Bankroll Management for Baseball Bettors
How unit sizing, break-even math, the Kelly formula and simulated risk of ruin work for baseball bettors, with worked examples and a plain account of what bankroll management cannot do.
Bankroll management is the set of rules a bettor uses to decide how much money to risk on any one wager and how much to set aside for betting in the first place. It is a way of controlling risk. It is not a way of creating profit. No staking plan can turn a losing bettor into a winning one, and none can guarantee a profit: a sportsbook builds a margin into every price, and most people who bet on baseball lose money over time. This article is general education about the arithmetic, not advice to place any bet, and it contains no picks.
With that said, the arithmetic is worth knowing, because baseball is an unusual betting sport. A team plays 162 regular-season games, so there is a game on the board almost every day from spring to fall. A bettor who wagers on many of them, and who sizes each wager by feeling, can lose a bankroll to ordinary bad luck even if his judgment is sound. The rules in this article are designed to keep a bettor in the game through the swings that every baseball season produces.
Gambling carries a real risk of loss, and it is meant only for adults of legal age in places where it is legal. If gambling is causing problems for you or someone you know, call the national helpline at 1-800-GAMBLER, and see the resources at the National Council on Problem Gambling.
What a bankroll is, and what it is not
A bankroll is a sum of money set aside for betting that you can afford to lose completely. That last clause is the whole definition. It is not money for rent, bills, savings or debts, and it is never money borrowed to bet with. If losing it would change your life, it is too large to be a bankroll.
The bankroll gives you two things. It gives you a number to size your wagers against, and it gives you a point at which to stop. A bettor who has decided that his bankroll is $1,000 knows, before the first pitch, the most he can lose in total. A bettor with no bankroll has no limit, and the absence of a limit is where most of the damage happens.
Keep the bankroll separate from your everyday money, in its own account if possible, and track every deposit, wager and withdrawal. The record is not for taxes or bragging; it is the only honest way to know how you are really doing, because memory keeps the wins and edits out the losses.
The price of a bet: break-even math
Before sizing a bet it helps to know what a price demands. Every American-odds price implies a probability that the bettor must exceed to break even. For a negative price (a favorite), the break-even probability is the amount risked divided by the total of the risk plus the win. For a positive price (an underdog) it is 100 divided by the price plus 100.
| Price | Risk to win | Break-even win rate |
|---|---|---|
| -110 | $110 to win $100 | 52.38% |
| -120 | $120 to win $100 | 54.55% |
| -150 | $150 to win $100 | 60.00% |
| -200 | $200 to win $100 | 66.67% |
| +100 | $100 to win $100 | 50.00% |
| +120 | $100 to win $120 | 45.45% |
| +150 | $100 to win $150 | 40.00% |
| +200 | $100 to win $200 | 33.33% |
The -110 line is the standard price for a coin-flip game: the bettor risks 110 to win 100, and the 10 extra is the sportsbook's margin, often called the vig or juice. A bettor who wins exactly half of his -110 bets loses 4.55 percent of every dollar he risks. To break even he must win 52.38 percent. That gap, about 2.4 points, is the first hurdle every bettor faces, and it is the reason a staking plan cannot create an edge by itself. For how prices differ from one book to the next, see how to compare odds and shop for lines.
What an edge is worth
Expected value, or EV, is the average result of a bet if it were made many times. For a bet at -110 where the true chance of winning is p, the expected profit per dollar risked is p times 0.909 (the win on each dollar) minus (1 minus p). The table shows how small the margins are.
| True win rate at -110 | Expected profit per $100 risked |
|---|---|
| 50.00% | -$4.55 |
| 52.38% | $0.00 |
| 53.00% | +$1.18 |
| 55.00% | +$5.00 |
| 57.00% | +$8.82 |
| 60.00% | +$14.55 |
A bettor who wins 55 percent of his bets at -110 earns 5 cents on each dollar risked, which is an extraordinary result over a long period. Most people who track their records honestly will find their true rate is at or below the break-even line. The point of the table is that an edge, if one exists, is thin: a difference of a few percentage points decides whether a bettor makes or loses money, and the staking plan must be built for a world in which the edge is small and uncertain.
