Interview with Juan Pablo Alonso Escobar (ATC 2012)

Interview on Automated Trading Championship 2012 from 28.11.2012.

"Everyone who is struggling with programming and who were not able to participate in this year's competition, know that it becomes a lot easier in time", said Juan Pablo Alonso Escobar (JPAlonso), the hero of today's interview. Starting from the third week his trading robot has a leading position at the Automated Trading Championship 2012, attracting public attention and causing a heated discussion.

Hello Juan. Please tell us a little about yourself and your trading experience.

I was born in Mexico, obtained my bachelor’s degree in Canada, and currently reside in the USA. I worked for a few years as an equity trader for a Canadian proprietary investment group, where I developed and monitored automated and semi-automated analysis and execution systems. I am currently completing a Masters in Financial Engineering program at Baruch College in New York, and I plan to continue working on algorithmic trading after graduation.

You have participated in Automated Trading Championships 5 times starting from 2007 so far. And each year you've got better and better from 520th place in 2007 up to 1st (so far) this year. In your personal opinion, does this progress at ATCs reflect your own progress as a developer?

When I prepared to participate in the first championship it took me days of sweat and tears just to get a working EA, let alone one that could make money. Now, the technical details of implementation are a lot easier to deal with, and I can focus on making the algorithms stronger and more profitable. Everyone who is struggling with programming and who were not able to participate in this year's competition, know that it becomes a lot easier in time.

The description of your EA is "An optimized counter trend expert". What this counter trend is about? And how do you identify a trend?

I use a series of technical indicators to identify when a short-term trend has over exhausted itself, and then enter a trade against it. The concept is simple. Finding the right indicators for the model was the real challenge.

What criteria have you used when you optimized your EA?

I used back-testing and statistical analysis to find the right combination of technical factors and risk tolerance that would maximize the probability of winning the competition. I did my best to prepare, but I know there are many talented developers participating, and a favorable market swing for a few of them could still radically change the rankings.

What market does not suit your EA? For example, if EURO will go up high like in 2009.

My system is a counter trend expert, so trends are not ideal. High volatility, unstable trends are still acceptable, but any consistently trending markets could be harmful. It would not necessarily be dangerous over the longer term, but in the 1 month window left on the competition it could distort the results.

Please tell us about your prop-trading experience. How it is correlated with programming of Expert Advisors?

Participating for the first time in the 2007 championship inspired me to focus on an algorithmic trading career. The development and trading skills acquired from building EAs and participating in the competitions have been very useful for my professional experience.

Why have you kept silent for so long in your Championship profile and didn't respond for messages?

I blame the hurricane and the elections. But seriously, I wanted to keep a low profile until I saw more concrete results. We are halfway through the competition, so this seems to be a good time to show up.

Your EA trades using the Stop & Reverse principle. It can also close unprofitable long position and immediately enter the market in the same direction. Is this a randomness or a planned behavior?

It is not random, and not necessarily in the same direction. Part of optimizing the EA and making it "Championship ready" was making it simpler and much more aggressive than usual. The original Expert Advisor I was building had more flexible stop/profit targets and it was selective when entering positions. It was stable and consistent... but it had no chance of winning the competition.

The final version still followed the same basic rules when finding the "direction" of the trade, but it was much more liberal about opening positions. While the original EA may have, after closing a position, waited to find a high probability trade before going back into the market, the optimized EA is willing to take medium probability trades as well. That is why the program seems to enter orders so soon, and so aggressively.

Unfortunately, part of what made the EA so profitable may be risky in the last few weeks of the competition. The forcefulness of the orders means that the variance of the portfolio will be extremely high, with average daily swings of 10,000.

Given the volatility of the EA and the expected results of the other participants (based on previous year’s results), I estimate, despite my current 30,000+ lead, only a 50% chance of winning the competition, 15% of ending in 2nd or 3rd place, 25% of staying elsewhere in the top 10, and 10% of leaving the charts altogether. The competition is still open, and I will be watching the results closely.

Do you use Stop Loss and Take Profit set strictly from opening prices?

