Tuesday, October 30, 2007

The Value of a Trading Setup

I've received a few emails that lead me to believe I wasn't sufficiently clear in my recent posts investigating the odds of hitting such price targets as the prior day's high and low, the average trading price, and pivot-derived support/resistance levels.

Those probabilities that I reported in SPY from 2004 - present were just that: baseline probabilities. I did not mean to suggest that, in and of themselves, these probabilities offer a trading edge.

Rather, by establishing baseline probabilities, we now have an objective basis for determining the value of a trading setup. The setup could be a chart pattern, an indicator reading, or an intermarket event. The question is: does this setup significantly alter the probabilities of hitting these target prices during the trading day? If the answer is no, either you have to alter or toss the setup or you have to redefine your price targets.

In my subsequent posts, I'll begin looking at the value of adding variables to price by seeing how odds change as a function of those variables. In general, my experience is that many indicators that seem to be valuable end up not adding predictive value to odds simply because they are so correlated with price itself.

But if we *can* find variables that affect the odds of hitting these targets, *then* we have the beginnings of a trading edge and idea. It is from those beginnings that we can add money/risk management rules and generate more rule-governed approaches to trading.


Odds of Hitting Support/Resistance Levels

Odds of Closing Opening Gaps

Odds of Hitting Previous Day's Levels

Trading With the Odds


Brandon Wilhite said...


I'm glad you posted this, as it clarifies things for me, since the statistics you've been posting certainly don't give us any kind of tradable setup. I was just going to hang along for the ride and see where you were going with all this, but now I can anticipate a bit and mull it over.

I think this is an excellent idea, using the baseline to measure how much any given indicator/setup/whatever adds to an 'edge,' kind of like the E-ratio measure in Way of The Turtle.

Thanks for clarifying.

In all, I've really enjoyed this series of posts, and the comments too. Thanks.

Brandon W.

Brett Steenbarger, Ph.D. said...

Hi Brandon,

Thanks for the note. It struck me that you can't really know an edge unless you know what average odds (and variability around them) look like. I also appreciate the excellent comments to the posts--