Mismatch between live forward testing and back testing | Forex Factory

Options

Permalinks
Print Thread

Similar Threads

Back and forward testing EAs aka Trading Robots 6 replies

Data window information and Journal data mismatch 5 replies

Wormholes - stochastic mismatch system 305 replies

Discrepancy between forward testing 1 reply

MT forward testing and going live 5 replies

Mismatch between live forward testing and back testing

  • #1
  • First Post: Oct 15, 2021 5:57am Oct 15, 2021 5:57am
Hi,
There is a mismatch between live and back testing of my EAs' order open and order close executions. I use 10 EA's on 4 instances and at various TFs. Each written by myself and each use different TFs' OHLC data as well as indicator data within themselves. (so I keep these TFs chart always open so that there will be no missing data).
But still EA s close or open trades 20 % of the time wrongly. I am suspicious about iCustom() function which I frequently use for the indicators I have written. (Should I better embed the indicator algorithm in the EA?)
I never use NULL and always write the symbol name instead to avoid any problem.
(Should I give each variable different names with different suffixes so that each variable set is unique for the particular instance and TF so that there will be no conflict?)
But the problem is still valid.
What do you think I should check else.
(I read yesterday in internet that each indicator should be attached to each chart which I did today and am monitoring.)
  • #2
  • Oct 15, 2021 7:05am Oct 15, 2021 7:05am
Quoting saunders
Hi, There is a mismatch between live and back testing of my EAs' order open and order close executions. I use 10 EA's on 4 instances and at various TFs. Each written by myself and each use different TFs' OHLC data as well as indicator data within themselves. (so I keep these TFs chart always open so that there will be no missing data). But still EA s close or open trades 20 % of the time wrongly. I am suspicious about iCustom() function which I frequently use for the indicators I have written. (Should I better embed the indicator algorithm in the...
You can't compare Live with Backtesting ... You don't have the real live market conditions in backtesting. As simple as this.

Backtesting should be used exclusively to test the accuracy of your code (technical rules) ...
MT4/MT5 EAs/Indicators/Alerts coder
  • #3
  • Oct 15, 2021 7:36am Oct 15, 2021 7:36am
Quoting saunders
Hi, There is a mismatch between live and back testing of my EAs' order open and order close executions. I use 10 EA's on 4 instances and at various TFs. Each written by myself and each use different TFs' OHLC data as well as indicator data within themselves. (so I keep these TFs chart always open so that there will be no missing data). But still EA s close or open trades 20 % of the time wrongly. I am suspicious about iCustom() function which I frequently use for the indicators I have written. (Should I better embed the indicator algorithm in the...
Adding to the true comments by FerruFX, you first need to ensure that your backtesting represents how your EA will work in the active market. E.g. If you are not controlling trades based on candle open/close times then you will need to backtest on a good set of tick data. Also, does your EA filter for conditions such as spreads or swaps? If so then these conditions are dynamic in active trading but static when backtesting. This difference can lead to trades being delayed or skipped completely in an active market.
Always read Post #1
  • #4
  • Oct 15, 2021 7:45am Oct 15, 2021 7:45am
All brokers are dishonest check this out All Traders MUST WATCH!! MT4 Meta Trader Forex Conspiracy Proof - YouTube, most traders dont know about this, It is better if you back test your strategy on the futures market for more accurate results since its centralized and more regulated (they canīt change candlesticks like FX brokers).

AG
SDTAG
  • #5
  • Last Post: Oct 15, 2021 4:14pm Oct 15, 2021 4:14pm
today my Ea's opened 15 positions and closed 10 positions. among the 15 positions that were opened only 1 of them did not opened properly when I compare with the daily backtesting. And again among the 10 positions that were closed 1 of them did not close properly (meaning, did correspond to the daily backtesting). For example for the latter case the positions should have actually be not closed according to backtesting . but in real time it closed.
In all my ea's backtestings I always used "open prices only", they have nothing to do with tick data.
Spreads, latency and such things do not play a role in my problem cause I am not comparing yield difference, I am just checking if the ea opens at right time and closes at right time.
I do not think I am comparing apples with oranges cause all these things happen at one sole demo account terminal only by the way.
how can an Ea closes a trade in live testing and not closes in daily backtesting when they both use the same and identical OHCL and indicator data of different TFs like M30, H1 and H4 of one unique symbol.
0 traders viewing now