Interview with Nikolay Kositsin: multicurrency EA are less risky (ATC 2010)

Interview on Automated Trading Championship 2010 from 27.07.2010.

Nikolay Kositsin has told us about his developments. He believes multicurrency Expert Advisors are a promising direction; and he is an experienced developer of such robots. At the championships, Nikolay participates only with multicurrency EAs. His Expert Advisor was the only multicurrency EA among the prize winners of all the ATC contests.

In 2006 you were the Third Winner with your multicurrency Expert Advisor. Was this just luck, or did your EA behave exactly as you had conceived it?
At that time, the foreign exchange market was fairly easy to predict, and therefore it was quite easy to select the desired strategy even in the multicurrency implementation. The problem was only how accurately the Expert Advisor would fit into the market - whether it would give hundred, fifty, or only thirty percent. As a result, the EA came exactly in the middle of what was required of it.

In 2008 you participated with a multicurrency Expert Advisor again, but unfortunately it was unprofitable. How can you explain it?
For some reason I decided that at the end of the year there would be no significant trends, and I used a non-trend algorithm for the trading system. Moreover, that time I used 10 currency pairs instead of 6 as before. It was natural that in the trend market my Expert Advisor was pretty good for only one month, and then quit the competition.

The conclusion is that the market always has some trends, and one should first of all take into account these trends.

Why do you develop multicurrency robots, and don't do just as most of the participants who write single-currency EAs?
The multicurrency Expert Advisor maximally expresses the essence of an automated trading system. With a conservative approach, a multicurrency EA shows much more interesting trade results. Perhaps, the only serious drawback of a multicurrency strategy is that it requires a more laborious and pedantic preset.

Do you intend to participate in the ATC2010?
Yes. Naturally, again it will be a multicurrency Expert Advisor. Moreover, the rules of the Championship 2010' seem to be leaving very little chance for single currency strategies.

How would you explain the fact that you were the only winner with a multicurrency Expert Advisor? And why are there so few multicurrency EAs?
The thing is that with the rules of the previous championships allowing only three open positions at a time, it was much easier to create a robust single-currency Expert Advisor. This time the situation may be reversed to the opposite! Carefully analyzing the rules of the Championship 2010, I would suppose that this time single-currency EAs are even in a worse position.

What is the functional difference of a multicurrency Expert Advisor from a single-currency one?
It's hard to tell. The code of a multicurrency EA is much more complicated and therefore it must be originally represented in a more functional form. The whole point is on what principles trading systems of different currency pairs interact inside the multicurrency robot.

You can create a functional monarchical center, which analyzes the profitability of your Expert Advisor on individual currency pairs and limits its activity on bad assets. Or you can build a multicurrency system based on the complete anarchy, which will incorporate independent trading systems, each of which will have its own portion of margin in the Expert Advisor. And it is absolutely impossible to say in advance, which of the two variants would be more rational. And that's only two versions of the general concept of constructing a multicurrency Expert Advisor. In fact, there are much more degrees of selection in the design of such EAs.

What are the required skills for developing multicurrency Expert Advisors?
First of all, one should properly learn to write the code in a single-currency implementation, and be able to wrap this code in universal functions. In a multicurrency Expert Advisor, it is much more convenient to first write a code inside the OnTick() function in the form of calls of universal custom functions for the formation of trading signals, functions for trade operations and other auxiliary functions. And after that you can fill these functions with the required program content.

What is the acceptable risk of a multicurrency Expert Advisor?
For the Championship 2010, I will use the following mathematics. The minimal lot is 0.1. Assume that the Stop Loss is equal to 2 000 points. So, the position of 0.1 lots was closed by the Stop Loss. The estimated loss is $200. If the deposit is equal to the initial one ($ 10,000), the minimum risk for this deal is two percent of the deposit when using the minimal lot. A lower risk is impossible in this situation, so this limit seems to be quite acceptable.

It is believed that multicurrency EAs allow to diversify risks. Did you implement this idea in your EAs, or did you focus on the maximal profit?
Well, it's clear from my previous answer that it is yet too far from the maximization of profit. Still, I dare say that multicurrency Expert Advisors are less risky. Simply try to observe how such an EA works. In general, the probability of a simultaneous drawdown of positions on various currency pairs is quite unlikely.

Could you estimate the realistic level of profitability of multicurrency Expert Advisors?
This level of profitability is significantly higher than that of the single-currency EAs. Still, everything is pretty relative. So we can assess only specific Expert Advisors.

What timeframes should be used in multicurrency Expert Advisors and why?
The logics is the following: the more currencies are used in an EA, the more presets it requires; the lower the timeframe is, the more often it needs to be set up! It may happen so that for small timeframes you will need to set up your EA all the time. And, perhaps, it would be more rational to focus on larger timeframes.

What analytical tools do you use in your multicurrency Expert Advisors: standard indicators or your own tools?
I don't use standard analytical tools in their original form, because custom ones are often much more effective. For example, you can take any standard indicator and replace the classical averaging algorithm with some more advanced algorithm. And it can greatly improve the trading system implemented on the basis of this indicator.

Do you develop Expert Advisors on a commercial basis?
Alas, one has to choose what to do - either to sell trading systems, or to develop programs for one's own purposes. The fact is that programming is quite a time consuming thing, while the resources of an individual are not limitless. Moreover, the complete meditative concentration on the solution of one problem creates the insight into the understanding of the solutions in the most ideal way. Well, sometimes people contact me to order Expert Advisors; and I do this if I have time. But no one has ever ordered multicurrency Expert Advisors.