Inspired to better understand the probabilities

I've utilized my limited time recently by going back and re-exploring mechanical trading programming from well over a decade ago.  For the longest time, I have passed by this area of trading with the not-so-accurate attitude of "been there, done that, not interested."

Automated systematic trading is a broad, complicated and complex field that I will never be able to fully grasp, especially on the institutional side.  Now days, this field of trading has fancier names -- such as algorithmic, algo, black-box, robot, HFT, and so on. 

For the individual retail trader, I'm still confident that a very good discretionary trader will be able to consistently outperform a solid mechanical trading system.  This is one of the primary reasons why I've always wanted to become a good discretionary trader and moved away from mechanical systems development.

Then why am I re-exploring developing mechanical systems?

I have been recently inspired by Electronic Local's blog, which is authored by a former pit trader on the LIFFE exchange who made the transition to become an upstairs trader.  I spent the past week going through his entire blog, and can report back that he has been very generous explaining his trading system.  I especially appreciate his "how to learn to trade" process.  Although he now sells training courses, just the free content on his blog is a gold mine for those who want to learn how to day trade the ES, forex, DAX, and other liquid trading products. 

The initial purpose of EL's blog was to document his journey of training his daughter how to become a consistently profitable (CP) trader.  It just happened to be that his daughter was from the IT field, so it wasn't long before she started to utilize technology to help her take what her father taught her and codify it as much as she could into a mechanical trading system.

Interesting use of trading systems

Soon, EL realized that by incorporating the basic overview of his trading methodology into a mechanical trading system had some compelling advantages.  He realized that there isn't a way to distill and program all of his discretionary trading knowledge into a mechanical system.  However, he could setup a "hybrid" methodology where the mechanical system can effortlessly monitor several markets, and alert him or even automatically enter a trade.

Once a position is established in the market, he can quickly scratch the trade if the picture didn't feel right.  He also discovered that he can manage/scale out of the trade much more effectively vs. a static set of mechanical  rules.  One reason is that there just hasn't been a way yet to code his ability determine market "context" from his years of trading experience.  Therefore, the hybrid trading method will enable him to cover and trade more markets, with less work.

EL has stressed many times in his blog, your experience and ability to determine market context is critical if you want to go beyond the performance of a typical mechanical system and approach that of a good discretionary trader.  For example, using the most appropriate algo for the current market conditions is key.  So if you pick a trend following algo on a low volume chop day -- well, I think we know what the outcome will be.

I like the hybrid approach, which seems to take the strengths from both technology and humans to leverage the best of both worlds.  Another approach would be the totally automated approach to utilize several different styles of algos covering various markets, which would be similar to having having your own group of prop traders (or multi-portfolio managers), each with differing strengths and weaknesses.  Ideally, this would smooth out the profit equity curve over time.

Dusting off my programming hat to better understand probabilities

I've spent over the past week dusting off my EasyLanguage programming hat, and have realized that my learning curve has gotten quite steep again.  I actually like this new challenge.  In a short period of time, I've been able to reproduce some of his basic systems in TradeStation and have also been discovering some very interesting observations. 

Another advantage of this return back to programming has been my ability to test out various setups and patterns used in my discretionary trading.  There have been a few examples of setups based on indicators or price patterns that I thought would be profitable over time, but my recent tests have shown them to be lackluster. 

By recently learning those trading setup probabilities, I'm sure I have saved myself countless $'s in losses over the long term.  It's an interesting direction I'm headed -- coming up with various trading concepts, theories, and hunches, testing them, and then determining whether it would work in a mechanical and/or a discretionary trading system.  Being able to code, test and understand my trading setup probabilities has given me a greater sense of confidence.

Where to from here?

It's interesting how EL's "inside out" setups (pullback in the direction of the trend), as well as the "outside in" setups (catching tops/bottoms) leverage and builds nicely upon what I have learned over the past year.  I've only scratched the surface so far with my exploration of systems development, but I believe this will be quite an interesting path on my way to becoming a better and consistently profitable trader. 

Although my summer non-trading schedule is getting even crazier, I hope to report some interesting updates soon...