Unit sizing: the flat-stake method
The most common approach is to define a unit as a fixed percentage of the bankroll, usually between 1 and 2 percent, and to bet one unit on each wager. With a $1,000 bankroll and a 1 percent unit, a bet is $10. If the bankroll grows to $1,200 the unit may be reset to $12, and if it shrinks to $800 the unit is $8. Many bettors recalculate the unit only at fixed intervals, such as monthly, to keep things simple.
The reasoning behind a small unit is variance. Even a bettor with a real edge will have losing streaks, and the streaks are longer than intuition suggests. At -110, with a 52.38 percent win rate, the chance of losing five bets in a row is about 2.4 percent for any particular five bets; over a thousand bets, a streak of eight or more losses is common. In the simulation described below, the median longest losing streak in 1,000 bets was 8 for a bettor winning 55 percent of the time and 9 for one winning 50 percent. A 5 percent unit loses 40 percent of a bankroll in an eight-bet streak. A 1 percent unit loses 8 percent.
Some bettors vary stakes by confidence, using one unit for a lean and two or three for a strong opinion. That can be reasonable if the confidence levels are tested against results. The danger is that confidence is a feeling, feelings are weakest after a hot streak or a cold one, and the larger stakes tend to land exactly when judgment is worst.
How much can go wrong: a simulation
A simulation shows the stakes better than a formula does. We ran 4,000 trials for each case, with a bettor placing 1,000 flat-stake wagers at -110 and starting with a bankroll measured in units. "Ruin" means the bankroll fell below one unit at any point. These numbers are the results of one model, not a forecast, and each carries a margin of error of about a point.
| Bankroll in units (unit size) | True win rate 50.0% | True win rate 52.38% (break-even) | True win rate 55.0% |
|---|---|---|---|
| 100 units (1%) | 5.1% ruined | 0.1% ruined | 0.0% ruined |
| 50 units (2%) | 55.9% ruined | 10.7% ruined | 0.5% ruined |
| 20 units (5%) | 91.5% ruined | 54.0% ruined | 10.9% ruined |
| 10 units (10%) | 97.2% ruined | 75.0% ruined | 34.6% ruined |
Look at the middle column. A bettor with zero expected profit, who wins just enough to cover the margin, goes broke in more than half of the runs when he stakes 5 percent per bet, and in three runs out of four when he stakes 10 percent. A bettor with a real 55 percent edge and a 10 percent stake still goes broke one time in three. The edge is not what saves him; the size of the stake is. Even for the bettor with the 55 percent edge and a 1 percent stake, the typical worst peak-to-trough decline over 1,000 bets was about 21 units, and one run in ten dropped 34 units or more. A 55 percent bettor must be prepared for a fifth of his bankroll to disappear at some point.
The 50 percent column, finally, shows what happens to the bettor who is simply guessing: the vig grinds him down, and even a conservative stake only slows it.
The Kelly formula, and why people fear it
The Kelly criterion is a formula for the stake that maximizes the long-run growth rate of a bankroll, given a known edge. With b as the net payout on a one-unit stake (0.909 for a bet at -110) and p as the true probability of winning, the Kelly fraction is (b times p minus (1 minus p)) divided by b.
A worked example: a bettor who is certain that a bet at -110 wins 55 percent of the time should stake (0.909 times 0.55 minus 0.45) divided by 0.909, which is 5.5 percent of the bankroll. At a true 53 percent the formula gives 1.3 percent, and at the break-even 52.38 percent it gives zero.
The catch is the word "certain". Kelly assumes that the bettor knows the true probability. No one does. Suppose the bettor believes the chance is 55 percent and stakes 5.5 percent, but the true chance is 53 percent. His stake is more than four times what Kelly would recommend for the real edge, and the expected long-run growth of his bankroll is slightly negative (about minus 0.07 percent per bet in logarithmic terms), even though each bet individually has a positive expected value. Over-betting past about twice the proper Kelly fraction produces a negative growth rate, however good the individual bets look.
For that reason, those who use Kelly at all use a fraction of it, often half or a quarter, and cap the stake at 1 to 3 percent of the bankroll in any case. A fractional stake protects against overestimating the edge, which is the most common mistake in the whole field.
Mistakes that sink bankrolls
Four errors do more damage than any poor choice of side.
