Friday, February 27, 2009

Combating Recency Effects in Trading

This morning's trade offered a nice example of how an understanding of historical market patterns can help put a brake on cognitive biases that lead to poor trading decisions.

We moved sharply lower during pre-opening trading hours, breaking below the multi-day trading range. I noted in my morning Twitter comments (free subscription via RSS) that an important issue for the day was how we traded around the market's opening price.

The recency effect is a cognitive bias in which we overweight the most recent events and allow them to unduly color our future expectations. Understandably, traders caught up in the pre-opening weakness might anticipate further downside market action early in the day, particularly given the downside breakout.

At such times, I like to check my biases against actual market history. Since 2007 in the S&P 500 Index (SPY), for example, the correlation between overnight market moves and moves during the day session is only .06. In other words, these function as largely independent trading periods. Indeed, we've had 59 occurrences of SPY opening more than 1% lower than its previous day's close; the market's subsequent day session was up 27 times, down 32 times, for an average loss of -.21%. That compares with an average loss of -.06% (236 up, 247 down) for the remainder of the sample.

So, yes, particularly during a period of lower market prices, there can be downside follow through to overnight weakness. This is hardly a significant edge, however. Knowing that a different category of traders participates in the day session vs. the overnight helps us avoid the recency bias of assuming that overnight moves will continue into the day.



Val said...

Thank you for posting data every morning even when you work with traders. Have a nice weekend, Val

Trader M said...

Wish I had read this before I went short this AM at the 740ish level. Would have saved myself a $2700 dollar hole I spent the entire rest of the day digging myself out of to break even by 4PM. :-)

More seriously Brett, would you recommend that we ignore the overnight and premarket futures, even the immediate hour or so before open? I was viewing the price action this morning before the bell as a distinct and legitimate violation of the November intraday lows and final last-ditch support, and thought it likely that the market would revisit those prices relatively soon. We did, of course, come close to those levels, right at the end of the trading session. And I did use the overnight action to suggest that the rally today we saw through the morning would be weak and have great difficulty defeating the former support / new resistance at 750.

I'm a new trader and I'm not entirely sure how much attention to pay to the overnight action in the ES contract. The one thing I have found to be true over the past few months, most of the time, is that if the biggest extent of the gap occurs right at the open, and the gap is widening rapidly into the open, the gap is likely to close pretty quickly. And conversely, if the gap is smallest at the open, it is more likely to "gap and go" than usual, and less likely to close. But then again my gap trading is quite mediocre to date.

Stick Hansup said...

I too, wish I'd read this article before today's action.

Thanks for writing your new book by the way. Looking at the sample chapter right now.

Brett Steenbarger, Ph.D. said...

Hi Trader M,

I think overnight action and action from the previous day are helpful in framing trade ideas, but it's perilous to use them to make absolute predictions--