Hello, Jimmy. Your EA is performing well in the contest. Could you, please, tell us about yourself? What do you do? What is your education? How long have you been trading?

Yes. First of all, I'd like to thank you for this opportunity to be interviewed by the Organizer of this great competition. I'm a 45-year-old male. I have a bachelor degree in Informatika (Computer Science) from Institut Teknologi Bandung, Indonesia, and the degrees of Master of Computer Science and PhD from Texas A&M University, USA. Until 2003, I worked for our national aircraft industry (PT IPTN) as a software engineer. Currently, I'm self-employed in IT-related jobs. I was introduced to Forex trading around 2004 by a friend of mine when I was looking for a new opportunity to earn my living. Although Forex trading has not given me a steady income yet, its potential is very good. Especially, when we can use automated trading, where we don't have to watch the market day in and day out.

You participated in all our Championships. What lessons did you draw from them? What differences do you see between them?

When I was introduced to Forex trading, I knew immediately its potential to be automated. However, I mostly use Unix/Linux-based systems and rarely use Windows-based ones. So I searched around brokers that support either *nix platform or Java/Web-based platforms. Unfortunately, none has good programming features for automated trading. I didn't consider MetaTrader 4 at that time because of that.

However, the announcement of the first Championship in 2006 really forced me to look at the platform. To my surprise, it is very easy to install MetaTrader 4 under Wine (by the way, you can also see my name on the old list of Wine contributors). So I quickly learned how to use and program on this platform because I thought it was only one or two months before the Championship began. I created a very simple EA based on CCI. Although my result was not very good, I was fired up to join the second Championship in 2007.

I coded almost everything for the second Championship. That was a mistake because the EA became bloated and too complex. It failed miserably. This year actually I almost didn't submit any EA for the Championship. Only within the last few months, I programmed a new EA which was basically a simplified version of the old EA plus using as much as possible supporting library from the platform.

The lessons I got? Well, the Championships themselves are very important as proven milestones for the progress of automated trading.

In the first Championship, I saw that almost everybody didn't know what to expect, and wrote EAs based on their ideals. Many important facts are revealed, such as the quality of strategy, good money management, and other important characteristics of good (competition) EA.

The second Championship showed us what good basic strategies are. We know the champion used some kind of AI at the time. However, using a proper approach, other strategies also revealed. Now we know a very long-term open position to be possible, as also a very short-term trading (scalping) can be profitable. Even very small number of trades can also put an EA at a high rank (as one showed by the disqualified EA from one of my countrymen).

This year's Championship is another progress of the previous year. People use and expand strategies from last year, which are thought to be fruitful. Just look at the top 4 or 5 ranks this year. All of them either open a very long trend-following system or a scalping system.

If next year there is another Championship, people will build new EAs based on what strategies/systems/whatever shown to succeed this year.

People tend to follow approaches that they think were good. But if people think the approach was not that fruitful or deem too complex, they will abandon it. There was a hype during and after the first Championship to use multicurrency systems. Some people used them in the second contest, but it was not shown that fruitful, so almost no good progress in multicurrency systems is shown this year. Another example is AI (Artificial Intelligence)-based system. It had very good results during the second competition, but it failed miserably during the current Championship. I don't expect there will be many AI-based EAs next year. I'm not saying those approaches are bad, it's just they may have to step back a little. Probably someday, these approaches will make a come back. Probably next year will be the year of multicurrency-based systems? Who knows.

You chose GBPJPY in ATC 2006 and this year. Why did you do that? Why did you choose EURJPY last year (2007)?

It's a good question. Well, competition is competition. It is a do-or-die thing. To win, we have to take chances and accept higher risk. I chose GBPJPY because so far it is the most volatile pair. A few-hundred-pip movement can happen within a day. So I thought more pips meant more profit (or loss for that matter). But being fun, I try a different pair each year. Although not the fault of my choice of EURJPY. Unfortunately, last year my EA failed miserably. So I think going back to GBPJPY that I am familiar with is a good decision. Probably next year, if I join the Championship again, I'll choose a different pair, just for the fun of it. So there is no big reasoning behind the decision, just picking high volatile pairs.

According to Statistical Report #2, EAs trading GBPJPY, in average, have been profiting for two months already. What do you think: are they winning "by a fluke" with this currency pair, or is this symbol really distinctive regarding its conduct on this Championship?

Well, I don't know. We do know that this pair moves even more wildly than usual. Gorez does a good job by predicting that GBPJPY will go south for many weeks. Hopefully, it is good enough for him to earn a good position at the end. I don't know much about other pairs. But because it is a high volatile pair, for many EAs there are only two choices: either make lots of profits or go broke. Probably you can check how many EAs using GBPJPY that currently have very little balance leave. I expect quite many of them. My EA actually performs rather bad during such too high volatility. It was almost dead during that time. Fortunately, the market calms down later on, so it can trade better. The profit is not determined by the pair alone, but more by a good strategy and clean code.

Jimmy, GBPJPY traded by your EA fell rather than grew during the Championship. However, your Expert Advisor made 25 BUY trades against 19 SELL trades. Does this mean that it is trading against the trend?

The EA didn't go against the trend. It follows the trend, but the smaller one, which normally happens several times a week. It tries to find peaks and troughs and start to open positions near those areas. Unfortunately, sometimes it makes a mistake, skipping good signals or taking a wrong signal. What happened in early October shows a bug in the program. Because the movement was so fast and so big, the EA missed the train, and it was almost dead stand still, waiting for the price to return. Fortunately, every end of week, many important parameters in the program are auto reset, so it can start fresh the next week.

