What Is a Middle Bet? The Math Behind Middling Odds

A middle bet is two bets on opposite sides of the same market at different numbers, so both of them win when the result lands in the gap between the two. It is the only bet in sports betting where two tickets on opposite outcomes can cash together. Middling is not free, though, and the price of admission is exact: a middle breaks even when the gap lands as often as the hold on your two prices. This guide derives that number, shows what actually moves it, and hands you the tool that computes it on any pair you find.

The Middle Outcome Map

Enter two numbers and two prices. The widget names the exact results that make both bets win, prices the two branches, and shows the hit rate the gap has to clear. Drag the estimate slider below the break even line and watch the verdict flip.

The defaults are an NFL spread straddling 3. Laying the favorite at −2.5 at one book and taking the underdog at +4.5 at another, both at −110, costs $9.09 on the outcomes that miss and returns $181.82 on a margin of 3 or 4 points. The bar it has to clear is 4.76 percent.

What a Middle Bet Is

Two bets, opposite sides, different numbers. That is the whole definition.

A normal hedge takes both sides at the same number, so exactly one ticket wins and the other loses. Move the two numbers apart and you open a window between them where neither side has lost yet. Inside that window, both tickets are winners.

The window only exists when you take the sides in the right order. You need the side that wins on low results priced at a lower number than the side that wins on high results. On a spread, that means laying the favorite at the shorter number and taking the underdog at the longer one. On a total, it means taking the over at the lower number and the under at the higher one.

Get it backwards and the two bets overlap rather than gap, which is a different bet entirely. That one is an arbitrage opportunity, and it is worth reading next because the two shapes come from the same board.

The Three Outcomes

Every middle settles into exactly three branches, and the middle case is what you are paying for:

Stake the two legs for an equal return, meaning the payout is the same whichever single side wins, and the first and third rows collapse into one. That is not a stylistic choice. It is what makes the whole thing computable, because you no longer need to guess how the probability splits between a result below the gap and a result above it. You only need the chance of landing inside.

SmartStake sizes the second leg with the same function it uses to seed a real matched bet slip, so the widget above is doing exactly what the product does when you click into a pair.

The Break Even Hit Rate

Here is the part no competitor page states plainly.

Write o₁ and o₂ for the two decimal prices and T for everything you put up across both legs. Equal return staking pays R on either single winner, where

$R = \frac{T}{\frac{1}{o_1} + \frac{1}{o_2}}$

That denominator is the summed implied probability of the two prices. It is also, exactly, the number SmartStake sorts its arbitrage board by. Call it A.

Outside the gap you collect R and gave up T, so you are down T(1 − 1/A). Inside the gap both legs pay, so you collect 2_R_ and you are up T(2/A − 1). The difference between those two branches is one full payout, R, no matter what the two prices are.

Set the expected result to zero and the algebra collapses to one line:

$q^* = \frac{1}{o_1} + \frac{1}{o_2} - 1$

The hit rate a middle has to clear is the hold on the two prices. Not the width of the gap. Not the sport. Not the stake. Two sides at −110 add up to 104.76 percent, a hold of 4.76 percent, and that middle needs the gap 4.76 percent of the time. Two sides at −105 add up to 102.44 percent, and the same middle now needs 2.44 percent.

If you have read what a low hold bet costs, this is the same quantity doing a second job. There the hold was the price of running money through a market. Here it is the hurdle a middle has to jump.

Width Versus Value

The result above has a consequence that surprises most people: widening a middle does not lower the bar.

Hold the prices at −110 and stretch the gap as far as you like. The payoff on a hit is still two payouts, the cost on a miss is still the hold, and the break even is still 4.76 percent. What a wider gap buys you is a better chance of clearing that same bar, because more results fall inside it.

Notice the middle column. A gap does not pay off on a range, it pays off on a list of whole numbers, because a game lands on a margin of 3 or 4 and never on 3.5. SmartStake's middle engine computes exactly that list, rounding the low end up and the high end down, and the product then says it in plain words: if the favorite wins by 3 to 4 points, both of your bets win.

