Accurate Backtesting
Is it possible to write an external Strategy Tester using tick data to interface with MetaTrader?
- Yes, using mql4 (combined language), script, indicator, expert, dll libraries.Some developers don't bother using the Strategy Tester or DEMO
accounts. Does this imply the only accurate testing environment is LIVE
forward testing?
- If right is written code and with same supplied (tick) data Strategy Tester does not makes mistakes, same to similar results will be shown on LIVE parallel run.- Live BUT Demo is only accurate without risk, if for the same name testing, but takes time
So your saying the Strategy Tester is not the issue... the issue resides in the Historical Data?
Not totally convinced in the Strategy Tester with ambiguous bars, algorithm fitting and over optimization.
Better (winner ATC2007) wrote is own Strategy Tester (combined with reliable historical data) which were a pivotal part of his success. I don't have the capability to touch Neural Networks but would prefer if backtesting was as accurate as possible.
I see time and time again wonderful EA's on backtest fall on their faces after going LIVE.
Pivot combined with other calculations is very precise, as we saw (Mr./Mrs. Better there had entries with 0 points missed - 100% accurate), so that is that because more than 50% of traders are old school (to say) and are trading by pivot points.
If in the Strategy Tester is used "every tick data" on any time chart up (that is M1, not actual ticks), it could be nearly close to Real testing - Forward Demo.
Falling after going LIVE, by my opinion, can be from other more reasons, maybe sometimes not same data and over optimized with different history data AND without strategy for trading build or even sometimes without trading experience - what is very important when choosing strategy, strong strategy and program can maintain the discipline sometimes missing with traders, pivot is strong calculation - but as singe approach to trading is not enough often.
On previous thoughts for external Strategy Tester, I have made many combinations for testing, here is one example of Optimizer/Indicator for existent indicator in the Code Base but, this is again not enough accurate because it is not using M1 or tick data if used on higher time charts, and again as alone stand strategy is not enough - we know history repeats itself but in trading it is not the same as people creates and makes the history so also must be made strong (not long heavy - but strong as strategy possible by experience) calculations. In difficult periods what we call Period Where No/Or Yes volatility when is changing often Take Profit and pause fixes the situation, but that is again as currency strategy approach good from my side.
By my decisions if I share it, when I loose in guessing, I often change the period on bigger time period, and I always relly more on bigger Time Frame Chart - Directions than the lower/smaller, often small entry and waiting for swing for main entry. And also early catch trade talking as type of successful made trade is better than latter one, which could probably start at the end of trend, now we are talking about currency trading what is not same in case Stocks and rest (there apply other rules - stocks: overvalued and later leveling-back as often moves and happens).
If someone wants to put all that in auto calculation and trading, without mistake or at least 99% as planned, there is lot of codding work and checking if it is right as strategy that if it is right as working code program as wanted - it doesn't matter if the program will even not go Live. What ever chosen if someone wants to achieve the target and something what is planned have to work a lot and hard to succeed in that, if he/she wants job to be done.
If in the Strategy Tester is used "every tick data" on any time chart up (that is M1, not actual ticks), it could be nearly close to Real testing - Forward Demo.
Falling after going LIVE, by my opinion, can be from other more reasons, maybe sometimes not same data and over optimized with different history data AND without strategy for trading build or even sometimes without trading experience - what is very important when choosing strategy, strong strategy and program can maintain the discipline sometimes missing with traders, pivot is strong calculation - but as singe approach to trading is not enough often.
On previous thoughts for external Strategy Tester, I have made many combinations for testing, here is one example of Optimizer/Indicator for existent indicator in the Code Base but, this is again not enough accurate because it is not using M1 or tick data if used on higher time charts, and again as alone stand strategy is not enough - we know history repeats itself but in trading it is not the same as people creates and makes the history so also must be made strong (not long heavy - but strong as strategy possible by experience) calculations. In difficult periods what we call Period Where No/Or Yes volatility when is changing often Take Profit and pause fixes the situation, but that is again as currency strategy approach good from my side.
By my decisions if I share it, when I loose in guessing, I often change the period on bigger time period, and I always relly more on bigger Time Frame Chart - Directions than the lower/smaller, often small entry and waiting for swing for main entry. And also early catch trade talking as type of successful made trade is better than latter one, which could probably start at the end of trend, now we are talking about currency trading what is not same in case Stocks and rest (there apply other rules - stocks: overvalued and later leveling-back as often moves and happens).
If someone wants to put all that in auto calculation and trading, without mistake or at least 99% as planned, there is lot of codding work and checking if it is right as strategy that if it is right as working code program as wanted - it doesn't matter if the program will even not go Live. What ever chosen if someone wants to achieve the target and something what is planned have to work a lot and hard to succeed in that, if he/she wants job to be done.
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Two things...
Is it possible to have the Strategy Tester use the history data from a CSV file of ones choosing? Without converting?
Is it possible to write an external Strategy Tester using tick data to interface with MetaTrader?
Some developers don't bother using the Strategy Tester or DEMO accounts. Does this imply the only accurate testing environment is LIVE forward testing?