Thursday, November 17, 2005

The Fcuk Factor

A few years ago, after a number of customers suffered what one might call grave financial contusions, BARRA capitulated and included their first non-fundamental factor in an otherwise fundamental factor model. This factor was called "Momentum" (note the capital 'M'). Never mind that it couldn't be explained. Or that consternation doesn't stop at it's puzzling properties. Though they weren't consulted for comment regarding this post, I do ocassionally wonder at precisely how much sleep the financial engineers at BARRA lost over the issue. That's because it (like the left-right political divide) causes visceral emotions best exemplified by reports of mild-mannered tweed-donning academics nearly coming to fisticuffs about it. It causes otherwise rational people to anthropomorphize or even deify it in stark terms of good or evil, and virtue vs. vice. I even have some good friends and colleagues who to this day get completely spastic over the fact that it exists at all, even though it seems like it shouldn't.

In the short few short years since they've added Momentum to their model(s), the world has come a long long way. Microprocessor clock speeds are now measured in gigahertz. The price per megabyte of memory has long since put a lid on EMC's share price. Telecommunication costs are less than the price of a colourful gumball. Much financial data is now virtually free (though quality remains dubious), and "data-mining" has supplanted "strip-mining" as a concern amongst thoughtful graduate students. Even momentum, once thought of as inexplicable, has, with the help of graduate student slave-labour, neural nets and non-linear models, now been found to alias more plausible sounding things such as sector return, forms of growth (or is inverse), and various nuances of relevant first derivative-like variables.

Yet, even with all this power imbued in the cooked sand of the microchip, in combination with all the quantitative model-building prowess of the best and the brightest, some securities continue to defy our abilities to systematically explain why they are up (and won't go down), or why others are down (and won't go up). And so to insure that my PhD. hasn't been an utter waste of time, effort, paper, and federal & university endowment grant money, as well as to altruistically further the knowledge of academic finance, I would like to propose that BARRA introduce what would be their second non-econometric factor: "the Fcuk Factor". Once introduced, I am confident that this factor will systematically explain the previously inexplicable returns on tails of the cross-section that have been baffling rational investors and financial sleuths for years.

With momentum, BARRA set about to measure and fit an observed (technical?) factor into their otherwise fundamental framework of systematic risk descriptors. They could have just as easily used "Relative Strength" or MACD, but that is a hornets nest for another day. The Fcuk Factor WOULD have a-reasonable r-squared to Momentum. But not all stocks that load highly on momentum will have a high loading on the Fcuk Factor. But since their factors are not orthogonal, this is of little consequence anyway. For some momentum stocks should have momentum. They DESERVE it at the time of measurement by virtue of their attributes, irrespective of whether they will live up to them in the future. FOr THAT moment, they have earned the right to bask in whatever expectational glory is they call theirs. But the Fcuk Factor would describe otherwise inexplicable or difficult-to-explain returns on the tails that - to date - BARRA calls "residual", since they are not explained by the model. As such the returns of securities with let's call it "undeserved momentum" are believed to be idiosyncratic and therefore specific to that stock. But I would propose to you that "the Fcuk Factor" will SYSTEMATICALLY help to explain large anomalous recent (~12mo) returns amongst otherwise historically uncorrelated and econometrically unrelated securities. The individual pieces are already out there in the body of public domain research - visible for all to see. BARRA need only take their stockpot and mix the composite soup of final derivation based upon the inputs, including, but not limited to: short interest ratio, short interest as percentage of float, degree of separation of short interest from market price action, probability-adjusted tradeable float, concentration of institutional ownership, change in institutional ownership; clustering of institutional ownership by investor type; # of inappropriate funds in fund family holding the security; the # of phone calls between the Company CFO and leading analyst; divergence from most forecasted absolute & relative equilibrium prices on the majority of rational models; scale and persistence of window-dressing activity; anomalous microstructure activity (end of day marks & other non-random footprints); "tip-sheet" popularity; # of positive mentions by "Cramer" in last 3 months; membership in IBD 100; Large change in IBD's Proprietary ranking, aggregate option open interest, and anomalous expiry activity.

The most important contribution of accurate measurement of the Fcuk Factor will be a dramatic increase in the efficiency of markets as well as its numerous applications to the investment management indsutry in general - especially the largest of long-only managers (Fidelity, Cap Research, etc.) and incentivized & well capitalized equity hedge funds. It will, for example, allow the largest marginal buyers to justify why many of their largest relative "active" positions in their portfolios are in some of the most implausibly overvalued crap & shite in the universe. I can already picture a large Boston firm's compliance officer's conversation: "Well you see Mr Spitzer, we were short the Fcuk Factor according to BARRA, so we purchased a million shares at the close of the quarter to insure our portfolios were balanced...". Uh huh. And it will be a bonanza for hedge funds who no longer will have to fish for justifications for ramping shares: "Dear Investors, Q3 returns were robust as we were correctly and aggressively bullish on the Fcuk Factor....". But the real boon will be to short-sellers, who can dramatically improve their returns by recognizing in advance as to which stocks to short and when (the answer to the latter, for those with elevated "Fcuk Scores", categorically being NEVER!). And with fewer shorts "getting in the way" of the largest investors having their way with their fav stocks, market efficiency will be much improved since the equilibrium price for such a security will be reached far quicker than might be the case without the fitting and estimation of a Fcuk factor into BARRA.

So I urge you BARRA, implore you BARRA to get to work beginning yesterday and sort it out. Measure it. Tweak it. Smooth it. Include it. And in so doing save hapless shorts from further one-way tickets to hell.


Anonymous said...

Real nice! Keep up the great work. Good work. Enjoyed the visit!
- g
spaghetti alla carbonara

Anonymous said...

Nice site man! Very nicely done. I will be back!
- 0
spaghetti alla carbonara