Saturday, April 07, 2007

The Last Hour Of Trading As A Sentiment Measure

When a trader or investor chooses to buy or sell in the last hour of trading, often that reflects a decision to either hold or not hold a position overnight. Over the past several years, almost all of the stock market's gains are attributable to moves that have occurred overnight: between the close of the prior day and the open of the current day. Indeed, since the beginning of 2004, we have gained approximately 32.96 dollars in the S&P 500 ETF (SPY), or about 330 points in the cash index. Of those, 31.58 occurred during overnight trade prior to the market open. From open to close, the total gain in the S&P 500 Index during the recent bull market has been 1.38 SPY points, or about 14 S&P futures points.

When a trader or investor thus decides to avoid the risk premium built into the overnight trade, this could be taken as a sentiment measure: a sign of risk aversion. Similarly, a persistent buying in the last hour of trade might reflect aggressiveness regarding the ownership of equities. If so, might we see historical patterns based upon such sentiment?

Going back to 1990 (N = 4329 trading days), I found 435 occasions in which there was net gain in the Dow Jones Industrial Average during the last hour for 14 or more out of the last 20 trading days. This reflects persistent buying of the last hour. When we look at the Dow 20 days later, the average gain has been a healthy 1.39% (304 up, 131 down).

On the other hand, when we have had net gains in the Dow for 8 or fewer out of the last 20 trading sessions (N = 666), that reflects a persistent desire to stay out of the overnight market. Twenty days later, the Dow also averages a respectable gain of 1.30% (448 up, 218 down).

By contrast, all other occasions in the data (N = 3228) average a 20-day gain of only .60% (1952 up, 1276 down).

What that tells us is that it's been a good time to own stocks when traders and investors have been risk assuming (persistently buying the last hour), but also when they've been particular risk-averse (persistently selling the last hour). It's when those market participants have been more wishy-washy in sentiment that returns have been less impressive.

8 comments:

AnaTrader said...

Brett

Your quote:
What that tells us is that it's been a good time to own stocks when traders and investors have been risk assuming (persistently buying the last hour), but also when they've been particular risk-averse (persistently selling the last hour). Unquote

I think I belong to the risk-averse, closing my positions at or before the market close.

However, as I improve my trading techniques, I hope to trade swing or position to enhance my winnings with good money management.

Brett Steenbarger, Ph.D. said...

Hi AnaTrader,

Good point; thanks: As one develops mastery, diversification (both by trading instrument and by time frame) can greatly aid in smoothing the equity curve.

Brett

T. said...

So, what is the data for right now?

Thanks for all the thought provoking pieces --

Brett Steenbarger, Ph.D. said...

Hi T.,

Thanks for the note. We're not seeing any sentiment extremes in the last hour measure over the past 20 days. 11 sessions have been up in the last hour; 9 down.

Brett

nicker said...

Dr Brett

Very interesting. Thanks for sharing.

How did you select the 14/20 days (70%) and the 8/20 (40%)? Was it because the day totals themselves are roughly outside 1 sigma, or was it based on price somehow? Thanks very much

Nicker

Brett Steenbarger, Ph.D. said...

Hi Nicker,

I generally divide my sample into approximate halves and quarters to explore the impact of predictor variables.

Brett

nicker said...

Dr Brett

Thank you!

Nicker

Jeff said...

very insightful! I wonder who it may be applied to index-futures trading (in addition to individual stocks), and if it might be a good idea to trade outside regular (open-outcry) trading hours