Taleeb rocks

[peterbirks]

When I first read Taleeb's "Fooled By Randomness" I did not give it the credit it deserved. I felt that we were looking at a player who made his money betting on unlikely events (in his case, taking cheap positions deeply out of the money and waiting for one in a 100 of them to pay off big time) but who did not really "get" the fact that this principle only worked provided only a few people were trying it.

Rereading his work, as I have been over the past few days, I have decided that I was unfair. Taleeb talks of "alternative histories". These, of course, are precisely what poker players are talking about when it comes to EV. If you make a play with a positive EV of $50, it doesn't matter whether you lose $600 or win $700 -- it was still the right play; even though that exact situation will never be repeated again. But in all the alternative universes, you are, in sum total, up.

But, as Taleeb says, try telling that to an accountant.

All this turned out to be serendipitous, because it appears that Man Financial is in a small bit of shit over Philadelphia Alternative Asset Management, a hedge fund set up last year which has comprehensively done its bollocks. Philadelphia was set up by Paul Eustace, a (wait for it) "star manager".

The details of the Man problem are fairly tedious (the old "secret accounts set up to conceal losses" tale - yawn), but the story of Paul Eustace is interesting.

Perhaps the investors in these funds really ought to go back and reread the opening chapters of Taleeb as well. Because here we have mention of just such a "star trader" (called "John"), who suffered the same fate as Paul Eustace. As the law of large numbers proves -- if you get 1m squirrels looking for nuts, and all of the squirrels are equally bright, then 100 of those squirrels will find a very large number of nuts, while another 100 will probably find none at all (and, on returning home, find that someone has set fire to their nut stash, making them down on the whole deal). If you choose to back the lucky 100, you might as well put all your money on red, "because that's the number that came up lst time".

What worries me is that some of these investors might have control of your pension.

Comments

[geoffchall.livejournal.com]

It's between gritted teeth that I find myself defending 'star' fund managers, but I do think your squirrel anaology is pertinent. You should back the 'lucky 100' to find the nuts, because they have not just shown luck but talent. They know where the nuts are buried.

In truth you should back the squirrel who finds nuts on a regular basis rather than just that season. But on a pure earnings value basis you should still be ahead from backing the 'lucky' squirrels than a random collection of squirrels. Yes you'll be backing some of them happened to be lucky bastards, but you will be more likely to hit some talented 'star' squirrels.

[jellymillion.livejournal.com]

I'd be happier with this thought if I hadn't worked over a number of years with a bunch of "squirrels" who wouldn't recognise a nut if it was rammed in their faces.

I have also worked with some major-league nut-gatherers about whom I would happily assert that their success was genuine talent. The problem is that they are as likely to be managing my money as my one-eyed cat is to rediscover the wonders of parallax. It just isn't worth their while to manage huge piles of pleb money.

[jellymillion.livejournal.com]

I thought the whole idea of Taleb's proposition was that he was operating on a different investment horizon to the majority of market players.

The shorter-term Johnnies are (in effect) laying the outsider at slightly too long a price. Which quite probably will work for long enough for them to have ensured their personal financial security. At which point everything else they make is jam - unless they believe their own hype and put their capital to work in the same way (an all-your-eggs and one-basket scenario, I submit).

Taleb, on the other hand, keeps backing the 225-1 long-shots at 250-1. He doesn't win very often but as long as he is accurately assessing the odds he'll win over the long term. But it has to be a long-term proposition. Fortunately, as he's managing a hedge fund of other peoples' money, he gets his annual management fee to keep the wolf from the door while he waits for the inevitable.

What worries me is that some of these investors might have control of your pension

They probably do. I'd strongly advise anyone who asked me (unlikely to happen, mind you) to ensure that their funds were well-diversified across a range of security types, markets, fund managers and any other dimension that sprang to mind. Keep the correlation down.

In the short term this obviously has the potential to reduce returns. In the long term it should reduce the risk of one's retirement income being $74 a week, although there's the obvious increased likelihood of one or more individual components going pear-shaped.

Taleb

[peterbirks.livejournal.com]

Yes, this was why I said that I misjudged him. I thought that his view was that the outsider line was always correct. In fact, what he is saying is that, because "star" traders work on short-time (personal) horizons, then they are far more likely to put $100m of someone else's money on a 1-to-100 shot that is really 1-to-150 than they are on a 100-to-1 shot that is really 66-to-1. I mean, wouldn't you, if the offer was 20% of any profit?

This is a reasonable - indeed, almost undeniable — analysis.

On Geoff's point. Taleb's point was that any star trader must, by definition, have been lucky, and that fact that he is a star trader (i.e., off the scale in terms of standard deviation) would tend to indicate that he is a risky trader. Coming back to poker, I note that no chip leader at the end of day one has ever gone on to win the world series Big One. This is because to become chip leader by the end of day one you have had to take some serious all-or-nothing gambles. Eventually these catch up with you. However, of the final table this year, I think that nine of the first 10 were in the top half.

One has to argue that a risky trader is by definition not a good one. Now, I guess one could argue against that, on the grounds that if the risky trader has a long-term average above that of "steady eddie", and if he has an unlimited bankroll, then your long-term return will be better with Jimmy "lump-it-all-in" Jayglo.

I was going to say that I would pick the trader with the steady record of outperforming by a few points. The problem is that, these days, that is probably the self-same guy who is taking on too much risk for very little extra in absolute return!

Personally, I pick my own investments and the only fund I am in is my 15-year endowment with Sun Life, and I think that all of that is in Treasuries.