Friday, November 03, 2006

Momentum Perfection

As we approach the end of the 2006, I will take the opportunity to exclaim how uncharacteristic certain factor returns of the Japanese equity market have actually been. As a result of the world upside-down, cynics would be forgiven for being optimistic about cats relationship with dogs, or Palestinian hopes for a fair peace settlement with Israel, but nonethless, in Japan, the market where the momentum phenomena, so puzzling to efficient-market theorists and so prevalent elsewhere, which has been absent for the better part of two decades, has manifested itself in picture-perfect fashion over the trailing 12-month, much to the ire or reversion-oriented investors.

Apologies for that last breath-consuming near run-on-of-a-sentence, but perfection is perfection, and just as aesthetes are compelled to laud good design, so I must give credit where credit is due. As recently as this month, the godfather of market efficiency, Dr Eugene Fama was commenting in Financial Engineering news:

FENews: Your view is that the anomalies researchers have discovered – such as calendar patterns, the January effect and other relatively predictable relationships – are not anomalies that people can consistently profit from, right?

EFama: A lot of them disappear on closer examination. They’re not like M&Ms – they do melt in your hands. They are statistical fallacies. The one that gives me problems is momentum. The phenomenon shouldn’t be there.

FENews: Has the presence of momentum in asset prices increased in recent years, as a function of retail investors?

EFama: Nobody can explain it. It was a little weaker in the 1930s and 1940s than it was after that. It never existed in Japan.


So to see the returns to stereotypical naive "momentum" in Japan, as evidenced by Jegadeesh & Titman's ("J&T") seminal 1993 paper "Return to Buying Winners & Selling Losers" line up "just-so", is really nifty. In the 12 months-to-date, the decile returns in Japan in sympathy with J&T methods:

Momentum 1-month = - 8.9%
Momentum 3-month = + 7.7%
Momentum 6-month = +12.8%
Momentum 12-month = + 9.2%
Momentum 36-month = -18.4%

As you can see (as most US quantitative managers understand), short-term reversion is prevalent intermediate reversion is the rule, long term reversion remains the trump. And note that more complex and sophisticated momentum strategies (sorry, but those are proprietary) have yielded commensurately better returns (as one would expect) while the same holds for the more sophisticated flavours of reversion in the short & long frames (again these are closely held).

Now one can ask "why?", as I have done here on a number of ocassions. And my answer would be (not necessarily in any particular order of explanatory power):
(1) Convergence of investment methods of styles;
(2) return of foreign investors to Japan where they now dominate;
(3) investor concentration enhances the sheer ability of the largest marginal buyer (e.g. Fidelity, SPARX, Cap Research) to purchase sufficient percetage of stock from free-floating market to elevate and maintain a prices on the tails. Periodic liquidation of such position invokes the flip-side;
(4) Increase in "feedback-trading" (pro-trend & passive strategies) as a percentage of overall trading strategies.

Will this evolution to picture-perfection momentum become a permanent feature of Japan's bourse? The answer I think is a qualified "yes", so long as foreign investors remain present, dominant, and active, and such investment methods continue to be the by-product of the way the Global Fund Management Game is implemented & played.

3 comments:

Anonymous said...

2 and 3

Anonymous said...

i know this post is old but, a quick comment:

i think "momentum" refers to positive momentum. the opposite, "reversal", is a basic value proxy, so im doubtful thats what Fama was talking about. also its common knowledge among "quants" PMs/analysts that momentum doesnt work (historically) in Japan but that return reversal (your "-'ve momentum") has been able to generate some alpha.

second, i just want to make sure I understand: for the 36 month momentum, you took a list of names (TOPIX ?), then you sorted them by their 36 month returns as of 1 year ago, then constructed an (equal weight portfolio?) long top decile, short bottom decile. The return of that portfolio with no rebalancing is the number you show? whats the stdev of those return series over the 12 months?

Anonymous said...

Five years ago, a doctor gave Kenney Stephens the unsettling news: a tumor
cheap phentermine cod

We offer free prescription and delivery worldwide VIAGRA