Wednesday, March 22, 2006

What's Happening on the Larger Time Frame: Does It Matter?

A reader recently made the valuable point that, in addition to looking at such key aspects of the market as volatility, momentum, trend, and sentiment, it is necessary to look across different time frames. That raises an interesting question: Do time frames matter? If we see a historical pattern on one time frame, does what's happening on the larger time frame make a difference?

To address this issue, I looked at yesterday's market, in which SPY was down -.63%. Going back to March, 2003 (N = 765), I found 95 occurrences in which SPY was down between half a percent and a full percent. The next two days in SPY averaged a gain of .20% (55 up, 40 down), stronger than the average two-day gain in the SPY sample of .12%.

I then divided the same down day sample of SPY into two categories based on time frame performance. One group was down between half and a full percent and was making a five-day closing low (just like yesterday; N = 46). The other group was down by the same amount but not making a five-day closing low (N = 49).

When the down day in SPY was making a five-day low, the next two days in SPY averaged a gain of .40% (29 up, 17 down). When the down day in SPY was not making a five-day low, the next two days in SPY averaged a flat performance (26 up, 23 down). Thus, the bullish implications of a down SPY day are entirely attributable to the fact that they're five-day lows. The larger time frame matters quite a bit.

This is a moderately bullish consideration for today, especially if early action fails to take the averages below yesterday's lows. Hats off to the reader for an excellent observation.


John Wheatcroft said...

On a single-stock basis how is sentiment measured? I always felt that the sentiment towards any given issue is implied in the price and reflected in the volatility.

Brett Steenbarger, Ph.D. said...


Volatility is different from sentiment, in that you can have volatile bullish or bearish periods. Sentiment can be assessed through put/call ratios, of course, but also by the ratio of volume traded at the bid vs. at the offer. I use the Market Delta program for the latter measurement. There are also ways of creating TICK measures for individual stocks just like the NYSE TICK. NeoTicker is an excellent program for that purpose.