Thursday, April 03, 2014

Examining the Edge of the Discretionary Trader

Traders commonly speak of having an edge in financial markets--a strategic advantage over other market participants.  When trading systematically, or when using systematized ideas as a backbone for decision-making, that edge is captured through an empirical process:  testing over a market period, applying the test to a fresh period in the market, and validating in real time.  One risk in such a strategy is testing so many ideas over so many periods that eventually some will provide good results merely by chance.  Such "overfitting" can be difficult to detect--and is more common than generally realized

On the discretionary side of trading, it is especially difficult to recognize those with an edge, as audited track records and detailed performance metrics are rarely available to the trading public.  Over the past couple of months, however, I have played market sociologist and used posts to Stock Twits, as well as material published on websites, to identify what active and experienced market participants are doing.  Several of these discretionary participants are actively involved in the training of traders:  these include AlphaTrends, Crosshairs Trader, and SMB Training

(Note:  none of these enterprises are aware that I am writing about them; none solicited my post and I have not accepted--nor would accept--any compensation or promotional consideration for mentioning them in the blog.  I've focused on them because I have interacted with each of them in the past and found them to be serious about what they do.  They have been doing this for a while, have an extensive history of posting ideas to StockTwits, and have gained a substantial number of followers.  All put significant work into their market activity, creating and posting videos and offering specific trading ideas in real time.  I know some traders look askance upon those who charge fees for educational services--"those who can't do, teach"--and Lord knows there are some sketchy providers out there.  One advantage of a platform like StockTwits is that it provides a track record of idea-related productivity and the value of such ideas to a trading community.)

So what are common elements in the strategic edge sought by these discretionary stock traders?  I observe several:

1)  A discipline of pre-market preparation:  All emphasize the importance of process and preparation: sticking to what you do best and being prepared for fresh opportunity--and threat--each market day.

2)  Selectivity:  All have some methods for screening stocks and focusing on a core group that offer opportunity.  Often, these screens focus on stocks that are trading actively, that show good movement, and that are setting up for directional price moves because of earnings reports, breakout patterns, etc.

3)  Patience:  This follows from the first two.  The experienced traders emphasize risk management and waiting for high quality trades, rather than overtrading.  All stress understanding the current market environment and adapting to it.

4)  Diversification:  These traders don't focus on one or two opportunities, but look at a range of promising shares and setups and trade more than one thing at a time.  All the proverbial eggs are not in one basket.  

5)  Simplicity:  My sense is that the traders are focused on understanding what is happening now, not predicting what will happen in the future.  If I had to guess, I'd say that they are talented in detecting the flow of activity in and out of shares and are riding moves as they are getting under way.  They don't appear to be researching deep value and holding for long periods to wait for that value to be realized.

So much of this kind of market edge boils down to pattern recognition, and so much of pattern recognition boils down to practice and immersion in markets.  This may be why the traders I selected all make extensive use of videos in their training:  they are attempting to hasten learning curves by providing more practice time in pattern recognition.

Having observed the tweets and posts of these and other traders in the StockTwits community, I'm left with the question of testing and systematizing some of the elements that they are using for decision-making.  Would adding a backtested component to the discretionary pattern recognition aid selectivity and trading results or would it interfere with a process that is largely implicit and intuitive?  There are considerable promises--and pitfalls--associated with such hybrid trading, but it is a direction I see many skilled traders taking in the money management world.

Further Reading:  Implicit Learning and Trading Performance


Rich said...

Hi Dr. Steenbarger,

First - It's great to see you back posting on Traderfeed. I started my professional prop trading career 8 years ago, and I don't think I'm exaggerating when I say the information and perspectives you shared were absolutely vital to my success during the first few years. Thank you and welcome back!

If you haven't yet, I think you'd enjoy a look at the journal/statistics/analysis site I use:

Doug A. G. said...

Dr. Brett, you are really a great help for us, 'discretionary' traders, I've attended one of your free seminar a few years ago in Pasadena, CA, nice to see you're back and continue helping us improved our learning...

Curtis said...

* Primary risk in backtesting isn't spurious results. But rather that the assumptions that underlie the evaluation may not be true in the future. This second class of problem is far more serious because it applies to all systems regardless of any results they produce.

* Key "edge" of this discretionary is my ability to predict the market both direction and the market's overall regime. Most trading systems can't predict regime. Instead, they trade in a way such that they tend too win more on a probability basis.

* Because of the ways systems operate, they will tend to take fewer trades and make more per trade then discretionary traders. Discretionary traders, however, should be able to profit in more types of conditions. As such, patience is not really a virtue for the discretionary trader but over trading can still be a problem. The one does not imply the other. Over trading is a function of the spread and cost per trade.

* Trading is a specialized activity. There are many classifications and types of graybox systems. There are many ways of improving performance through technology.

"Would adding a backtested component to the discretionary pattern recognition aid selectivity and trading results or would it interfere with a process that is largely implicit and intuitive?"

Below I list the factors that will determine the viability of this conjecture,

1. How many signals does the graybox produce?
* More signals are better then fewer signals. Algorithms need to to produce a good number of trades every day to be most valuable.

2. Are the signals high probability?
* Systems that produce high probability signals are far better then low probability signals. Win ratios of 75%-80% would be desirable.

3. Does the trader have competing signals too choose from?
* For best results, the trader should be able too choose among competing signals. Obviously, every signal shouldn't be competing or there wouldn't be any value. A good mix of long/short signals will give the trader the opportunity to trade with his or her bias easier.

* Note the above refers to process-based trading with graybox assistance. The systematic trading of a strong performing system is a different question.

James said...

Why do you think HF returns have been horrible?