MT4 + Tickstory vs MT5 Backtests

Hello,

how would you judge the backtest results of MT5 compated to MT4 + Tickstory ? I found both platforms to deliver total different backtest results, not even similar at all. I know one it time based and the other one tick based. Just .. which one is better or more "real" ? I would have hoped to get at least similar results to validate my trading systems but its kind of confusing ..

Run both of them (the ea for MT4 and MT5) on a demo live account. Then re-run the backtest for the period you have the live data from. Compare the results.

I have recently switched the MT5 so I would be very interested to hear your results.

PS, when you say back testing are you using the optimization? One thing that popped in my mind is the use of the genetic algo, are you using this in MT5? If so it doesnt display all results (including positive ones) it just converges to the best parameters so the results might appear different if compared to MT4 optimization.

Live demo accounts make sense to compare, thx. With backtest I mean to take the best optimization result from MT4+Tickstory and let it run on MT5 to doublecheck the results. MT5 is very powerful for mutli-currency testing, just I'm not so sure if the backtest results can be trusted if the data quality is kind of unkown.

Please post your results once you have done the test. I have been meaning to do it myself to confirm which is more accurate, but I keep getting distracted with other stuff... I have read plenty about it but it would be great to have a real life test to verify.
Also note that MT5 will possibly (does on my broker) have tighter spreads which could effect entries and exits.

Best of luck.

So here a MT4 result with Tickstory data, 4 years:



http://www.myfxbook.com/strategies/randomfun/70312

Original Disussion: http://www.forexfactory.com/showthread.php?t=503399


Same thing in MT5, same input params and date range:




MT5 EA included in the full report:



Now I'm not a born MQL5 coder but hope the MT5 EA is bug-free. If it is, than either MT4+Tickstory or MT5 show wrong backtesting results.
The last 40% of the curves look similar I would say. The final ends are very different. And the first 60% are completely different, too.

I heard that Dukascopy had a messy time in the past regarding tick quality. Just we have no clue if MT5 tick quality is any better ..

Well, its nearly 5am so someone better repeat these tests due to possible sleep-bugs here ..

Thats really interesting, in Mt4 you made a fortune in Mt5 you are almost broke.

I'll have a closer look later when I get a chance, but analysis aside the only way you are going to know is running the ea on live demo account(s).

(I havent looked at the code, but I suspect a bug as its unlikely the results should be THAT different).

Demo accounts running, Alpari UK, MT4 Market account vs MT5 ECN. Spread is worse on MT4 for EURUSD but with a TP of 165 $ that should not be important .. ​

oh good stuff. Well the results of that testing will answer things. Will take a look later when I get home from work.

Yes, but we have to wait 4 years for the results lol

Maybe a high frequency scalper EA would be more useful. Let it trade for a month, than backtest both and we should have something to compare.

This has been a good exercise for me to do for my own peace of mind. I am of the opinion that there is no real difference between a 99.9% MT4 strategy test VS the standard MT5 strategy test.

You therefore have an error in your mt5 code. I had a quick look at it and nothing stood out but I am the wrong person to check code. MT5 is new to me and I am not a programmer.

To verify the strategy testers are the same I used an ea that I have in both mql4 & mql5. I have copied the results below. Sorry the MT5 came out as an excel file so I just cut and pasted it. Both had a closing balance of approx $80k but there are a few differences. One thing is MT5 has 'deals' where it combines entries to a single price point. This has effected the win/loss percentages. Also I noticed the relative DD is higher in the MT5 test, I cant explain that one.

I believe MT5 strategy tester is far superior in functionality and now I have this to back up its accuracy I wont be bothering with all this downloading of tick data nonsense you have to do in MT4.

IMHO backtest should only verify the strategy is working as intended, not mentally bank money you could potentially make. Therefore MT5 is good enough to do that job.

Anyway, best of luck with it.

