MLB Bankroll Management: Unit Sizing, Variance and the 52.38% Threshold

Editorial Team «how do you bet Baseball» June 10, 2026 Reading time: 22 min

The number every MLB bettor must learn first

If you only remember one number from this entire article, make it 52.38%. That’s the win rate you need to hit, betting -110 lines flat across a sample, just to break even. Anything above is theoretically profitable. Anything below — and that includes 50%, the rate that intuition says should be neutral — is a long-term loser. Most punters never internalise this. They feel like they’re winning when they hit half their bets and lose money for years without quite figuring out why.

The maths is mechanical. At -110 odds you risk £110 to win £100. To break even, your wins (£100 each) must exactly cover your losses (£110 each), which happens at a win rate of 110/(110+100) = 52.38%. The 2.38% gap between 50% and 52.38% is the bookmaker’s margin in raw form. It’s the price of admission. Before you’ve factored in any skill, model, or insight, the bookmaker is taking 2.38% of your action just for setting the line. Beat 52.38% and you’re profitable. Don’t and you’re not.

Now layer on the structural reality of MLB. The regular season runs 162 games per team, 2,430 games league-wide, plus the postseason. Most punters bet some number of those games every day from late March through October. That’s 200 days of action. Even at a strong win rate of 55%, you’ll have stretches of fifty bets where you go 22-28 and feel like you’ve forgotten how to handicap. The variance over a baseball season is genuinely enormous, and the bankroll-management discipline that survives it is unromantic, repetitive, and the difference between cashing out a profit in November and refilling your account in July.

This whole article is about that discipline. Not the maths of edges and models — that’s a separate problem. The discipline of how much you bet, when, and how you survive the inevitable bad weeks. Because the punters I’ve watched succeed at MLB long-term aren’t always the ones with the best handicapping. They’re the ones whose bankroll discipline is boring enough to compound.

What a “unit” actually is in baseball

Walk into any sports betting subreddit and you’ll see grown men post results in units, claiming “+47 units this season” with no context for what a unit actually represents in their bankroll. Half of those posters are betting 5% of bankroll per “unit” and don’t realise that’s three times what professional handicappers do. The other half are using “unit” as a synonym for “any number I felt like betting”, which makes the metric useless.

The professional definition of a unit is straightforward. Professional sports bettors typically wager 1-2% of bankroll per single bet, treating each as a “unit”. If your total betting bankroll — the money you’ve consciously set aside for sports betting, separate from your everyday finances — is £2,000, a unit is somewhere between £20 and £40. You bet one unit by default on every standard play, scale up to two units (4%) on high-confidence plays, and very rarely go to three units even on what feels like a lock. That’s it. There’s no clever variation hiding in the system.

The reason this small percentage matters comes back to variance. Even at a 55% win rate, you’ll occasionally lose seven of ten bets in a row through pure bad luck. At 1% units that’s a 7% drawdown, painful but recoverable. At 5% units that’s a 35% drawdown, which destroys both your bankroll and your psychological ability to keep handicapping clearly. The 1-2% rule isn’t conservative for the sake of being conservative. It’s the size that lets you withstand the inevitable variance without cratering.

One adjustment most beginners miss: a unit is a percentage of your current bankroll, not your starting bankroll. If you start at £2,000 and grow to £3,000, your unit is now £30 to £60, not £20 to £40. If you draw down from £2,000 to £1,500, your unit drops to £15 to £30. This re-pegging compounds your wins and protects your downside, and over a full season it makes a meaningful difference to long-term performance. The friction of recalculating after every bet is annoying, which is why most bettors update their unit size weekly or after every 20 bets, rather than continuously.

Flat staking versus percentage-of-bankroll

The first real bankroll choice every MLB bettor faces is whether to bet flat — the same pound amount on every bet — or as a percentage of their current bankroll. The right answer is “it depends on your goals”, and the wrong answer is “the one that lets me bet bigger when I’m winning”. Both methods have a place; let me walk through where each works.

