How To Graybox

Graybox Trading

Jun 06

The difficulty of building your graybox is that there’s no blue print. But, looking to identify both strengths and weaknesses in your trading may be a start:

  1. What do you do already do well?
  2. How can you improve what you already do well?
  3. Where do you struggle? What do you do poorly?
  4. Where are the mistakes?

As per building the graybox, there are many possibilities:

  1. You can systematize your best setups into signals.
  2. You can create automaton strategies that will trade mean reversion or trend very well and then you can limit those strategies to only trading in certain price ranges, times, or turn the bots on or off based on your overall read.
  3. You can create micro-order management strategies that will help you get into and out of trades.
  4. You can run fully quantified/working strategies but manipulate the direction, style, and aggressive that the system can trade.
  5. You can create decision support tools that helps the trader to know how to execute.

Some of the unique challenges of building graybox systems versus traditional quantitative systems:

  1. You need to build a GUI or interface to work with the strategy in real-time.
  2. Many systems that work trade super selectively: these systems are of little value to the discretionary trader for purposes of graybox trading (though can be useful for scaling out/diversifying). You need predictive measures that can give “continuous” edge at all times or at least several signals per day.
  3. You need sophisticated order management capabilities: the ability to manage strategy orders manually and the ability to manage manual orders in a semi-automated fashion.

Discretionary traders tend to be good at predicting direction but poorer at execution. Systems tend to be better at execution but may not always recognize the best conditions to execute. Discretionary traders tend to lose more often when they try to use tiny stops: on the other hand, it also difficult to program systems that will work with tiny stops. On the other hand, the “scale up” approach to futures trading requires extremely tight risk management to be viable. And, even though stop management is difficult to make work for any trade: entry, stop and target management seems like one area where systems might be able to do a better job. One idea is a meta-strategy that can optimize the stop/target based on conditions such that you know the system will find the optimal exit strategy for any market conditions.

About the Author

The author is passionate about markets. He has developed top ranked futures strategies. His core focus is (1) applying machine learning and developing systematic strategies, and (2) solving the toughest problems of discretionary trading by applying quantitative tools, machine learning, and performance discipline. You can contact the author at