Multi-Book Bankroll Rules for +EV Bettors
One bankroll, several balances, and the rules that keep them honest
8 min read
The Quick Answer
You have one bankroll and several balances, and the two are not interchangeable.
Size every bet from the total across all your books. Then check whether the book with the good price can actually fund it. Aggregate your exposure by outcome rather than by account, so the same side at two books counts once. Treat pending bets and promo balances as unavailable. And keep enough money at each book to take a real bet, which usually means funding fewer books properly rather than many of them tokenly.
None of this replaces ordinary bankroll management — it is what ordinary bankroll management leaves out, because it assumes one pooled number and a bettor with one account.
What this page inherits rather than redefines
- What a unit is, and how big it should be. Defined in bankroll management. This page deliberately quotes no percentage of its own — use the band you already run.
- Sizing by edge. The mathematics of proportional staking lives in the Kelly criterion. Everything here is about where the money sits, not how large the bet should be.
- Why you hold several accounts at all. Line shopping and account longevity are covered in avoiding limits. This page starts after you have opened them.
One disambiguation, because the titles collide: the “allocating funds” section of the bankroll-management guide is about varying stake size with confidence. It is not about allocating capital across books, which is this page.
Running several books is what makes expected value reachable in practice — the best price is rarely at the book you happened to open first.
The Rules, and What Breaks Without Them
| Rule | Do this | Failure mode |
|---|---|---|
| One bankroll, many balances | Size from the total. Check fundability per book before betting. | Sizing off a single balance, so your unit shrinks at the book you use most. |
| Aggregate exposure by outcome | Same side at two books is one position, sized as the sum. | One unit per ticket, four tickets, four units on one game. |
| Pending is not available | Track committed and available capital separately. | Overstating spare capital on a slate where bets settle late. |
| Promo balances count as zero | Exclude anything with playthrough from the sizing base. | A bankroll figure inflated by money you cannot move or withdraw. |
| Limited accounts have capacity, not stake size | Bet the book’s maximum; do not scale up elsewhere to compensate. | Abandoning +EV bets because they cannot be placed at full size. |
| Keep a usable float per book | Fund fewer books properly rather than many tokenly. | The best price appears at the account with $15 in it. |
| Rebalance on a schedule, not mid-slate | Move money weekly, between slates, with transfer time budgeted. | Chasing a number while a withdrawal sits pending for three days. |
| Measure drawdown on the total | One equity curve across every book. | Per-book curves that each look survivable while the total does not. |
Stranded Capital: Bankroll on Paper, Useless in Practice
The single-account model assumes your bankroll is available. Across books it usually is not. Money at the book that does not have the good number is bankroll in your spreadsheet and dead weight on the bet in front of you — and because it still counts toward the total, it inflates the unit you think you should be betting while reducing your ability to fund it.
Rebalance between slates, never during one
Withdrawals take days, deposits have caps, and some methods charge. Budget transfer time as a fixed cost of running several books and move money on a schedule — typically once a week, between slates. Trying to shift funds while a number is live means you get the transfer and miss the price.
A float, not a token
Decide the smallest balance at which an account is worth having, and keep every funded book above it. An account below that line is a price feed rather than a betting account — useful for knowing where the market is, useless for acting on it. Funding three books properly beats funding six badly.
Two Books, One Outcome
This is the error with the largest blast radius, because it feels like diversification while doing the opposite. Spreading a bet across accounts spreads the execution. It does nothing to the risk — if the side loses, every ticket on it loses at the same moment.
How it happens: you flag a game, take a unit at the book with the best number, then notice a second book is nearly as good and take a unit there too. Later the line moves and a third book becomes attractive. Each decision was individually reasonable and you now have three units on one result, with no single screen showing it. The fix is an exposure log keyed on the game and side rather than on the account.
The mirror image is worth watching too: taking opposite sides of the same game at different books, usually weeks apart and for reasons that made sense at the time, quietly converts two bets into a flat position that still pays vig twice. Neither error is visible from inside either account, which is the whole reason this page exists. For the related case of bets that are logically linked rather than identical, see the correlated-risk discussion in bankroll management.