Flat staking is the simpler option and the one I recommend for any bettor in their first full season. You set a unit size (say £25) at the start of the year, and you bet that amount on every standard play regardless of how your bankroll has shifted. The advantage of this approach is psychological — you’re not constantly re-evaluating the size of your bets, which keeps you focused on the bet selection itself. It also dampens the swings of variance: when you’re down, you keep betting the same amount and don’t risk grinding your bankroll down faster; when you’re up, you’re not exposing yourself to bigger losses on what might just be a hot streak.

The disadvantage is that flat staking under-compounds. If you’re profitable and your bankroll grows from £2,000 to £4,000 over the season, you should arguably be betting more — your edge hasn’t changed, but your buffer for variance has doubled. Flat staking keeps you betting £25 a unit even when your bankroll says you could comfortably bet £40 to £50.

Percentage-of-bankroll staking solves the compounding problem at the cost of more emotional volatility. You set a unit as 1.5% of current bankroll, and you re-peg either continuously or weekly. When you’re winning, your bets get bigger naturally — you bet £30 a unit on a £2,000 bankroll, £45 on a £3,000 bankroll, £60 on a £4,000 bankroll. The compounding is meaningful: a 20% return on bankroll grows much more aggressively than the equivalent flat-stake season. The downside is that bad runs hit harder emotionally because the dollar (or pound) amount of each loss starts shrinking too, which can feel like you’re “playing with house money” — and in fact you are, but the framing has caused more bankroll mistakes than I can count. My personal compromise is to flat-stake intra-week and re-peg my unit size every Monday. That gives me the compounding benefit without the constant calculation.

Kelly Criterion in plain English, and why fractional Kelly is safer

The Kelly Criterion is the mathematically optimal bet sizing formula, derived from information theory in the 1950s and adapted to gambling shortly after. The formula is simple: f = (bp – q) / b, where f is the fraction of bankroll to bet, b is the decimal odds minus 1, p is your estimated probability of winning, and q is 1 minus p. If you can plug in honest numbers, Kelly tells you the exact stake that maximises long-run growth.

The catch — and it’s a big one — is that Kelly is honest about probabilities. If you genuinely have a 55% chance of winning a bet at +100 (decimal 2.0), Kelly says bet 10% of bankroll. That’s an enormous bet by professional standards. If your actual probability is 53% rather than 55%, that 10% bet has a much smaller positive expectation and bigger drawdown risk than the formula suggested. The maths assumes your probability estimate is correct. In real-world betting, your estimate has uncertainty, and that uncertainty makes full Kelly aggressive to the point of being dangerous.

This is why almost every professional bettor uses fractional Kelly — usually quarter-Kelly, sometimes half-Kelly, almost never full. Fractional Kelly takes the size the formula recommends and divides it by two or four. The mathematical sacrifice is some long-run growth; the practical gain is dramatically reduced drawdown risk and far less sensitivity to errors in your probability estimate. Quarter-Kelly on a 55% bet at +100 is 2.5% of bankroll, which sits within the 1-2% professional range with a small upgrade for high-confidence plays. That’s roughly the equivalent of “two units” in a flat-staking system.

For most UK punters betting MLB, I’d suggest skipping Kelly entirely and using flat 1-2% units instead. Kelly is mathematically beautiful but practically unforgiving when your probability estimates are imperfect (which they always are). The 1-2% framework gets you 90% of the bankroll-protection benefit of Kelly with almost none of the modelling complexity. If you want to go further into Kelly territory, do it after a season or two of tracked results, when you have empirical data on how accurate your probability estimates actually are. There’s a useful related read on closing line value as a long-term metric that pairs naturally with Kelly thinking, because CLV tracking gives you the empirical feedback you need to calibrate probability estimates before betting them.

Variance over a 162-game season

I want to share a quote that changed how I think about variance, from a panel of professional handicappers writing for professionalgambler.org: “Professional sports betting isn’t as glamorous as people think. With 1,000 plays annually, we know our win percentage will be about 54 to 55 percent. We expect our bankroll to reach a new high only 5% of the time, meaning most days we will be below our all-time peak.”

