In my last post, I took a look at herding as a form of sentiment and found a characteristic pattern at market bottoms. That post used a measure of volume concentration to detect when traders were leaning significantly to one side of the market (advancing stocks, buying) or the other (declining stocks, selling). This post will take the same volume concentration measure and cutoff points to detect herding to see what has happened historically following days of significant herding. (Please consult the previous post for details on the Volume Concentration Index and its cutoff points).
Going back to 1965 (10,715 trading days), we find 159 days of significant herding into rising stocks. Three days later, the S&P 500 Index (SPY) averaged a gain of .62% (105 up, 54 down), much stronger than the average three-day gain for the remainder of the sample (.09%; 5699 up, 4857 down). Indeed, when we go out twenty days, we find that the average gain in SPY following a single upside herding day has been 1.86% (109 up, 50 down), far stronger than the average 20-day gain of .61% for the rest of the sample (6222 up, 4334 down).
In short, a single day in which investors significantly herd into rising stocks has tended to bring bullish follow through in the near term. That pattern has held during the period of 2004-present, particularly at that 20 day horizon. Of the 20 days of bullish herding since 2004, 16 have shown positive returns over the next 20 days, averaging a healthy gain of 1.55% in SPY.
Now let's turn to herding days in which volume is concentrated in the shares of declining issues. Going back to 1965, we find 220 days of such bearish herding. The following day, the S&P 500 Index (SPY) averages a loss of -.19% (93 up, 127 down), weaker than the average single-day gain of .04% (5500 up, 4995 down) for the remainder of the sample. Interestingly, as we go to a 20-day horizon, we find no significant bullish or bearish implications of downside herding days. The bearish follow through to a single day of concentrated volume to the downside has been limited to the very short term.
Even this conclusion must be qualified, however. Since 2004, single downside herding days (N = 28) have had significant bullish prospects going forward, including the very short term. Twenty days after a downside herding day, SPY has averaged a gain of 1.60% (22 up, 6 down), much stronger than the average 20 day gain for the remainder of the sample.
One implication of the above finding that bears further investigation is that bull markets treat downside herding days differently than bear markets. Indeed, that might help to make bull markets bull markets (panic selling leads to buying among value-oriented, longer time frame participants) and bear markets bear markets (panic selling feeds on itself). Examining the trajectory of market outcomes following herding days might thus yield insights into the longer-term trend and sentiment of the market. This is particularly the case when we examine sequences of upside and downside herding days (i.e., upside herding day following a downside herding day and vice versa; successive upside or downside herding days).
This is a fruitful area for research, as I'll be elaborating in my next Trader Performance post. In my next blog entry, we'll look at volume concentration as a kind of "overbought-oversold" indicator to see if it yields any historical insights about prospective market moves.
Making Decisions From Current Stock Market Data - Relevant to real-time monitoring of volume concentration.
Trading With Sentiment Bars - Relevant to very short-term market patterns involving volume concentration.