Comments on: God and RenTech’s black box

By: slocklin

slocklin — Tue, 10 Jul 2012 11:12:56 +0000

I dunno if you read old comments, but Rich Jim and company used to be codebreakers. Yeah, you could call weird stuff a “potentially spurious correlation that makes no sense.” Or you could know something about information theory and code breaking, and know that your spurious correlation is statistically certainly telling you something about some underlying market mechanism you don’t quite understand. That’s more or less my point about their trading weird patterns. They probably don’t have economic models for what is happening, but I guarantee they have sound probabilistic models.

Not that I’d know anything about this and stuff.

By: JimmyJimmington

JimmyJimmington — Sat, 26 Jun 2010 23:31:22 +0000

Isn’t this what the Fama-French three factor model does? It generates what looks like alpha according to the CAPM for no good reason. I’m not sure what the point of this post is. How can we discuss their models or their trades if we don’t even know what they are?

By: hsvkitty

hsvkitty — Fri, 25 Jun 2010 20:01:53 +0000

I slept on this and I am still back to the original thought i had, about the man who had made a perpetual motion machine and everyone was astounded.

That is might not be real meant nothing to the throngs of educated, uneducated, scientist and engineer who gawked in awe as it moved seemingly to infinity.

The fact is it had a little winder inside that an insider secretly wound up every night with a special key

The perpetual money machine. Eureka!

By: dWj

dWj — Fri, 25 Jun 2010 18:48:37 +0000

I’m reminded of neuroscience evidence that people tend to decide what they’re going to do subconsciously and only use the conscious mind to rationalize the decision after the fact. (Much like financial journalism.) Some things are just harder to pretend we understand than others.

By: HBC

HBC — Fri, 25 Jun 2010 16:57:23 +0000

Renaissance, formerly synonymous with the resurgence of learning as a means to escape the gravitational pull of theocratic darkness… From that to this:

“We’re not sure and frankly don’t care whether money revolves around the sun god, or vice versa. Just pay The Man, and don’t forget to tell him the Oracle sent you!”

By: chaetodon

chaetodon — Fri, 25 Jun 2010 14:51:34 +0000

I’m a bit saddened by this one Felix.

Agathocles is right on point. Typical spewing from quants.

By: loopguy

loopguy — Fri, 25 Jun 2010 13:53:02 +0000

Now Felix, let’s not get metaphysical.

It could just be the old shell game: shape the return curve so that returns increase, volatility decreases and third moments & higher all increase. Setting yourself up for a larger disaster, sometime later.

By: Rikh

Rikh — Fri, 25 Jun 2010 04:34:59 +0000

1. Isnot there asimple explanation. You can start with something in 2 ways:
-start from the theorie
-start with statistical info.
2. Subsequently you assure yourself by testing the theory in practice resp. find a logical explanation for the statistical findings.
3.The latter you can of course also do by new (future) testing if it shows that it works not only in the past, there is a high probability that there is a correlation.
4.Of course spreading your risk if you donot have a logical explanation yet reduces the overall risk even more. Limit your exposure on one bet and keep looking for the logic.

By: agathocles

agathocles — Fri, 25 Jun 2010 04:15:09 +0000

“Some signals that make no intuitive sense do indeed work…the signals that we have been trading without interruption for fifteen years make no sense. Otherwise someone else would have found them.”

The speaker’s simply contributing to the aura that surrounds quantitative finance. Quants love to portray their work in the imagery of alchemists divining the secret laws of the heavens and the world. Unfortunately, many journalists are more than happy to oblige (cue Scott Patterson of the WSJ).

The reality is that the inference techniques used by quant funds are not too far from those used by academics in the natural and social sciences, bioinformaticians at Pfizer, engineers at Boeing, market researchers at Nielsen, or search engine designers at Google.

Also, the whole idea of using factors and signals that are non-intuitive happens all the time across many fields. Consider principal components analysis (PCA). Would you consider the factors it outputs to be non-intuitive? Yes, it lacks the robustness of models that progress from an intuitive theoretical model, but I would consider it a fairly standard, understandable, and robust (under specific cases) technique for statistical inference.

Lastly, I would recommend that you rarely believe quants when they boast that their signals are unique. The original ideas are sometimes brilliant, but a large majority of signals I’ve come across have been variations on a theme. Often, what makes a difference between a profitable signal and an unprofitable signal is not the actual signal itself but rather the technology with which you can implement it, the operational sophistication of the firm, the coding ability of its IT developers, and the culture of the team within which they work.

By: jiminy_c

jiminy_c — Fri, 25 Jun 2010 02:59:12 +0000

Reminds me of this Radiolab segment about computers that spit out algorithms that explain relationships, even though we can’t understand/intuit why the relationship exists:
http://audio.wnyc.org/radiolab/radiolab0 41610c.mp3
I <3 Hume