Comments on: Deep makes more money than Wide

By: Zoe Lindesay

Zoe Lindesay — Tue, 10 Aug 2010 09:54:37 +0000

As a starting place for an alternative, what about:

IR = sqrt IC * sqrt N

By: Zoe Lindesay

Zoe Lindesay — Tue, 10 Aug 2010 09:18:51 +0000

A number of points:
1. What is ‘IR’ ? I am assuming it is returns above the average market return of a passive fund invested in a portfolio replicating a benchmark index?
2. The equation ignores the impact of other participants. What happens if the majority of other market participants are trading off another set of (false) premises? (i.e. Can you be too clever / have too much information) ?
3. What is this equation trying to say about ‘N’?
I seem to have a choice between:
£100 * IC of 1 * sqrt 1 = £100 (Given trading opportunity A)
or [£50 * IC of 1* sqrt 2] *2 = £141 (Given trading opportunity B and C)
Therefore increasing the number of trades always increases my returns.
But this assumes my ‘IC’ is equal for A, B and C and remains constant regardless of how many trades I carry out. If the IC differs between trading opportunity, then couldn’t the relationship between increasing ‘N’ and increasing returns break down very easily:
£100 * IC of 2 * sqrt 1 = £200
[ £50* IC of 1 * sqrt 2] * 2 = £141
Substituting, ‘research effort’ for ‘IC’: Does this imply if I spent twice as much effort researching one stock, I could significantly increase my above market returns compared to splitting my effort equally between two stocks?

By: Andrew Nelson

Andrew Nelson — Sun, 16 Mar 2008 17:42:44 +0000

Charlie Munger, ““A Lesson on Elementary, Worldly Wisdom As It Relates To Investment Management & Business”. Way back in 1994.

By: Baruch

Baruch — Fri, 14 Mar 2008 07:25:50 +0000

B, if the point is that our ability to understand complexity is not great, surely out schmetas will be low. Can we square root those schmetas? We may also need to introduce a term H ( or hubris)for our PERCEIVED ability to understand complexity, where H does not equal Schmeta. N rises in proportion to H, as we feel we are jolly clever and can understand more things. But of course, this adds to N which at higher levels reduces our IC! Throw that in somewhere, pet mathmo of the Spinozists, and I think we may be ready to publish.

Felix, interesting you bring up Soros, who is a prime inspiration of the reduction of complexity point. He would work intensively to reduce his N. His m.o. was always to isolate a strategic issue out of the many available, and then throw as much resource as possible at it, really concentrating on it, to see if it was worth playing one way or another. I don’t know how many themes he would play at one time, but I would be surprised if it was very many.

My complexity point is also very informed by Popper and Hayek, and as you know, Soros and Popper in a tree.

But in general, you are right: we don’t hear so much about the big macro funds any more. So long as they have epistemologically sound approaches, I see no reason why they shouldn’t prosper.

By: Felix

Felix — Fri, 14 Mar 2008 01:01:01 +0000

Does this mean that all Global Macro funds are bollocks, and that George Soros never made money?

By: B

Fri, 14 Mar 2008 00:16:34 +0000

The hope is that IC is dependent on your ability to focus and determine detail, where schmeta is our ability to understand complexity. IC is our inner accountant (grey and nameless), while schmeta is our inner steve jobs.

The far more intriguing question is how to measure all of this. To do so, I think we need luck An appropriate model might thus be
IR=IC*N^(-schmeta)*1/N*L^(1+schmeta).

As we all know, luck is always amplified by schmeta. Better yet, I have a 4 component model to fit lines, which makes me feel that even I can manage this.

By: Baruch

Baruch — Thu, 13 Mar 2008 22:05:50 +0000

Indeed, B, the impact of diversification, which, as you say degrades returns at higher levels of N, would seem to cock it all up. We would need to have a term, say aN, where a is the level of N after which it all went pear-shaped. Otherwise I liked your version of the equation very much indeed, and had no idea what it was on about by the time I got to the end of it.

Wouldn’t IC be partially dependent on Schmeta as well, however? Probably correlated, if not causal. If I can handle lots of complexity I may also have a large brain and therefore more IC than the average bear, therefore might there not be 2 schmetas in the equation were we to break down IC definitionally?

By: B

Thu, 13 Mar 2008 21:42:42 +0000

Isn’t risk proportional to 1/N, and that is what makes this equation mistaken? The more stocks, the lower the risk., but the lower the potential reward

There ought to be a ‘schmeta’ which measures your ability to understand complexity–so that the equation looks something like IR=IC*((N)^-schmeta)*(1/N). If you have a large schmeta, you can deal with complex multi-part problems well. If you have a really high schmeta (schmeta>.5). If you have a schmeta of <.5, wider spread always causes you to do poorly.

If we add enough variables, we can easily fit all possible opportunities perfectly. Maybe schmalpha is schmeta-(market-average-schmeta).

I am, however, totally uniformed in the monetary arts, and merely here for the philosophy.

regards