Mauboussin, and my assessment of the portfolio theory work ahead
Mauboussin does a great job at summarizing the two schools and portfolio theories that I've read about so far:
http://www.capatcolumbia.com/MM LMCM reports/Size Matters.pdf
1) kelly (and thorp and others), which he calls also "maximizing the geometric mean"
2) markowitz (and samuelson and others), which he calls also "maximizing the arithmetic mean" and "mean/variance efficiency"
Based on information theory, the Kelly Criterion says an investor should choose the investment(s)
with the highest geometric mean return. This strategy is distinct from those based on
mean/variance efficiency. Importantly, however, you can calculate geometric mean using the
same arithmetic mean and variance from mean/variance models.
Furthermore, the author says that even among the CAPM theorists there's disagreement and that Markowitz himself did not disagree with the "geometric mean approach" (I don't even know what it is precisely):
There are two other problems with utility theory and investing. The first comes from the father of
mean/variance analysis, Harry Markowitz. In his famous Portfolio Selection, Markowitz advocates
the geometric mean maximization approach. In spite of arguments by Jan Mossin (one of the
founders of the capital asset pricing model) and Samuelson in the 1960s, Markowitz reconfirmed
his endorsement of the geometric mean maximization strategy in the preface to his second
edition published in 1970.
So, ok: kelly and thorp and poundstone all agree on pretty much everything, and say kelly is better than markowitz. But the academics and the CAPM disagree (more so for everyone who came after markowitz), ignore or look down on kelly supporters, and have affected the fund managers greatly and in a negative way, as Mauboussin says, but not to the extent of what Scott Vincent says (cfr.
this post).
Mauboussin is both an academic and a fund manager, so he knows theory but doesn't waste time on bull****, and although he has read everything there was to read, he sums it all up, practically, for me. Still not enough for me, though. But it's best thing I read so far in terms of conciseness.
While reading Mauboussin's paper, I redefined the issues at stake as far as I'm concerned.
1) Diversification is good, but adding more systems does not necessarily produce diversification. By adding correlated systems I may actually be
decreasing the diversification of my portfolio. That is why portfolio theory has become so crucial.
2) Now, just as diversification might actually not diversify but
concentrate, also the study of correlation to avoid faulty diversification, might not be enough and be misleading, because past correlation (unless we're talking about YM - ES - NQ, and ZN - GBL, and currencies, and similar other correlated futures) does not imply future correlation, and viceversa.
3) It is better to build a portfolio through good and proven empirical/practical stastistical work (including resampling), like I've been doing so far, than using a faulty theoretical approach - which is all I could afford right now if I went the other way. On the other hand, a theoretical understanding of the issues is my ultimate goal, because it will allow a faster and more coherent building of my portfolio - but, as i said, I can't do it yet. Until then, I'll be sticking to my present empirical approach. Until then, I'll keep on studying the ingredients of portfolio theory. Until then, however, I'll keep on using portfolio
practice.
...
Speaking of learning the ingredients, this is the kitchen I need to hang around:
4. Portfolio Diversification and Supporting Financial Institutions (CAPM Model) - YouTube
I am still stuck with the summation notation. It has the power of driving me crazy and making simple things appear complex to me. I'll only be able to digest 10 seconds of this class per day. But there's no way around it: all the portfolio theory books are full of such formulas.
[...]
Trying to find out if the sharpe ratio is enough to do the efficient frontier calculations, in that they both use profit over variability, and since I'm already acquainted with the sharpe ratio, it would save me a lot of work.
For a few minutes I was looking for "efficient frontier" calculators and excel sheets, but they're all pretty crappy and poorly made, and even the best one is crappy, the one in Solver:
Example: Portfolio Optimization - Efficient Frontier Markowitz Method - Frontline Systems
Besides, I'd much rather understand things and then implement them on my own, so I moved on to another search:
sharpe ratio vs efficient frontier
If these two are the same thing, then I can handle the sharpe ratio. Been doing it for a year, with the investors.
I need to squeeze my last few resources left, and I can't afford to waste time doing unnecessary work.
[...]
Good link with an efficient frontier calculator, but nothing for futures:
Efficient Frontier
Still, an awesome job they did.
Inno dei Sommergibili - Italian Submarines in WWII -Tribute- - YouTube
[...]
Ultimately the concepts I have to focus on are those expressed by these three people:
1) markowitz
2) kelly
3) sharpe
If I can find a way to fit together their three slightly different recipes, or to create a recipe of my own that uses their main ideas, then I am set. The problem to tackle gets clearer and clearer...
I just wish there were an intermediate step for me, something like "portfolio theory for dummies". But there isn't. I feel like a child who tries to climb a set of stairs with high steps. I get tired. And I feel like quitting. It's like learning a foreign language: you can't learn Arabic overnight. It doesn't matter if you immerse yourself in their culture. This is not as hard as learning Arabic for me, but it renders the idea.
What would be ideal is if I could find another 20 papers like the one Mauboussin wrote.
