The Lower Time Frame Problem : Why Metatrader Cannot Accurately Simulate Systems Below the 30 Minute Time Frame (NOT even With 99% Testing Quality)

When I started my journey in forex automated trading one of the first things I wanted to do was to develop systems to scalp the forex market and get massive amounts of profit. I wanted to have a system that traded a lot everyday, a system that was able to ride trends in the 5 minute time frames and scalp to give me a very large amount of compounding and a very big amount of returns every year.  Sooner than later I had a bunch of systems with huge profits in back testing that failed miserably in live testing. On today’s post I am going to explain why you simply CANNOT obtain reliable simulations on the lower time frames for MT4, why there is no way to circumvent these limitations and why developing systems for such time frames on this platform is simply an overall WASTE of time.

Why is it that systems on the lower time frames cannot be accurately backtested? After I developed my initial set of “holy grails” it became apparent that several evident  limitations of the metatrader 4 platform were obviously limiting the accuracy I could get. The first and most evident one was one minute interpolation (the absence of ticks within the platform) that makes the trading within each one minute bar a “guess” and therefore the reaching of small profits/losses and the signals of indicators based on intra-bar trading below the 30 minute time frame a waste of time.

However it seemed that such an approach was evidently “savable” as I could simply use only bar OHLC information (Open/High/Low/Close) to avoid interpolation and achieve “clean results”. The results however where exactly the same, “holy grails” that took thousands to millions in a few years which utterly failed to perform the same way in live testing.

Then about a year ago I thought there might have been some hope with the introduction of “tick data” simulations in Metatrader 4 – through the use of dukascopy tick data and a few tricks – which made it possible to run simulations without interpolation that would give us quite precise intra-bar results even on very high resolution (1 minute) bars. Several people also saw this surge of the 99% quality simulations as the “answer” to their questions about the influence of interpolation.

The truth however was that I could still achieve the same results, holy grails that crushed live accounts. Why was this the case ? Why is there such a high overestimation of profitability and underestimations of draw down when developing strategies on the lower time frames, even when using 99% simulations on many systems? The answer has to do with broker tick data and the nature of the forex feed.

The problem on the lower time frames is that the spread becomes relatively important in relation with the size of the bars. However this not only changes the cost of trading but it also changes the positions and signals generated on each bar as Bid/Ask feeds are DIFFERENT something that can ONLY be taken into account if you can do a simulation with tick data for BOTH feeds. The fact that MT4 does NOT have this option makes this simulation impossible as accuracy in simulations on the lower time frames (below the 30 minute time frame) REQUIRES the use of separate Bid/Ask feeds not only to determine trading costs but to actually NOW the OHLC of each bar and the place where trades are actually entered. Often there appear to be obvious, BIG inefficiencies on the lower time frames that are effectively eliminated when taking into account Bid and Ask feeds SEPARATELY.

Another very important point here is that people underestimate the importance of broker dependency in forex trading when it comes to the lower time frames. Simulating with one broker’s tick data (eve if Bid/Ask separate simulations are used) would still only yield results that would be usable on THAT broker since differences between brokers on the OHLC values of low time frame data can represent as much as 20% of each bar. When the value of bars can be 20% different on different brokers, what type of reproducibility can you expect to have amongst different brokers in the long term? The answer is little or none.

For the people who are interested in the development of very low time frame systems (below the 30 minute time frame) I would have to tell you that you should NOT waste your time with Metatrader 4 since accurate simulations below the 30 minute time frame  (even on tick based 99% modeling quality) are not possible due to the absence of adequate Bid/Ask feeds which should be absolutely distinguished when performing simulations on such sensitive time frames, not only because of the spread – as I mentioned before – but because trade entering positions as well as signals WILL depend on their accurate evaluation. It is very EASY to develop a “holy grail” in Metatrader 4 on the lower time frames due to this reason, something which has been shown by a myriad of commercial systems that have even shown good 99% quality simulations (which are limited to 3 years anyway if using the freely available Dukascopy tick data)  but failed in real testing. Every aspect of the market you ignore on the lower time frames (like separate Bid/Ask feeds) leads to HUGE apparent inefficiencies which are – in reality – not exploitable.

I am not saying here that long term profitable systems on the lower time frames cannot be achieved but if you want to do this then please BE AWARE of the fact that these cannot be simulated accurately in MT4 and that a platform that can accurately run these simulations should be used. Using a platform capable of directly evaluating tick data with SEPARATE Bid/Ask feeds is not only important but VITAL to the realistic simulation of lower time frame systems. Even then, variables such as slippage may play a substantial role that may affect the real performance of the systems something that also needs to be taken into account (assuming a certain slippage on a certain percentage of trades).