Read that twice. A professional, betting at a 54-55% rate over a thousand annual plays — which is genuinely strong, well above the 52.38% break-even — expects their bankroll to be at a new all-time high only on roughly 5% of days. That means 95% of days they’re below their previous peak. They’re spending most of their year in some kind of drawdown. Those are the people who are unambiguously winning at the sport. The variance they sit in is the variance you’ll sit in too, but probably worse, because your win rate is probably lower.

The MLB-specific version of this is brutal. The 162-game season produces enough samples that your true win rate eventually emerges, but the journey is rough. A 55% bettor playing two games a day for a full season can have a streak of 25-30 losing days. They can have a stretch where they go 35-50 over 85 bets, watching a chunk of their bankroll evaporate while doing nothing wrong. Their judgement starts being questioned — by friends, by themselves. The discipline of bankroll management doesn’t make these stretches less painful; it just makes them survivable.

The stat that grounds this all: at a 55% win rate with disciplined flat staking, a bettor’s bankroll reaches a new high only on roughly 5% of days. Most days you’re not winning. Most weeks you’re not winning. You’re just slightly tilting the long-term trajectory in your favour, and the slight tilt compounds. The mistake is treating each day as a referendum on your ability — chasing losses on bad days, going small on good days because you’re “due for regression”, changing your unit size based on how you feel. The bankroll plan is the entire point. If you don’t trust the plan during drawdowns, the plan can’t help you.

Tracking ROI without lying to yourself

Here’s a confession. The first three years I bet MLB, I told myself I was profitable because I remembered my big wins better than my drip-drip losses. Then I started tracking every single bet — date, market, odds, stake, result — in a spreadsheet, and discovered I’d been roughly break-even, not profitable. The losses I’d been mentally writing off as “small” had been adding up to more than my “big” wins. I had been comfortably losing money while telling everyone I was up.

This isn’t unusual. Memory is a lousy P&L. Cognitive biases run heavily in favour of recalling successes more vividly than failures. A spreadsheet — or any tracking app — is not optional for serious bettors; it’s the entire foundation of knowing whether what you’re doing is working. The fields I track on every bet: date, sport, league, market type (moneyline, run line, total, prop), team or player, odds in decimal, stake, result, profit/loss, and a free-text note about why I made the bet.

That last field is the one most people skip and the one with the biggest long-term value. Six months later, when you’re sorting your spreadsheet by market type and discovering that your run line bets are -8% ROI while your totals bets are +12% ROI, the notes are what tell you why. They reveal patterns — “I bet too many home favourites at -200 last summer”, “my totals plays driven by weather are profitable, my totals plays driven by gut feel are not”, “I lose money on every Sunday day game”. Those patterns are where real improvement happens, and they’re invisible without tracked notes.

Some longer-tenured data is worth knowing too. In a long-run MLB study by Boyd’s Bets, April underdogs finished 1,487-1,866 — a 44.43% win rate, at average odds of +131.3, returning a +1.0% ROI on flat $100 wagers. That’s a small, durable, season-specific edge that took thousands of bets to identify. You won’t find anything like it in your own data without tracking. You probably won’t find it after a single season. But three or four seasons of clean records will start showing you what kinds of MLB bets are actually profitable for you specifically — and that’s the point of the exercise.

How to survive a 1-in-20 drawdown

Statistically speaking, a sample of 1,000 bets at a 55% win rate has roughly a 95% chance of containing a streak of at least eight losses in a row. That number startled me the first time I saw it modelled out. Eight losses in a row, on a strong win rate, just from variance. If you’re betting £30 a unit, that’s £240 in losses across a streak that says nothing about your ability to handicap.

The drawdown survival kit has three parts. First, the rule that you don’t change your unit size during a drawdown. The temptation is to either chase — bet bigger to win it back faster — or panic — cut your unit size in half because you’ve “lost confidence”. Both responses turn variance into permanent damage. The bet size you’ve calculated is right for your bankroll size. Stick to it.

Second, the rule that you don’t increase your bet count during a drawdown. The same temptation that produces chasing also produces “I’ll just bet a few extra games this Sunday to make some of it back”. This adds variance to a bankroll that needs less variance, not more. If you bet two games a day on average, keep betting two games a day during drawdowns.

