Don't use software. Also I would not bother whith anything over 6 months as the market is not the same anymore, I gurantee you that.