I am all for developing long term profitable trading strategies but in order to do so simulations must be ACCURATE and RELIABLE, in the future you should keep in mind that in Metatrader 4 this totally excludes short time frame simulations, even on 99% modeling quality, due to the problems mentioned above. If you would like to learn more about how I develop trading systems and how you too can learn to develop likely long term profitable strategies with reliable simulations please consider joining Asirikuy.com, a website filled with educational videos, trading systems, development and a sound, honest and transparent approach to automated trading in general . I hope you enjoyed this article ! :o)

Print Friendly, PDF & Email
You can leave a response, or trackback from your own site.

6 Responses to “The Lower Time Frame Problem : Why Metatrader Cannot Accurately Simulate Systems Below the 30 Minute Time Frame (NOT even With 99% Testing Quality)”

  1. Brian says:

    Thanks for the excellent article. Even though I don’t comment every day, please know that I read all your blog posts.

    Do you know of a way to acquire the Dukascopy tick data all at once (either free or for $)? I have only found their data split into hours which would require a whole lot of clicking and reassembling.

    In the mean time, I will use the excellent historical data that you have on Asirikuy!

    • admin says:

      Hello Brian,

      Thank you very much for your comment :o) I certainly appreciate the fact that you read all my articles ! I am very glad that you find them useful and informative. Regarding the Dukascopy tick data you might need to use a script to make this practical although Dukascopy can ban your IP from downloading if you do this. When I did this almost one year ago I used such a method, you can google “99% modeling quality tutorial” and you should find some useful results with scripts for downloading, etc.

      Either way I do not believe that you would get much benefit from getting this tick data as it is limited to just 3 years and evaluation in MT4 is extremely time consuming and difficult when attempting to use it (only one year at a time, huge memory consumption, need a hacked terminal.exe, etc). Also – as I highlight on this post – most systems that would benefit from such data can often be affected by other bad problems. My advice would be – if you are interested in tick data evaluation – to buy 5-10 years of tick data for the pair you want, this can be expensive (2000-4000 USD per pair) but certainly worth the cost if you are interested in accurate evaluation using separate Bid/Ask spreads (although you might also have to develop the simulation framework). I would also encourage you to use FXCM’s strategy trader which – although without tick data – does allow you to separately evaluate Bid/Ask feeds from this broker. I believe that you win much more in accuracy by doing this in a much more efficient manner.

      Thank you very much again for your comment Brian :o) I hope that Asirikuy and my blog are helping you a lot on your journey towards long term profitable trading !

      Best Regards,

      Daniel

  2. Jon says:

    You’ve addressed the biggest and most underestimated problem of forex automated trading, with Metatrader or not: non homogeneus data feeds.

    I certainly recommend you to try futures markets, where market data feed is generated by an official market, and is built from real trades made by all market participants, with an official timestamp.

    It’s an independent source, same for all, and whaveter you backtest there you can make real with a reasonable slippage if the market is liquid enough (which you can see because you see traded volume and market depth).

    You may not be able to get 200:1 leverage but 40-50:1 instead (no enough?). But everything else is far more better for serious trading, including transparent trading costs.

    • admin says:

      Hello Jon,

      Thank you very much for your comment :o) I do personally have traded futures and I do agree with all the advantages you mention above. However probably the main reason why most investors who trade forex do not consider the futures market – and use forex instead – is because minimal contract sizes are just much smaller. In the futures market minimum contract sizes are probably in the 10,000 USD mark while in forex you can get contracts from 100 USD (1 USD cent per pip per contract). This means that in practice much larger accounts are needed. The much lower capital requirement is the main reason why I have focused my efforts on forex, because most retail traders will be able to benefit more from this approach.

      Bear in mind that the lack of a central exchange does not mean that profitable system developments are impossible but merely that adequate precautions need to be taken into account to minimize its effects. However in the future you can expect me to expand my research into futures trading :o) Thank you very much again for your comment,

      Best Regards,

      Daniel

  3. […] (99% quality) – are not reliable. I have written a few posts in the pasts explaining the unreliability of their simulations and the inherent disadvantages of scalping strategies. Trading a strategy without knowing its long […]

Leave a Reply


Fatal error: Call to undefined function mcsp_html() in /home1/danielfp/public_html/mechanicalforex/wp-content/themes/modern-style/comments.php on line 71