Quote:
Originally Posted by ATsai
11t,
Thanks for the analysis. Seems very accurate.
I guess that I have been overly degenerate in these prop bets then because I have been giving even money coin flip bets to fish in order to keep them entertained. Basically, I adhere to the DGAF school of giving gamble action to fish to keep the game atmosphere light and gambly. My past experience also indicates that certain fish and regs will start tilting if they get stuck on coin flips, so there is often +EV metagame benefits for me to do coin flip prop bets because I don't tilt when I get stuck.
FWIW, I have done coin flip prop bets for significantly more than than 1k in the past.
Recently, however, certain regs have been offering me odds on my prop bets, so I was curious about the bankroll implications.
One thing to keep in mind is that the kelly criterion gives you the *optimal* amount to bet to maximize your expected bankroll growth. It is not intended to be an upper bound. If you have a logarithmic utility of money (kelly criterion), it is probably still worth taking bets with an edge that are for more than kelly would suggest for your bankroll, as long as they aren't a *lot* larger than the kelly amount. A quick log utility estimator in Excel shows that with a bankroll of 20k, the 1000 bet with a 51.25%edge is ok, just not as good for you as a 500 bet (which is close to your optimal bet size).
If you are an aggressive gambler, the Kelly criterion represents the bullseye of amount to bet, and there's a reasonably wide range that offers similar expected utility. For instance, with a 20k bankroll, your 2.5% edge bet has a utility adjusted gain of $6.25 at $500 which is right around the kelly criterion bet. Betting $300 has a utility adjusted gain of $5.25 which is nearly as much. Betting $700 also has a utility adjusted gain of $5.25. Betting $1000 starts getting dicey, cutting your adjusted gain to roughly zero.
Generally the utility adjusted amount rises gradually with your bet, until it plateaus near the kelly amount, then comes down slightly at first, and starts going over a cliff usually somewhere around double the kelly amount.
This all assumes you have roughly logarithmic utility of money which should be relatively close to accurate for a very aggressive investor or an advantage gambler, but will be much too risk-loving for the typical person.
It's worth putting together a spreadsheet to analyze props like this. It isn't hard. Here's the formula I used:
Start by converting your bet to an EV equivalent bet of win X n% of the time, and lose X 1-n% of the time.
ln is the natural logarithm function in excel etc.
pwin is the probability (between 0 and 1) of winning the bet
bet is the bet amount
bankroll is your bankroll
util-adjusted-EV == exp( pwin*ln(bankroll+bet) + (1-pwin)*ln(bankroll-bet) ) - bankroll
Utility adjusted EV is basically logarithmic utils converted back into dollar units. For very small bets relative to your bankroll, this will be very close to linear EV (what we're normally talking about when we say EV).
The primary function of good bankroll management is to keep your bets at sizes where your utility of money function is close enough to linear that standard EV calcs are close enough. If you put too much money on the table your utility of money function starts to become important, and you become exploitable by people whose utility at those stakes is still close to linear (scared money).
For crazy prop-land, it's a good idea to have a sense of your real utility of money function, and what kinds of bets/edges will maximize it, or at least be in range of maximizing it.
Short answer: your 1k bets are not terrible and probably don't hurt you much unless you have a hard bankroll or are more risk-averse than a typical poker player.