It is great that parameters can change. Could you give us an example of a changing parameter? How did it change?

Well, for example, when the bug occurred, actually the EA had already made a decision to buy! Supposedly it waited to buy at the very bottom. So it drew a line and was waiting for the price to cross the line before it buys. The line was supposed to be adjusted every few hours or so. However, it didn't update the line. So the line was never crossed, and the EA looked dead. Fortunately, during the weekend the line and the decision to buy were expired, so the EA went back to its initial state, looking for other good signals.

Your EA is performing well in the Championship: a rather high percentage of profitable trades (over 60%) and, at the same time, your average profit trade of $5 326.73 exceeds your average loss trade of $4 802.55. How can you explain this? What did the test show?

So far the statistical data is quite similar to the results of the test. Profit is around 60% (2 times losses), the drawdown has similar values, too, and so are the average profit and loss trades. And its PF is around 2.0. So its behavior is predictable so far. I'm not saying I've predicted the EA will have a big win. I just say its behavior so far doesn't deviate much from the test results. We have to remember that tests are based on historical data, and the test results are also influenced by the quality of the data. Furthermore, the total actual profit or loss is determined by fluctuations in the market. However, I expect if this EA loses, it will not go broke on the sudden.

Your MAE indications are good, too. We can see in the chart that you are using stop orders. How did you select the size of your protecting StopLoss?

SL and TP are determined together with the decision to buy or sell. The TP should be within the predicted range of volatility and the SL should be outside the calculated range. The SL and money allocated for the trading determine the lot size. So, if SL is close to the market price, lot size can be bigger than when the SL is far. Just compare the last two trades of my EA. The last one uses a bigger amount because its SL is small, while the previous one has a smaller amount because its SL is very far from the market price. I think that makes its loses controllable, and shows in the MAE values.

Some orders were closed by SL, others were closed by TP. However, there are orders closed by market. What alerts/situations underlie closing an order by the market?

Yes, in general my EA will leave the closing to SL and TP. However, if it takes too long, the audience (i.e., me) gets bored. So the EA will close it, if it lives more than two days. It also closes a position if it keeps a floating (small) profit for too long or if it is near TP but hasn't touched it for some time. There is a lot of improvement needed to be done in this area, though.

Orders closed by SL have losses that don't exceed 170 points. This value is very small for such a volatile currency pair as GBPJPY. Did you intentionally work for these results?

All SLs and TPs are calculated. Well, except for TP: if the calculated value is too big, the real value is reduced. And yes, most trades that hit SL have too small SL. The program needs an improvement.

Your EA seems to use a Trailing Stop. Is it a simple algorithm of trailing the protective stop level, or do you use a specially developed, complicated algorithm?

It is just a protective stop level that is adjusted from time to time. So it is not really a trailing stop. I don't want to use trailing stop in a traditional way because SLs are very often unnecessarily hit by the market.

The size of positions to be opened is sometimes quite different. What is the reason for choosing one or another amount of lots in an order to be opened?

There are two components: The distance of SL from the market price, and the allocated money. If the balance preservation hasn't been kicked in, the allocated money is just the same as its equity. It is a rather simple calculation so that, when the market hits SL, I am only willing to lose certain percentage of the allocated money without being stopped out. From that, we can get the lot size to be used in this particular trade. So yes, the amount of lots is not fixed but recomputed based on the two components above.

How does the EA open positions - by the market or by placing pending orders?

By market (mostly). But if the precomputed amount of lots says it can open more than one trade, the latter one is placed as a pending order. No particular reason for such pending order, just for fun. Profitwise, probably it could get more profit by placing all orders as market orders.

If it is not a secret, what do the numbers mean that we can see in the comments on each order? They don't look like a normal MagicNumber.

They are for the purpose of debugging. I use them because the comment length is rather short and so is it in log files. I do know what the total size will be, and I don't want to violate the clause about the maximum log size. For example, the first number (before colon) is the capital preserved. The last trade says it is 12, meaning it has preserved (12x5000$=) 60000$. So it is unlikely for my EA to drop below top 10 (hopefully) unless many other EAs' balances reach the values above this size, because my balance will not go below this value. The other numbers are just a marker in the program so that I can check back, why it makes such decision to buy or sell.

Your EA gained its largest profit in November, at the time of no strong trend for GBPJPY. Is there any connection between these two facts?

Most probably, there is some. After the big move in September, I didn't expect the fall to continue as big in early October. So I didn't check the EA against continually big moves as such.

Your Expert Advisor has made 44 trades. They all were opened in the daytime. What is the reason for this?

It is evening and night time in my country. But yes, it trades during European and American market time because of rather low volatility during Asian market.

Jimmy, what about real trading? Have you got any results there?

No, not yet. After I see the results of this EA and fix many bugs I found during the Championship, I'll try it on real trading.

What EAs participating in ATC 2008 seem to you to be most attractive? Why?

Hmm. Those three amigos are really good scalpers occupying the top three for sometimes. They have a steady growth of balance. I really like them, although it is not my style of trading. Strategywise, I like EAs, such as Gorez, abeiks, and the disqualified strelec, considering their approach to the market. To me, they give us a new direction of strategy for automated trading. That is putting some kind of fundamental analysis into the program. Probably, in the next year's contest we will have more EAs with very good and more elaborate fundamental analysis in the codes.

Thank you, Jimmy, for the interview. Good luck!