Everything You Wanted to Know About Mechanical Trading Systems
Hey there traders,
Terms like mechanical trading, algorithms, ‘expert advisors’ or EA’s as they are called have become pretty common in the trading world in recent years. But the whole idea of computer generated trading and automation remains a bit unclear to many.
I recently sat down with Daniel Sinnig PhD, to get his thoughts on this exciting field. We discussed the advantages/disadvantages as well as how he helps clients come up with custom solutions or assist them on writing their own trading code.
1. For those who might be new to coding and algorithms give me the elevator pitch.
Mechanical trading systems are powerful devices in the toolbox of both discretionary and fully automated traders. Consistent trade entry and exit, 24/7/365 availability and ability to backtest a strategy prior to using it in a live environment are some of the key advantages.
Mechanical trading systems are useful for traders who have a proven trading system, but are too emotional to trade it in a discretionary manner. They are also useful for discretionary traders to help validate a trading idea or intuition. Let’s say your discretionary approach consists anticipating the close of a price gap after market open or the hit of a pivot. Translating the trading approach into a fully automated trading algorithm will allow thorough testing. Often it can be found that the trader’s discretion or intuition was ill-guided or only applicable in certain situations.
2. What are some of the traps people fall into when building out EA’s?
The ability to test and evaluate an mechanical trading system is both a blessing and a curse. If done correctly, testing allows to forecast the expected profitability and capital requirements of a trading system. If done incorrectly, test results may give completely unrealistic and wrong expectations. For example, using parameter over-optimization almost every trading system can be made profitable during backtesting – but most of them (with almost certainty) will not be profitable in a live environment.
3. I am of the belief that the combo of computer horsepower and trader experience is the best combination ESPECIALLY in this day and age. Do you agree? If so, tell me about some specific tasks/routines you help traders build.
Both automated and discretionary trading have their individual strengths and weaknesses. Automated trading guarantees emotionless and accurate trade execution but lack in “contextualizing” or interpreting trading setups. The latter is the strength of the discretionary trading approach. A human being is vastly superior in identifying a chart pattern or in fundamental analysis.
There is trend towards so-called “hybrid” trading systems where the trader will manually identify a set-up. This could be as simple as drawing a trendline or a support / resistance zone. The robot will then use the trader’s input to execute the trade based on a set of predefined entry and exit rules – thus leveraging the best from both worlds.
I was happy to see Dan go into this. My friend James Cutting over at Nautilus Capital is a big proponent of combining manual and discretionary trading. His opinion matters….his client list includes the likes of Paul Tudor Jones and other mega hedge fund managers. (James will be presenting at the Trader’s Round Table at The Ranch too!)
4. So, assume I am new to the whole programming scene but I have a Meta-Trader, MotiveWave or ThinkorSwim platform. Where does one start? What can a guy like you do to help jump start that process?
ThinkOrSwim is great for trading options and market scanning. MetaTrader is supported by almost every Forex broker and is a reliable and robust platform for 24/7 trading. MotiveWave is probably the best charting software for technical analysis and provides a rich set of built-in indicators and studies. Each platform can be customized and extended by built-in studies, indicators and strategies. Picking the right platform depends on the needs of the trader, the degree of automation, and the equity classes traded.
5. What do you see right now as one of the better ways to use computing power to help uncover trading opportunities?
One of the key advantages is the ability to scan thousands of symbols across different markets for trading setups. Once the computing software has identified a handful of promising setups the trader will use his discretion to “cherry-pick” the best setup most promising. Next a trading algorithm can be used to execute the setup.
6. You are going to be joining 6 other pedigreed traders at The Round Table at The Ranch – what can attendees expect to hear from you in your presentation – what will they walk away with that they can apply immediately?
Unlike the other traders, my background is not in finance but in software engineering and math. I will be sharing concrete insights on the technological requirements to develop, test, deploy and monitor an automated or semi-automated trading system.
7. What percentage of your clients are completely mechanical versus a combination mechanical and discretionary?
Roughly 80% of my clients are looking for a fully automatic trading system that does not require any discretionary intervention. Fully automated are typically easier and cheaper to implement. They are also easier to backtest as they do not have to account for the trader’s manual discretion.
Semi-automated trading system are usually more complex as they need to interpret and react to the trader’s manual input. This could be as simple the trader drawing a trendline or more multifaceted where the trader provides a multi-level Elliott wave count.
8. To get a basic mechanical system in place or a data mining script, what can a new trader/programmer expect in terms of time to completion and having it be reasonably effective?
The turn-around time from an initial vision of the trading system to a fully tested and deployed solution can be as fast as 1 week depending on the trader’s availability and the complexity of the trading system. A typical development cycle includes frequent interactions with the trader to discuss design alternatives and optimization opportunities.
9. What are the common pitfalls new coders make? And what would you offer in terms of advice
I think the single most dangerous pitfall is to implement or customize a trading system which is ‘curve-fitted’ to perform well during backtesting. Backtesting is always based on historical data which is not going to be repeat itself in the future. A curve-fitted trading system is almost guaranteed to fail during live trading. Luckily, there are other effective techniques to evaluate the true performance of a trading system such as walk forward testing, Monte Carlo analysis and live testing.
10. What will your presentation at The Trader’s Round Table cover? What can attendees expect to gain from it?
During my presentation, I detail the various steps involved in developing and deploying an algorithmic trading system. I will explain and demonstrate the do’s and don’ts when it comes to testing trading systems. During the breakout session on “developing algorithms” participants will learn step-by-step how to code an automated system from scratch and how to evaluate its profitability.
Want to learn more from Dan and 6 other industry experts? Attend our upcoming Round Table at the Ranch. Click the image below to learn more.