- Chasing losses. Raising the stake after a loss to win the money back turns an ordinary downswing into a crisis. The next bet has the same odds as the last one; it does not owe you a win.
- Doubling after each loss. The Martingale system doubles the stake after every loss so that one win recovers everything. It fails on arithmetic. After seven straight losses at a $10 starting stake, the eighth bet would be $1,280, and the total lost would be $1,270. A run of seven losses at even odds turns up, on average, about once in every 128 sequences of seven bets, and a sportsbook's maximum bet can stop the doubling before the bettor's money does.
- Betting too many games. With a full schedule every day, the temptation is to find a bet in each game. Each bet carries the margin, so more bets mean more cost without any more edge. A bettor who bets only when he has a reason bets less often than the schedule allows.
- Parlays as a bankroll tool. A parlay combines several wagers, and the margin is applied to each leg. The payout looks large, but the chance of winning falls fast: four legs, each with a 55 percent chance, win together only about 9 percent of the time. They are entertainment priced at a premium, not a way to grow money.
How long it takes to know
A bettor who thinks he has an edge needs a long record to confirm it. The standard error of an observed win rate, for a true rate near 55 percent, is about 5 points after 100 bets, 2.2 points after 500, 1.6 points after 1,000 and 1.0 point after 2,500. A bettor who wins 55 percent over 1,000 bets is only about 1.7 standard errors above the break-even rate of 52.38 percent, which is suggestive but far from proof. A bettor who wins 56 percent over 100 bets has learned almost nothing.
This is the honest limit of the whole subject. A season is short, a hot month is common, and much of the apparent skill that bettors report is the kind of variation that luck alone can produce. Bankroll management cannot tell you whether your edge is real. It can only keep you solvent while you find out, and the answer is often no.
Baseball-specific points
Baseball has features that matter for staking, though they are about information, not predictions.
- The favorite is rarely a lock. Even a strong team wins only a modest majority of its games over a season, so a -150 price, which needs 60 percent to break even, asks a lot. Prices vary by game, but no baseball game is certain.
- Pitching changes move the price. A scheduled starter who is scratched can change the line, and a bet placed on a listed pitcher may be treated differently depending on the book's rules. Read how listed pitchers versus action affect bets before betting a game.
- Weather affects totals. Wind and temperature change how a ballpark plays. The effects, and the limits of what a bettor can know, are in weather, wind and baseball totals.
- Shop the price. A small difference in price, repeated over a thousand bets, can be worth as much as a real edge. The odds page shows the lines this site tracks.
Honest limits and when to stop
Bankroll management does not tell you what to bet, whether a price is fair, or whether you should bet at all. It is a risk control and nothing else. It cannot overcome the margin built into every price, and it cannot protect someone who is betting to escape stress, chasing a loss, or hiding the losses from family.
Set limits before you start and keep them: a deposit limit, a time limit, and a hard stop at which you walk away for the season. Treat the money as the cost of entertainment, the way you would treat a ticket to a game. If you find yourself borrowing, lying about it or betting with money you need, stop and talk to someone. The helpline at 1-800-GAMBLER is free and confidential, and the broader picture, including warning signs, is in responsible gambling and baseball. The long relationship between the sport and wagering is covered in the history of baseball and gambling.
Common questions
What percentage of my bankroll should I bet? There is no universal answer. Many bettors who track records use 1 to 2 percent as a unit, and the simulation above shows why larger stakes raise the risk of ruin quickly. A beginner may choose a smaller unit than that.
Is Kelly worth using? Only with caution. It requires an accurate estimate of your edge, which almost no one has, so fractional Kelly with a hard cap is the conventional version.
Should I increase my stake when I am winning? Only on a schedule, by recalculating the unit from the bankroll at fixed intervals, not by feeling.
Does bankroll management guarantee I will not lose? No. It limits how fast you can lose, not whether you can.
What if I want to bet for fun? Decide the amount in advance, treat it as spent, and use a unit small enough that a long losing streak does not matter.
Where to go next
Read responsible gambling and baseball for limits and warning signs, how to compare odds and shop for lines for price differences, and weather, wind and baseball totals for the information that moves totals. Current lines are on the odds page, and the topic page for baseball betting collects more articles. If you or someone you know needs help, call 1-800-GAMBLER.