Strategy Tester Report
Alpari-MT5 (Build 975)
Settings
Expert:SuperSimpleSystem
Symbol:EURUSD
Period:H1 (2012.08.01 - 2014.08.01)
Inputs:ATRTimeFrame=16408
ATRPeriod=14
ATRPercentage=40
MATimeFrame=0
MAPeriod=40
MAMethod=1
MAPrice=1
MinimumLots=0.010000000
MarginUsage=0.01000000
StepLots=0.03
CountStart=25
MaxTrades=43
AllowedDeviation=10
Broker:Alpari (UK) Ltd.
Currency:USD
Initial Deposit:10 000.00
Leverage:1:100
Results
History Quality:99%
Bars:12 388Ticks:52771 009Symbols: 1
Total Net Profit:68 362.79Balance Drawdown Absolute:0.00Equity Drawdown Absolute:2 943.32
Gross Profit:79 628.80Balance Drawdown Maximal:6 250.23 (12.34%)Equity Drawdown Maximal:21 054.77 (39.61%)
Gross Loss:-11 266.01Balance Drawdown Relative:12.34% (6 250.23)Equity Drawdown Relative:39.61% (21 054.77)
Profit Factor: 7.07Expected Payoff: 243.28Margin Level:95.67%
Recovery Factor: 3.25Sharpe Ratio: 0.29Z-Score:0.48 (36.88%)
AHPR:1.0077 (0.77%)LR Correlation: 0.99OnTester result:0
GHPR:1.0074 (0.74%)LR Standard Error:3 581.03
Correlation (Profits,MFE): 0.69Correlation (Profits,MAE):- 0.07Correlation (MFE,MAE):- 0.67
Minimal position holding time:0:00:27Maximal position holding time:115:00:03Average position holding time:23:15:54
Total Trades: 281Short Trades (won %):143 (95.10%)Long Trades (won %):138 (95.65%)
Total Deals:3 507Profit Trades (% of total):268 (95.37%)Loss Trades (% of total):13 (4.63%)
Largest profit trade:5 702.39Largest loss trade:-5 731.36
Average profit trade: 297.12Average loss trade:- 866.62
Maximum consecutive wins ($):68 (16 412.02)Maximum consecutive losses ($):1 (-5 731.36)
Maximal consecutive profit (count):16 412.02 (68)Maximal consecutive loss (count):-5 731.36 (1)
Average consecutive wins: 21Average consecutive losses: 1

Thanks for your report ! Very similar, indeed !

Seems I start to like MT5 now lol It's amazingly simple for multi-currency stuff ..

Dude its the optimizer that kills it! So so so powerful. I have been reading up on the OnTester() function and you can pretty much get the genetic algo to calculate whatever you want.

Just need to learn that pesky C++ better!!

Someone just asked me the following and I thought it would be helpful to answer in public so someone in the future my also benefit.

To: Innate | From: jkrick33 | Sent: Dec 30, 2014 1:07pm
---------------------------------------------------------------------------------
I noticed a post you had made regarding tickstory, i had done a search because im having some issues with it and looking for someone to help, maybe you might be able to . ive downloaded 10 years of tick data and wanting to launch the mt4 platform, but when i do launch, the backtest does not test for the entire 10 years, is this common, could you help by chance, thanks so much, jkrick33

Hi,
The issue is the fact mt4 cannot handle a data file size over 4gb (about 4.5 years). I am not aware of any workaround. The only way to do it is to break you tick files in to blocks of <4gb then "glue" your results back together. Not ideal.
MT5 does not suffer this issue.
Hope that helps!

Hope these links helps anyone....

http://www.futuresmag.com/2014/10/02...avoid-curve-fi

http://www.traderslaboratory.com/for...e-fitting.html

How could I get a mt4 for tickstory that support 4 gb of fxt files?
Thank you very much

Hi, I just purchased the Tickstory Paid Version (Standard) and I don't receive no email from these people. I have an activation key, but no use for it, as I don't know what to do.

Does anyone have any idea how to solve this and where I can download the Standard version of tickstory so to activate my account?

Thanks guys!