I have long suspected that the momentum anomaly is driven by earnings momentum. Chordia and Shivakumar (2005) have studied this extensively:
This paper narrows the search for an explanation to the anomalies by documenting that the price momentum anomaly is a manifestation of the earnings momentum anomaly.
The finding that price momentum is subsumed by a common factor related to the macroeconomy is significant since it does not rely on capital market frictions to explain the price momentum effect.
i.e. market inefficiency is not required to explain the momentum anomaly!
The paper uses a standardized unexpected earnings (SUE) metric: standard deviation of 12 month earnings change:
SUE for month t = (Eit- Eit-4)/sit , where Eit is the most recently announced earnings and sit is the standard deviation of (Eit-Eit-4) over the prior 8 quarters.
Decile SUE portfolios are formed (PMN = P10-P1):
from January 1972 through December 1999, the monthly holding period returns increase monotonically from 0.79% for the lowest SUE portfolio, P1, to 1.68% for the highest SUE portfolio, P10.
For comparison, the price momentum returns are similar:
Over the entire sample period from January 1972 through December 1999, the monthly holding period returns increase from 0.84% for the lowest past-return portfolio, P1, to 1.60% for the highest past-return portfolio, P10.
However, price momentum (WML) has a much larger deviation:
the monthly standard deviation of returns of PMN is 2.21% while that for WML is 5.62%. These results are consistent with WML being a noisy measure of PMN.
PWM is also successful at predicting 12 month ahead economic activity, for example Industrial Production Growth (IPG):
Note the negligible coefficients for the Fama-French size and value factors. See the paper for more details.
I have edited the previous post on DWAS and PDP top holdings for accuracy, based on a very illuminating email.
The Dorsey Wright momentum ETFs are value weighted, using a proprietary ranking metric.
Relative component prices are shown below. Positions are rebalanced quarterly.
DWAS (small-cap momentum)
(click to enlarge)
Top ten holdings comprise 15% of the fund, ranging from 1% to 2% of the fund’s size. Eight of these stocks appreciated over 50% in the last year, showing strong price momentum. EXAS and PIR are showing a small annual loss; it will be interesting to see whether they appear in the top ten after the next rebalancing.
PDP (large-cap momentum)
(click to enlarge)
The large cap fund has relatively larger positions, up to 3%, and the top ten cover a quarter of the fund.
Nine of the top ten stocks display a pattern of strong 12 month price momentum (>30%). The obvious exception is AAPL which lost a third of its value in a year. Clearly, price momentum is not the only ranking metric utilized in the fund.
Interestingly, AAPL is the strongest performer this quarter amongst the top ten stocks. That the ranking metric is able to select this recent winner is impressive.
Continuing this series on the Dorsey Wright momentum (or relative strength) ETFs DWAS and PDP:
These funds are designed around the momentum anomaly discussed extensively on this blog. Fama-French data shows a frictionless annual return greater than 20% for small-cap high-momentum US stocks.
However, there is a another anomaly named “long-term reversal” where poor performers over the past few years (excluding current year) outperform:
Next month performance is in inverse order of past 5 year return, excluding current year (i.e. price 13 months ago minus price 60 months ago).
Chen and Kadan showed that double-sorting on both anomalies leads to significantly increased returns:
(MQ = momentum quintile, RQ = LT reversal quintile, data covers 1926 to 2006)
Update: note that Chen et. al. use a different measure than Fama-French: price 13 months ago minus price 24 months ago.
Simply by excluding the upper half of long term outperformers, monthly returns are increased significantly. Monthly excess returns would rise to 1.30% from 1.15% (i.e. increase by 2 percentage points annually). This filter would be worthy of consideration in the Dorsey-Wright funds.
Essentially, the screen would find stocks with a price history similar to the illustration below, where time zero corresponds to the buy point. A fund manager could then filter that list by other criteria such as earnings or global macro drivers to satisfy the mandate of the fund.
An earlier post analyzed DWAS, the recently introduced small-cap momentum ETF from Dorsey-Wright. This post looks at PDP, an all-cap momentum fund with a longer history (> 6 years).
Comparing daily returns:
Frictionless Fama-French data has 98% of PDP returns, on average! i.e. PDP performs 2% better (R2 = 0.92, correlation = 0.96).
Performance since launch versus SPY and Fama-French momentum* is shown below.
* average of small and big “high momentum” series from the Ken French data library.
The theoretical series (red) corrected less in 2008 and therefore ends higher. However, the return from the 2009 low is identical between PDP and FF (see table). PDP sharpe and average daily return are about 50% higher than SPY over the last 6 years.
For comparison, daily returns versus SPY are below:
The gradient of the trendline is 0.85, therefore PDP daily returns are 15% greater than SPY, on average since 2007.
Four decades after Robert Levy‘s paper, past winners still outperform past losers. S&P 600 stocks rising over 20% in a year return 10% in the next quarter, on average. (This analysis covers the last 5 years and has some survivor bias due to index changes, unlike the Fama-French data showing similar results.)
More important than the magnitude is the distribution. Clearly, annual return is proportional to the next quarter return. Asking Steve Cohen for tips is unnecessary. Fama-French data confirms this effect over decades. My opinion is that the effect is due to accelerating earnings driving prices and is therefore built in to market structure.
An earlier post discussed Fama-French momentum, in particular small-cap with annual returns above 20%.
By adding the simple rule of holding cash when the previous monthly portfolio return is negative, remarkable improvements are realized:
Annual return is little changed but volatility is reduced and Sharpe is increased from 1.1 to 1.5 for small-cap momentum (upper green trace).
The rolling annual performance shows that down years are almost eliminated (red) but it is still possible to trail the S&P 500 in some periods (blue):