MetaTrader 5 Strategy Tester Released
Over the weekend MetaQuotes released yet another new beta version of MetaTrader 5. This one is build 268, dated April 24th, and as we speculated earlier this month, at long last MetaTrader 5 now comes complete with a strategy tester!
We had one or two problems getting it up and running. The first was that MetaTrader 5 still contains no equivalent of the MetaTrader 4 history center. That's because it's no longer necessary! We discovered via the MQL 5 forum that one way to load historical data is by opening up a chart on a monthly timeframe and then scrolling back into the dim and distant past. I managed to get back as far as May 1993 that way. Alternatively you can also write a script to grab the data for you.
MetaTrader 5 now comes with a couple of example expert advisors, so we quickly backtested them on our newly acquired historical data. This revealed that the MT5 strategy tester still suffers from one of the main limitations of the MT4 backtester. The highest available resolution of the historical data is still one minute bars. Still no tick data, and still not even 10 second bars. Another limitation is that visual backtesting and the option to display your backtested trades on a chart are not yet available, although apparently they are on the way.
MetaQuotes have introduced one interesting new innovation though, which is a form of distributed computing. If you are doing a long optimization run you can farm out chunks of computation to what MetaQuotes call "agents" running on remote computers accessible via the internet.
Then we tried to backtest one of our own "robots", which proved to be something of a disappointment too. We tweaked it to print out the spread between the quotes it was receiving from the MT5 backtester, which was another major problem with the MetaTrader 4 strategy tester, and it gave us its answer. Zero! It appears as though MetaQuotes still have some work to do in this area, and it doesn't fill us with much hope that MetaTrader 5 simulated spreads will be any more useful than they were in MT4.
The final disappointment was that we were unable to reproduce the surprisingly good backtest results we obtained for our own expert advisor using MetaTrader 4. An EA which has worked fine in "live" testing on a demo account since the day MetaTrader 5 was first released for public beta testing failed to function properly when backtested using the MetaTrader 5 strategy tester.
We'll keep you posted on developments, but our own entry for the MetaTrader 5 Automated Trading Championship has made zero progress today.
Filed under Trading Platforms by 
Leave a Comment
Comments on MetaTrader 5 Strategy Tester Released
MT5 will be a flop! I'm 100% sure of that unless some radical chages are made soon!
Ask yourself this:
1) Why would anyone bother swithing to MT5 and to code in new language that is not compatible with old platform that has very wide userbase.
2) Why would anyone switch to MT5 just for some additional time frames when the most wanted frames are still missing (Tick charts (constant volume/tick) and renko/constant range charts)
3) Also MT5 backtesting has no visual mode
I mean for God's sake didnt these people bother to ask anyone what traders and programmers want!!?? I understand that this group is not Metatrader clintel ( which is BROKERS!) but these people actually use the platform for trading and programming!! And I guaranytee I'm not alone with my opinion, this opinion is shared by ALL the programmers I knbow, and none of them is recommending MT5!
Here are some things that they actually managed to get right:
– Object oriented language paradigm
– Unlimited number of arrays and size
– Dynamic data just like C and C++
– Dynamic allocation of memory (not very usefull feature)
Hi Jani,
I'm not a huge fan of MetaTrader 5 either. However if I can play devil's advocate for a moment whilst I try to answer your questions:
1) Switching to MQL5 from MQL4 isn't easy, but switching to MQL5 from a serious OOP programming language like C++, C# or Java has become easier.
2) They almost certainly wouldn't!
3) The MT5 strategy tester does at last have visual backtesting plus 3D visualisation of optimisation results, and distributed optimisation is a useful feature. You can now backtest multi-currency strategies too.
Here are some more things that MetaQuotes managed to get wrong:
– If they added OOP why didn't they add decent event handling as well?
– If they beefed up the backtester why didn't they add support for ticks with realistic spreads?
– If they added the facility for non-programmers to create automated strategies why didn't they include features that might actually help those folks make a few bucks, instead of just the same old indicators?
Cheers,
Jim
Spot on guys. I can't get over my head on why time frames below 1minute are not offered here.
It's not just them; even TradeStation is missing that.
I recently switched to using Java 100% for my trading systems development. I still use MT4 to visualize some of my indicators though..:)
Hi Ismail,
TradeStation has a vastly better range of bar/tick charts than MetaTrader 4 or 5, although perhaps not everything one might wish for.
If you're just getting into Java then you might like to take a look at our Dukascopy Java source code examples.
Cheers,
Jim