Episode 129 ... Victor Haghani (42:49)

Father was a goods trader (Sephardic Jew born in Iran?)
Went to University in London (LSE)
His dad said go for the less bureaucratic firm (why he chose Salomon over JP Morgan)
John Merriwether asked him to become a trader, government bonds arb desk
Youngest trader on the desk
He had been in fixed income research at Salomon Brothers
Merriwether left Salomon in 1992, Haghani left in late 1992 ... founded LTCM
Started LTCM's London office
Worked for 13 bank consortium after LTCM failed in 1998 ... helped liquidate portfolio
JWM Partners hedge fund ... also helped start that
Founded Elm Partners five years ago ... "active index investing"
Lowenstein's "When Genius Failed" -- a good read, but not 100% accurate
Dunbar's book is also OK
Buy the Harvard Business School case studies on LTCM, by Andre Perold
We're a product of our experiences
Haghani wrote paper on biased coin flip
Gave 61 subjects (financial professionals) $25, coin biased 60% heads, could keep whatever they made after 30 minutes of flipping, capped at $250
[Sounds similar to Van Tharp's old experiment that he has given hundreds of times]
[Van Tharp gave subjects bag of ten marbles: Seven 1R losers, one 5R loser, two 10R winners. Subjects got 40 marble pulls and a $100,000 bankroll. Expectancy is 0.8R (positive) but most people end up broke because their bet size is too large and they revenge trade]
People bet a lot on tails :-) ... usually after a streak of heads [laughing]
People believe random things have some sort of predictability (human experience versus math)
People got bored of betting on heads [laughing again]
1/3 of people went bust betting on a 60:40 biased coin
1/5 reached max payout ... kids who could flip really fast with smaller bet size mainly
1/2 won $80
Using simple rule of only betting 15% of bankroll would give 95% chance of hitting max payout within 30 minutes
"Suboptimal behavior"
Nearly everyone voluntarily bet their whole stake at some point
Those all-in bets *always* happened after someone took a loss on an outsized bet, classic need to "get it back"
People who busted didn't want to talk about it
A whole range of bet sizes works (8 or 9% to 20%) to hit max payout within 30 minutes
Kelly Criterion number (optimal bet) was 20%
Optimal solution is very complicated, but just use heuristics (common sense)
Without the cap, expected value would be $3,000,000, 4% return on every flip (betting 20% of bankroll)
St. Petersburg Paradox ... expected value versus expected utility
People won't bother to play even if they have positive expected payout
Have to understand your own risk aversion
Betting 50% gives negative expected utility (with 60:40 coin)
Bet sizing is not simple, not secondary ... it's incredibly important [I say it's *everything*]
LTCM trade sizing was all screwed up (position sizes were way too big)
Global equities should have a positive expected return above the risk-free rate or inflation, trouble is the Sharpe Ratio
Thorp inspired the coin-flipping experiment
Haghani believes there are some rare people who can beat the market, trouble is finding them, identifying them in time
Past returns are not indicative of future returns (because we don't have enough data)
How do you identify the biased coin after only 30 flips? You can't, it takes 143 flips
Need to find an investor or trader with 143 year track record
"I don't have very much on the wisdom front"
www.elmfunds.com
Not on Twitter, "haven't figured it out"