When a Book Limits You, the Percentage Rule Stops Working
Win consistently and books restrict you. A limited account might accept twenty or fifty dollars a bet, and at any serious bankroll that is nowhere near a normal unit. The standard advice — stake a fixed percentage — simply cannot be followed there, and pretending otherwise leads people to abandon perfectly good bets because they cannot be placed at full size.
The reconciliation is to stop thinking of a limited account as having a stake size and start thinking of it as having a capacity. Bet what it will take. A small positive-expectation bet is still positive expectation, and the alternative is not a bigger bet, it is no bet.
The trap to avoid is compensating. If a book will only take a fifth of your unit, the answer is not to bet extra elsewhere to reach a nominal total — that is how a sizing rule gets quietly abandoned. Take what you can get at the good number, and let the total for that bet be smaller than usual. Keeping accounts usable for longer is a separate discipline, covered in avoiding limits.
Counting What You Actually Have
Three categories of money look like bankroll on a balance screen and are not:
- Pending wagers. Committed, not available. On a college Saturday with staggered kickoff windows, a large share of your capital can be in flight from noon until nearly midnight. Track committed and available separately or you will size against money that is already spent.
- Promo and bonus balances with playthrough. Closer to a coupon than to capital: not movable, not usable where the good price is, and not certain to convert. Count them at zero and treat whatever clears as a windfall.
- Money in transit. A withdrawal that has left one book and not arrived at another belongs to neither for sizing purposes, and this is precisely the window in which people forget it exists.
And one measurement note: keep a single equity curve across every book. Per-account curves are individually reassuring in a way the total is not — a drawdown split across four books looks like four survivable dips and is one serious one. For the underlying discipline this protects, see bankroll mistakes to avoid.
Know Which Book Has the Better Number
EdgeBoard compares books against a vig-free FairLine, so you know where the money needs to be.
Frequently Asked Questions
Is my bankroll the total across all my sportsbooks, or the balance at each one?
Your bankroll is the total, and your unit size comes from that total. But only the money sitting at the book with the good price is actually usable on that bet, so the total tells you how much to bet and each balance tells you whether you can. Treating a balance as its own bankroll leads to betting too small at the book you use most; treating the total as freely available leads to a unit you cannot fund where you need it.
How should I size bets at a book that has limited me to $20?
Bet the maximum it will take and stop pretending it is a percentage. This is the one place where the standard rule genuinely breaks: a limited account physically cannot accept a normal unit, so the choice is a small bet or no bet. A small +EV bet is still +EV, and the honest accounting is that the account now has a capacity rather than a stake size. Do not compensate by betting larger elsewhere to hit a nominal total.
If I bet the same side at two different books, is that one bet or two?
One position, sized as the sum. Two books do not make an outcome independent — if the side loses, both tickets lose together. Bettors who size each ticket at one unit because each is at a different book routinely end up with three or four units on a single game without noticing. Aggregate exposure by outcome, never by account.
Do pending bets count as part of my available bankroll?
No, and this is the most common accounting error when running several books. Money in unsettled wagers is committed. On a Saturday with four kickoff windows you can have a substantial share of your capital in flight for most of the day, and a dashboard that shows cash balances will quietly overstate what you have to work with. Track committed and available separately, or you will discover the difference at the worst moment.
Should bonus or promo balances be treated as bankroll?
Not until they are withdrawable. A balance carrying playthrough requirements is closer to a coupon than to capital — it cannot be moved, it cannot be used at the book with the better number, and part of it may never convert. Count it at zero for sizing purposes and treat any of it that clears as a windfall rather than as part of the base you are sizing against.
How many sportsbooks should a +EV bettor actually fund?
Enough that the best available price is usually reachable, and few enough that each account holds a workable float. Adding a book only helps if you keep enough money there to take a real bet when its number is the outlier — an account with a token balance is a price feed, not a betting account. If funding another book would push every balance below a usable float, the honest move is to fund fewer books properly.