Andrey Voitenko: Programming errors cost me $15,000 (ATC 2010)

Automated-Trading | 15 December, 2010

Interview on Automated Trading Championship 2010 from 15.12.2010.

How did you start using automated trading?

In I have come in trading through an Internet advertising and my close friends in 2008. At that time I had a non-intellectual job: despite of the higher technical education and my post-graduate study I couldn't find a proper job. Until recently I had been working as tiler and doing other construction works. So I wanted to try myself somewhere else.

I almost immediately opened an account and started trading. I've gone the same path as most traders do: the first disappointment, loss of my deposit and a complete misunderstanding of how to trade profitably.

At that time MetaTrader 4 was extremely popular, but I didn't appreciate the possibilities of automated trading and was learning to trade manually. Later I realized that there are lots of distractions that interfere from monitoring the market and trading strictly adhering to a trading system. From that moment I began to study MQL4. However, frankly speaking I didn't believe in automated trading until the ATC 2010!

There were also disappointments. Initially this was due to the fact that none of the trading systems could operate profitably in the long term. Now the frustration is connected with the fact that the market is like a living organism, it has its unexplored habits that I find hard to describe in a program code.

By the way, you're very active in our new Jobs service developing Expert Advisors to order. Can you tell us about this?

As I've mentioned, due to poor health I quit my previous job and started working at home. This service was an exciting opportunity for me to implement my skills and earn some income. The Jobs service gives me the opportunity to communicate with different people from around the world and participate in their projects. The main feature of this service is the ability to appeal to a third party to resolve difficult issues. And it is very important due to the problem of fraud related to payments for services ordered from freelancers on the Internet.

Andrey, you didn't participate in previous Championships. What prompted you to participate this year?

Oddly enough, it's not money. I wanted to experiment and make sure that my trading system, which was giving a small profit on synthetic tests, would be able to give it on a real account. The feeling that I could earn with it came when my EA entered the top ten.

Actually, if we take into account the results obtained in the previous two Championships, and extrapolate them to this one, my Expert Advisor wouldn't be able to give the desired profit to win the competition. I wanted to be in the top 50, and I discussed it at the start of the Championship with my friend LeoV.

Is it a surprise for you that the Expert Advisor of Leonid Velichkovsky is unsuccessful on this Championship?

Well, in fact I was involved in the unsuccessful performance of LeoV on the Championship. Because as I wrote for him the Expert Advisor and done a lot of errors. I hope that Leonid has already forgiven me.

On what principles does your Expert Advisor trade?

The work of my EA is based on the breakdown of the horizontal channel. First the levels of the channel are calculated on the basis of the minimum and maximum prices over a period of five-minute bars. If the price goes deep into the channel, two pending orders are placed on its borders. Then only wait for activation of one of them and fix the profit or loss. This system works well on the periods when the market moves from the area of low-volatility to the high volatility area. The choice of a five-minute period was made during the test optimization - I wanted my EA to trade quite actively.

I can also tell you about mistakes made in my Expert Advisor. I've already heard several times, from different people about one of them, which is though not a real bug, but rather a non-standard implementation of a trailing stop. Because of invalid conditions in the EAs code SL is immediately trailed too close to the market, and this prevents the EA from holding the position and fixing profits. I've corrected this part and published on my site the second version of the Expert Advisor with the implementation of the classical trailing stop. My frivolous experiments with trailing stop cost me the prize position in the Championship.

If not this, at this point I would receive an increase in the balance equal to $15,000 and would be in the third place, and hence could expect the cash prize in this Championship. So I bitterly admit that the programming errors cost me $15,000. This will be a serious lesson for me.

Why are you trading five lots from the very beginning?

A properly constructed system of money management can bring to a success even the simplest strategies. However, I didn't include it for the Championship, though it is implemented in my EA. As the saying goes, slow and steady wins the race.

