So AQR, one of the largest quant funds bought 12 million shares of MU between June and September after a huge earnings beat, almost doubling its share count from 16 to 28 million shares.
AQR uses quant models that takes into consideration earnings momentum, but they didn’t have a model for the weak DRAM/Nand pricing selloff that started in June.
While DRAM prices peaked and dropped 25% in that three month time, AQR, probably not knowing this fact, picked up 12 million shares of MU in that three month based off uptrending stale analyst estimates. By the next earnings report came off, MU cited weak DRAM/nand pricing and revised down estimates, causing the stock to tank. MU dropped from the 50s to the mid 30s.
It’s probable AQR will now start selling MU because of negative earning trends. I guess the lesson from this is it’s not so simple to beat markets with a dummy approach because you will be taken advantage of with those who have in depth knowledge of company/industry trends.
This is a good example of the competition that’s out there. As quant-driven investors, we are actually in many cases less informed than your run-of-the-mill analyst. However, the flipside is that our naivete allows to be more objective than the fickle analyst who is more likely to be swayed by constant ebbs and flows.
I did a lot of work trying to disentangle oil and gas accounting using the information we have on P123. Although I eventually abandoned the project, it taught about the lag (and dissonance) between real world economics and accounting statements. Maybe it’s a pipe dream, but still I hope to one day to be become both better informed and more objective than the competition.
I wouldn’t be in a hurry to dance on the grave of Cliff Asness or AQR. He’s been around for a while and has proven humself many times over and need not hang his head before anyone. Anybody who has been in this field for more than a trivial period of time has his or her share of bad misses . . . I’ve had enough of my own to know better than to point fingers at anybody.
No investing style can ever be perfect and different kinds of investors are necessarily vulnerable to different kinds of misses. MU and DRAM prices, along with out-in-the-filed oil/gas trends, are examples of how folks like us can wind up very badly on the wrong side of a particular trade
But the folks who get these things right definitely have their own kinds of vulnerabilities. One is the nature of what they do. Counting cars in a shopping mall parking lot, tracking DRAM prices, counting active drilling rigs, etc. is all well and good but assuming you correctly extrapolate from what you observe in the present into the future (which will make or break stock performance) – which is no easier than extrapolating from what we use and there’s a difference between a one-shot anecdote and a consistent track record of success – they still have the problem of specialization. Even the best DRAM price forecaster in the world can carry you just so far because nobody (other than a specialty trader here and there) is going to build an entire investment portfolio around the timing of DRAM prices. Sure you could pay for specialized research in a whole bunch of fields, enough to support a full investment effort, but then, you may find whatever alpha you think you earned more than swallowed up by research costs. (Remember that in the real world of live money, any alpha above zero is terrific.)
Which approach is better? Ask 10 people and wind up with at least 50 opinions. But bear in mind that it’s the overall work product of the in-depth fundamentals crowd that serves as the foundation upon which we find tons of research “proving” that stock picking is ultimately useless and that the only thing to do it buy the market.
Maybe Asness, and the rest of us, aren’t so dumb after all – despite the reality that every now and then, we serve up our heads in ways that entertain professional finger pointers.
The only possible lesson here is in the area of risk management. Any style will have this type of thing happen. It is only the level of investment (28 million shares) that may have been a mistake. Even then, I am not sure of their total investment or what their attempts to hedge were. I would have to know this before I had any criticism.
Whatever style you use—even detailed fundamental analysis—what are your stats? How fat are your tails?
I am happy to share mine. I cannot get an R^2 above 0.0015 on anything I do with regard to predictions for individual stocks in the universe . That leaves 99.85% to chance. At the level of one stock which is what the discussion is about.
And fat tails? Are you kidding? We are talking about stocks.
Note: probably just me on the poor ability to predict. But it would be easy to see—with P123—how well the average analyst’s recommendation correlates with anything. Even if a financial wizard could do 10 times better………
I have been holding MU for a long time through highs and lows. As it is one of 30+ positions, I am not really concerned by single ticker gyrations. Even in the assumption that we can improve performance with discretionary research (which is not certain on the long term with a diversified model), the right questions are about time: what time would it take, what else can we do with this time and how do we value it.