Third — and this is the one I wish I’d internalised earlier — schedule a forced break after any drawdown of 15% of bankroll or more. Take three days completely off betting, no matter what’s on the slate. The break isn’t because the games are bad; it’s because your judgement after a 15% drawdown is no longer trustworthy. You’ll be tempted to lean on outcomes rather than process, to “trust your gut” instead of working the inputs methodically. Three days off, then return with the same plan you started the season with. The drawdown will have been priced into your final results either way; the break is so the next stretch of bets is sized and selected with a clear head.

Adjusting unit size by market type

Not every MLB bet should be the same size. The 1-2% framework is the default, but some adjustments based on market type are sensible and used by most professional bettors. The principle is simple: bets with higher variance — wider price ranges, narrower edges, less stable inputs — should be smaller, and bets with lower variance and clearer edges can be slightly larger.

Standard moneyline and run line bets in the -200 to +200 range are full-unit territory by default. The variance is well understood, the prices are tightly contested, and a flat 1.5-2% unit size works fine. Underdog moneylines beyond +200 — say +250 or +300 — should be sized down. Not because they’re bad bets necessarily, but because the variance on individual outcomes is much higher and a clean run of bad luck can drain your bankroll faster than a loss on a -150 favourite.

Player props get sized down further, typically to half a unit (0.5-1% of bankroll). The reason is that prop markets often have wider implicit margins (the bookmaker holds 8-10% on many props versus 4-5% on game lines), the lines move based on lineup news that’s not always available, and individual variance on a single player’s performance is much higher than on a team’s score. Half a unit on props means you can build a portfolio of bets without overcommitting bankroll to a higher-variance market.

Futures — World Series outright, MVP, Cy Young — are a different beast. These are fundamentally illiquid bets that tie up bankroll for months, and the variance is structural rather than statistical (a single injury can wipe out an MVP futures ticket regardless of skill). I cap my total futures exposure at 5% of bankroll, distributed across multiple positions, with each individual ticket no more than 1.5% of bankroll. That’s smaller than a regular game bet because the time horizon is six months and the bankroll has to work elsewhere during that time.

The handicapper mindset: marathon, not sprint

I want to close with a quote that sounds simple but has been hard-won: “Always remember that it is a marathon, not a sprint, and the best way to grow our bankroll is to grind out profits over a period of time.” That’s Jerald, a veteran handicapper writing for Predictem. He’s been around long enough to have learned the lesson the expensive way, and he writes about it like someone who’s still occasionally tempted to forget.

The marathon framing matters because the daily noise of MLB betting is loud. Two games a day, every day, six months a year. Every day produces wins and losses. Every week produces a hot stretch and a cold stretch. The temptation to react to recent outcomes — by changing unit size, by changing handicap method, by switching markets, by chasing or quitting — is constant. The handicapper mindset is the discipline of treating this season’s results as one data point in a multi-year journey, not as a verdict on whether your approach works.

Practically, this means a few things. First, set a season goal in terms of process rather than outcome. “I will track every bet, run my pre-bet checklist on every play, and stay within my unit size for 162 games” is a process goal. “I will be up 20 units by August” is an outcome goal that you have only partial control over. Process goals are achievable; outcome goals have variance baked in.

Second, evaluate yourself in chunks of 100 bets, not in chunks of one week. A hundred bets at 55% gives you a 95% confidence interval of roughly 47% to 63%. That means even a strong handicapper can finish a hundred-bet stretch at 47%, look terrible, and have done nothing wrong. The real signal is in 500-bet chunks, ideally 1,000-bet chunks. That’s why bankroll discipline matters so much: it gets you to the sample sizes where your actual edge becomes visible, instead of cratering before you’ve collected enough data to know what you’re doing.

Third, accept that the 1-in-20-days-at-a-new-high reality applies to you too. Most days you’ll be in some form of drawdown. The marathon isn’t getting to the finish line; it’s accepting that the journey to the finish line is mostly uphill, and that the people who finish are the ones who keep their pace steady when others are sprinting and quitting around them.

Prepared by the how do you bet Baseball editorial staff.