On Catalysts and Quant Trading
-The resolution of an investment catalyst should be a carefully monitored function of the time horizon of the investment. You may have a high resolution catalyst or set of catalysts on stocks with a one year time horizon, but when the horizon drops to six months, extremely specific catalysts become all the more important because the time dependency is much higher.
-There may be hope yet for deep value investing with a quantitative overlay. Even though quant thrives in high resolution data over a large universe of liquid stocks, that doesn't mean it can't help you infer a first guess, comparables style, on how much a company should be worth. Think of volatility surfaces-- option writers don't necessarily have super specific valuation models to determine in a vacuum exactly what an option should be priced at in implied vol terms. Rather, they interpolate out the vol surface as a multi-dimensional function and see where the new options fit on that surface-- a glorified interpolation.
Quant, then, can essentially be applied as a more in-depth screening technique which may point you, a person with scarce resources and time, to stocks with the highest probability of being potentially interesting (just remember the market maker and his 'eval': this is more or less the same thing, except we are working for more than a few basis points). The quant gives you the scale benefits that you couldn't have possibly detected as a human, because it's doing calculations based on an arbitrary number of variables over an arbitrarily large number of stocks. And yet at the same time, you get the qualitative ability to wade through and infer based on all the idiosyncratic information which makes each stock so unique.
Deep value, then, becomes a multi-tiered analysis where the resolution of the different phases of the analysis are proper functions of (1) the number of companies being looked at and (2) the nature of the process being observed and the time interval being examined. That makes sense.
So then the question becomes what model to create...
-There may be hope yet for deep value investing with a quantitative overlay. Even though quant thrives in high resolution data over a large universe of liquid stocks, that doesn't mean it can't help you infer a first guess, comparables style, on how much a company should be worth. Think of volatility surfaces-- option writers don't necessarily have super specific valuation models to determine in a vacuum exactly what an option should be priced at in implied vol terms. Rather, they interpolate out the vol surface as a multi-dimensional function and see where the new options fit on that surface-- a glorified interpolation.
Quant, then, can essentially be applied as a more in-depth screening technique which may point you, a person with scarce resources and time, to stocks with the highest probability of being potentially interesting (just remember the market maker and his 'eval': this is more or less the same thing, except we are working for more than a few basis points). The quant gives you the scale benefits that you couldn't have possibly detected as a human, because it's doing calculations based on an arbitrary number of variables over an arbitrarily large number of stocks. And yet at the same time, you get the qualitative ability to wade through and infer based on all the idiosyncratic information which makes each stock so unique.
Deep value, then, becomes a multi-tiered analysis where the resolution of the different phases of the analysis are proper functions of (1) the number of companies being looked at and (2) the nature of the process being observed and the time interval being examined. That makes sense.
So then the question becomes what model to create...
posted by Dan McCarthy at
6:47 AM
0 Comments:
Post a Comment
<< Home