As mentioned before, I use the pre-defined targets found during the statistical analysis and optimization process. The Stop Loss and Take Profit targets selected were the ones that maximized the expected profitability given the counter-trend strategy. Taking into account expected EURUSD volatility, Take Profit levels are set in the range of short-term reversal of trend.

Why did you opted EURUSD after all? Does it fits better for technical analysis?

I chose EURUSD because I have a decent understanding of its volatility profile, and I feel comfortable with it. I have worked with many pairs, and I don’t think it necessarily fits better for the technical analysis tools I used. The question of using EURUSD was not so much of why, as it was of why not.

What is technical analysis in your understanding? Can it be used?

Technical analysis, using historical prices as a tool to understand the future, can definitely be used as long as you understand its limitations. A positive signal based on historical data analysis does not mean that the market will move in your direction. It only means that, assuming the distribution of returns remains consistent over time, there is a positive expectation that it will.

Are there any currency pairs similar to EURUSD behavior? If yes, why didn't you create a multicurrency EA?

Many pairs do have the same, or at least very similar, short term behavior to be analyzed and traded with the same indicators. Most of the tools used in the EA can be used, with some adjustments, for other pairs. I decided to focus on a single pair to better optimize it for the competition.

What do you think about multicurrency Expert Advisors?

I think that it is definitely an area worth exploring, and there are many profitable opportunities waiting to be found. Some interesting systems can be built using multiple currencies, but I would recommend to new developers that they become familiar with the mechanics of single-currency EAs before jumping in.

Is manual trading experience required to reach understanding of market?

Not required, but recommended. I opened my first live Forex account for manual trading before participating on my first championship, and it was a great learning experience. Manual trading gives you a better understanding of how the market actually works, which can then be applied to the automated tools.

Will trading robots eventually push human traders out of the market?

They will not push human traders completely out of the market, but they will (continue to) drastically reduce the number of human traders needed to maintain an efficient market. Regardless of how fast and logical they seem to be, robots are programmed and monitored by humans, so they still respond to human emotions, fear and greed.

Human traders will still be needed to examine the data, update the instructions on the automated trading programs, and monitor their performance. The trading profession is not going to disappear entirely, but it will definitely continue to become more technical and more specialized.

How can one earn some money using trading robots? Can you leave an EA working by itself for months and go on vacation to the sea?

In my view Expert Advisors require regular monitoring and evaluation. You should at least check them a couple of times a day, and set up a series of alerts to notify you (by email or text messages, for example) when important targets are reached, or when the market conditions have changed. Nevertheless, this is a major improvement over being glued to the screen for 12 hours a day using a manual trading system.

What skills are required to create a trading robot?

A good combination of programming skills and analytical skills is important, but these skills can be obtained through books, online resources, and practice. As long as you are passionate about trading and willing to put in the time, it is possible for anyone to create a trading robot.

What is the most helpful in trading - math or psychology?

Math is more important during the strategy analysis and design phase, and (personal) psychology is more important during implementation and execution phase. You need a system that works, but you also need to be able to make rational decisions when faced with losses and profits.

Are you ready for manual trading?

It really depends on the situation. I do enjoy manual trading and it is sometimes useful when dealing with unusual market situations or more fundamental trades. It is more exciting to execute a trade with a mouse click or a keyboard shortcut than just watching a program do the work, but the same excitement leads to more emotional involvement and more potential mistakes.

Automated trading is much better for timing-dependent and non-discretionary executions. When a trade has to be executed precisely at a price at precisely the right time given a specific set of rules, algorithmic trading is superior.

What can be automated in trading and what not?

A system that uses only numerical inputs (technical analysis) as trading decision factors is easy to automate, but a system that also uses written and visual inputs (fundamental analysis) is not. Some traders are building systems to parse news announcements and reports for automated analysis, with somewhat positive results, but we are still decades away from having software (AI) that truly understands the written language.

What principles is your EA based on (moving averages, oscillator, etc.)?

As some people guessed I do have a few moving averages in the system, but there are other decision factors involved. I will give a few more details if I win the competition.

Well, we wish you success in the Automated Trading Championship 2012 and thank you for the interview.