Bankroll
A bankroll is the money you have set aside for betting and for nothing else. How much of it you put on one game decides how long you last, more than your picks do.
Money kept apart
A bankroll is a fixed sum, kept apart from the money you live on. The rent is not bankroll. Neither is a credit card. If losing all of it would change your month, it is too big.
Fixing the sum gives every bet a size. Without it, the size of a bet comes from how you feel that day, and a player who sizes by feel bets biggest when he is losing. I watched that from behind the counter for years, and I never once saw it end well for the player.
Units
Bettors measure a stake in units, and a unit is a set share of the bankroll, usually 1 or 2 percent. On a $1,000 bankroll a 2 percent unit is $20.
Most professionals don't put more than 5 percent of the bankroll on any one bet. That is a ceiling for the strongest opinion of the month, not the everyday stake.
Seven in a row
Here is what seven losers in a row take out of a bankroll at a flat stake.
| Stake per bet | Seven straight losers cost |
|---|---|
| 1% of the bankroll | 7% |
| 2% | 14% |
| 5% | 35% |
| 10% | 70% |
Seven in a row sounds like something that happens to other people. It isn't. Take a bettor who wins 53 percent of his bets, which is good enough to make money at -110: over 250 bets he has about an even chance (49 percent) of losing seven straight somewhere along the way, and a 77 percent chance of losing six straight. For someone picking at 50 percent the chance of seven in a row is 63 percent.
At 1 or 2 percent a run like that is a bad fortnight. At 10 percent it is the end.
The vig eats it anyway
A bankroll does not protect you from the price. Bet $20 a game at -110, win exactly half, and each bet costs you 91 cents on average. A hundred bets cost $90.91, which is 9 percent of a $1,000 bankroll, gone without a single unlucky bounce. The vig is the reason a break-even picker goes broke slowly instead of never (and the reason the book doesn't mind him at all).
What an edge is worth
There is a formula for the right bet size, the Kelly formula, which divides your edge by the odds you are getting. I'll skip the algebra and give the answers at -110.
- Win 52.38 percent and you break even. The formula says bet nothing.
- Win 53 percent and it says 1.3 percent of the bankroll.
- Win 55 percent, which very few people do over years, and it says 5.5 percent.
So the 1 to 2 percent unit and the 5 percent ceiling are not folklore; they are roughly what the arithmetic gives a good bettor. Most people who use the formula bet half of what it says or less, because nobody knows his true winning percentage, and the usual mistake is to think it is higher than it is. I'd bet the half and sleep better.
Flat or a percentage
A flat stake is the same dollar amount on every bet, simple and easy to keep records with. A percentage stake is recalculated as the bankroll moves: drop from $1,000 to $800 and the 2 percent unit falls from $20 to $16, so stakes shrink while you lose and grow while you win. Either works. What does not work is the third method, the one nobody admits to, which is raising the stake after a loss to get even.
Setting yours up
Fix the amount and keep it somewhere separate from your other money. Set the unit at 1 or 2 percent and leave it there. Set a loss limit for the day and for the week, and when you reach it, stop.
I'd also write down every bet: the price, the stake, the result. It takes a minute, and it is the only way to know your real winning percentage, which as you saw above is the number the whole thing hangs on.
If the bankroll goes to zero, it is gone. Refilling it from the rent money is a different problem, and there is help for that one.
The book has to deal every game on the board. You get to pass.