Can We Run Reliable, Computationally Cheap System Simulations on Metatrader 4 ? : Using the Control Points Method Accurately

When you go online and read about running reliable and accurate simulations with Metatrader 4 the first thing you are told is that only 90% modeling quality run on “every tick” is useful – and even then not all the time – while other methods such as Open and Control Points are utterly worthless and should never be used. When I first started testing systems it seemed pretty odd that the guys at Metaquotes would invest time in allowing us to choose two additional simulation methods if there wasn’t any good reason to allow it. Then – after a lot of experience in trading and simulating systems – I realized that those methods were there because they are much more computationally efficient than the “every tick” method. However, the catch is that you must be VERY careful in how you design your trading systems such that the “every tick” and cheaper methods will yield EXACTLY equal trading results.

The people who claim that these methods are “useless” generally do so because they found out that the systems they tried to simulate gave “incredible” results on cheaper computational methods while they failed abruptly when simulated on “every tick”. This is not a consequence of the lack of “usefulness” of the cheaper methods but a lack of knowledge from the person who ran the simulations as awareness about the inherent limitations of each method is absolutely necessary to obtain good results. Within the next few paragraphs I will tell you what precautions you need to take into account and how you can design systems so that the results of the control points and “every tick” methods are exactly the same, with about one thousand of the computational cost.

What is the Control Points method ? This method is the cheapest way to simulate a trading system and uses information only of the current and upper time frames to run a simulation. This means that all the information you have is the high/close/open/low of the bars in the past on the same and higher time frames with absolutely no idea of what happens within each inside-bar’s movement. This of course poses great limitations on the systems that can be simulated accurately as they must not rely -whatsoever – on the information provided within the current bar. If you want to get simulations that can give you the EXACT same results on “every tick” and “control points” modes you need to take the following precautions :

  • The method should only open or close trades on bar open
  • Only information derived from the high/low/open/close of already CLOSED (past) bars can be used
  • Absolutely no information from lower time frames can be used
  • No pending orders
  • No take profit values can be used
  • No stop loss values can be used
  • No trailing or break even mechanism can be used

Basically if you follow the above guidelines you will be able to produce a strategy that gives reliable simulations – at least as reliable as the 90% modeling quality “every tick” mode does – using the Control Points option. As you see the use of lower time frames or any form of take profit, trailing stop or stop loss is forbidden as the accurate profit or loss determination of these trades depends on inside-bar movements (as we need to accurately tell when the TP or SL was reached). This effectively limits us to the use of systems with internal logic based closing mechanisms that use the last bar’s (or previous bars’) values on the current time frame to close positions. Even though this prevents us from evaluating a wide arrange of systems it allows us to very quickly evaluate systems that use most common indicators for opening and closing trades based on past crosses, divergence, derivatives, etc.

If you are wondering how large the increase in efficiency is I can tell you that a 10 year simulation using 90% modeling quality of an RSI cross strategy takes about 10 minutes while the same strategy merely takes 5-15 seconds to be evaluated on a Control Points mode. The results are exactly the same as the strategy takes all the above mentioned precautions into account (see image above for a comparison). When using such fast simulation modes we are allowed to use a great variety of different techniques – which are impossible to explore in “every tick” under current desktop computational power levels – to develop and explore new trading strategies.

The above idea opens up the way to many new and exciting projects which I will be sharing in Asirikuy during the next few months. It also allows anyone to quickly evaluate different ideas for the development of trading systems and to get accurate estimations of profitability and draw down levels when the above considerations are taken into account. Certainly this shows that it is absolute VITAL to have knowledge about the inherent limitations of the simulation methods being used and that no simulation method is “useless” but merely limited to certain system development constraints. As my grandmother used to say, a hammer isn’t useless because it cannot chop down a tree.

If you would like to learn more about system development and how you too can design and trade your own automated trading systems based on sound trading tactics and money management please consider joining, a website filled with educational videos, trading systems, development and a sound, honest and transparent approach 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.

2 Responses to “Can We Run Reliable, Computationally Cheap System Simulations on Metatrader 4 ? : Using the Control Points Method Accurately”

  1. […] few months ago I wrote an article about the “control points” simulation mode and how it could be used to produce reliable […]

Leave a Reply