The money management system that was disabled for the Championship is one of the simplest ones, the lot is calculated as a preset percentage of the current free margin. If you apply this system, the risk is no longer fixed and depends on the available funds. Thus the EA is switch to a more aggressive trading, which would be timely for the Championship.

How was your pre-testing? What were your results?

The preliminary tests showed even better results than I had expected. That is why, having sent my robot one month before the deadline, I the changed nothing in it. Though there were things to change. Now I wish I had done that

Firstly, I would correct the logic of the trailing stop mentioned above. And secondly, to win the Championship, of course, I should have used the maximum allowed lot size.

Since my Expert Advisor is available to everyone, I receive different suggestions of how to improve it:

• Like add trading on the pivot from the channel border - for example if it is quite wide. (Remember, now it is trading only on the breakthrough.)
• Or binding some EA parameters to the width of the channel instead of using a fixed value.
• A more advanced module for managing capital.
• And the most labor-intensive improvement is making the channel inclined rather than horizontal as it is now.

However, it will be quite another Expert Advisor. But I am ready to make efforts for its natural evolution. My Expert Advisor is an open project, and its future versions will always be available on my site.

Why are you trading only by stop orders, and not on the market?

Apparently this makes it easier to trace its logics which is an important point for the Championship. In fact, the levels are specified in pending orders; and they would be meaningless when trading on the market.

In this Championship you are using an Expert Advisor from the article. Didn't you have enough time to prepare something new or did you think it's good enough for the competition?

Well, to write the article I needed a profitable Expert Advisor. Profitable, because the idea of the article was to show the ways to withdraw profits. It was also planned for the Championship, but I didn't think about improving it because I didn't have time.

You have a quite smooth balance curve. What optimization criterion did you use?

True, I am praised for the most beautiful curve of the balance, after the EA of Manov of course. Although I didn't optimize by the minimum relative drawdown intentionally, only by the maximum balance - cause it's for the Championship. And the point here is not only optimization, but also timely profit-taking, that is, the algorithm of the EA.

Your Expert Advisor closes deals by Stop Loss, although Take Profit is also placed. What is it connected with?

Actually there were several WA versions: one version worked by SL and TP, and the second one was a derivative of it with a trailing stop, in which TP was like the tail of the previous versions. Since the trailing stop algorithm was modified several times, now it's implemented so that the TP will almost never trigger - it is set too far. That is, TP is not needed if there is a trailing stop. The parameters of the Stop Loss and the trailing stop were chosen as a result of optimization, and the Take Profit is set so large in order not to let it interfere.

Your drawdown is about 50%. When did it happen and why?

It happened at the beginning of the Championship, when the balance was small. Why? I haven't analyzed it, although the December 6 was a significant date: a large loss due to sluggish movement, with the price repeatedly returning to the border of the channel opening and closing orders by the stop. Now I'm blaming the trailing, because with the classical trailing implemented in the second version of my Expert Advisor, the equity curve is much smoother!

Do you like any of participating EAs?

I believe that the future belongs to multi-currency Expert Advisors, so I'd like to single out Manov's EA. It has a very beautiful balance curve. I would also like to mention my competitors - those who have been long listed in the top ten and their EAs are time-tested.

I also had an idea of creating a multicurrency EA, but there was not enough motivation - apparently, I initially didn't believe that I could get into the top ten.

Whose EA in your opinion deserves the first place?

You know, the market will determine the winner, while we will only have to agree. I find it hard to judge, but again I go back to the Expert Advisor of Manov: its parameters are optimized by the minimum drawdown, and if the author had set a larger lot, then probably there would be no one to compete with this EA.

You probably have noticed that the leaders of the former ATCs are not showing good results this year. What do you think, why? 

I've already mentioned that the programming errors didn't let me go higher in the Championship table. Perhaps the same reason influenced their performance. Although I do not see anything wrong in that different EA developers win the competition. In my opinion, the success of the new champions is in good motivation and time to prepare for the Championship.