Thursday, November 30, 2006

U-Turns In The Market: What Comes Next?

We've had an unusual situation in the stock market, in which we had a high momentum decline on Monday followed by a two-day recovery that erased that decline on strong positive momentum. The natural question to ask is: What has happened historically following such market U-turns?

Before I address this question, however, allow me to step back and reflect upon how to utilize these historical findings. For me, they are background: they tell you if there has been any directional edge to the market under current conditions and how large that edge has been. If I see a decent skew to the data, that forms a hypothesis in my mind that will help shape my thinking about the market 1-5 days out.

A hypothesis, however, is not a conclusion. A hypothesis is an idea that needs to be tested out. In trading, what tests the hypothesis is the actual supply and demand that occur in the market under real time conditions: how much volume is being done, at which prices, at the market bid or offer, and by whom. If my hypothesis shows a bullish bias for the next day of trading *and* I see large traders lifting offers early in the AM, I will wait for a pullback and enter in the direction of market sentiment--and the direction of my hypothesis.

So a historical pattern is really a heads up, telling you: "Be on the lookout for this". After we had the large market drop on Monday, my research said: "Be on the lookout for a rise". Heeding that heads up when you saw buyers become aggressive paid off. But that is different from treating the hypothesis as a conclusion and using the pattern as a mechanical trading signal.

OK, so we had our decline and then our recovery: what happens next? Since 2004 (N = 729), we've had 24 days in which the market recorded strong downside momentum (Supply > 1000) two days ago and strong upside momentum today (Demand > 1000), using the indicators from the Trading Psychology Weblog.

Five days later in the S&P 500 Index (SPY), the market was up 11 times, down 13, for an average loss of -.17%. That is weaker than the average five-day gain of .16% (415 up, 314 down). It appears that, overall, U-turns tend to turn all over again within a five day period.

Interestingly, however, when the U-turn has occurred during a two-day period of positive sentiment in the broad market (NYSE TICK) and in the large caps (Institutional Composite), all five of those occasions were up the next day and four of five were up an average of .98% three days later.

While five days is a very small sample, the hypothesis I'm left with is that the willingness of traders to lift offers as we approach the bull market highs will be an important factor in whether or not we sustain an upside breakout. If we see sector divergences and waning lifting of offers as we test new highs, I'll be much more likely to look for yet another U-turn than if participation in the rise is broad, with sustained volume at the offer.


Benjamin Boyle said...

G'day Brett,

I'm glad to have found your blog. It's been wonderful to learn from you. Love the psychology, and love to recognize my own psychological hangups as my trading matures.

Would you mind posting an article about your unusual-looking charts (and how to get them) and how to get the TICK?

I have a $TICK on my ThinkOrSwim charts, but I'm not sure exactly what it's measuring the ticks on.

Keep posting Brett! It's great :-)

Cheers / Ben

Brett Steenbarger, Ph.D. said...

Hi Ben,

Thanks much for your kind comments.

The symbol $TICK should represent the NYSE TICK, which tracks all NYSE issues. The $TIKI (sometimes rendered as $TICKI) tracks the Dow issues specifically. There is a platform called NeoTicker that enables you to create TICK measures for specific baskets of stocks, such as the S&P 500 issues.

The unusual chart you're probably referring to comes from Market Delta (, which is one of the tools I use in my own trading. There is a good amount of material on their site re: how to use the charts.

I appreciate your comments; do stay in touch--


cpptrader said...

Hello Dr. Steenbarger,

Thanks for the further insights into your methods. I enjoy your unique and time-sensitive tests. I was curious - do you ever continue on the results with significance tests or confidence intervals? I know with some of your results the sample size is too small to provide any reliable statistical significance, but that is not always the case.

Also, just recieved your book via Amazon, and looking forward to delving in.



Brett Steenbarger, Ph.D. said...

Hi Matthew,

Great question. I view the blog as a kind of sketch pad, where I lay out initial trading ideas and hypotheses.

If I want to develop a sketch into a full blown work of art, that's when I do the full array of backtesting and significance testing. My favored platform for that is the Salford System suite of modeling programs (

I tend, however, to not be a systems trader, so that isn't my primary focus on the blog.