That is why a 1 point gap sitting on 3 in the NFL is worth far more than a 1 point gap sitting on 8. The bar is identical. What differs is how much probability lives on the numbers inside. How a point spread settles covers why football margins pile up on 3 and 7 in the first place, and it is the right companion to this page when you are judging whether a gap is in a good neighborhood.

To turn that into a number rather than an instinct, price each side against a sharp book and devig the pair with the devigging calculator. The difference between the two devigged probabilities is an estimate of how often the result lands between them.

Middles That Push

Half the middles you will actually see are narrower than they look, and they behave differently.

Put a whole number on one side of the gap and no result can make both tickets win outright. Lay −2.5 and take +3, and a margin of exactly 3 pushes the +3 ticket, returning that stake, while the −2.5 ticket wins. The engine calls this scenario a push win rather than a both win, and the difference in money is large.

A both win adds a full payout to the outside case. A push win only hands back the stake on the leg that pushed. At −110 on both sides, that is $90.91 instead of $181.82, and the break even doubles:

$q^*_{push} = \left(\frac{1}{o_1} + \frac{1}{o_2} - 1\right) \times o_{push}$

At −110 that is 4.76 percent times 1.909, or 9.09 percent. That is the whole story of the classic 2.5 and 3 middle. A margin of exactly 3 is the most common single result in football, and the bar this bet sets lands close enough to the usual estimates of that frequency that your own read of the number, not a comfortable cushion, decides whether the bet is worth taking. Estimate it against a sample you trust rather than assuming it clears. That closeness is also why books guard the 3.

Totals behave the same way. Over 41.5 against under 42 pushes the under whenever the game lands on 42, so it carries the same doubled bar as the spread version.

The Price Is the Lever

If the bar is the hold, then shopping the two prices is not a detail. It is the entire lever you control.

Five cents of price, the smallest increment most books quote, cuts the required hit rate roughly in half. Nothing about the gap changed. You did not find a better number or a softer market. You just paid less to be there.

This is the same argument line shopping makes for straight bets, sharpened. On a single bet a better price adds a bit of expected value. On a middle it moves the threshold that decides whether the bet is worth making at all.

Middles and Arbitrage

The two shapes are neighbors, and the relationship is cleaner than most guides admit.

An arbitrage is prices that overlap: the implied probabilities sum to under 100 percent, so every result settles the same way. A middle is numbers that gap: most results cost a little and the ones inside pay twice. Nothing stops a pair from being both. When the two prices overlap and the two numbers still leave a gap, the break even hit rate computes negative, which is the arithmetic saying the pair covers its own cost before the gap is counted at all. The widget shows this as None needed, because there is no hit rate you have to reach. It still assumes both legs are accepted at the prices shown, which is where these pairs most often come apart.

SmartStake's board treats them as one feed with one column between them. A pair with no gap between its numbers is an arb; a pair with a gap is a middle; the Arbitrage Finder's Middles Only filter is literally a filter on that column, and both are ranked by the same summed probability the formula above uses. The arbitrage calculator sizes either one.

That single column, though, is a width, and this article is about why width is not value. The board can tell you the gap exists and what it costs. Only you can supply the third number, the chance the result lands inside it.

Where Middles Come From

Middles exist because the number moves and the old price stays on your ticket. Three sources produce most of them:

• Line movement over time. Take a side early, and news, weather, or money moves the market past you. The other side is now available at a different number, and your original ticket has become half of a middle.
• Disagreement between books at the same moment. Books post different numbers on the same game, and occasionally the gap between two books points the right way. These are the ones a scanner surfaces.
• In play scoring. A live total drops fast during a scoreless stretch, so an over taken early sits several points under the current number. Live arbitrage betting covers how these appear and how quickly they close.

The first source is the one most bettors stumble into without a name for it. If you are holding a ticket and the market has moved through your number, you are one bet away from a middle, and the question in this article is the one to ask: does the gap land often enough to beat the hold on the pair you would be creating?

Finding Middles on the Board

Scanning books by hand for gaps is the part that does not scale. The SmartStake Arbitrage Finder streams matched pairs across the books it covers and carries the middles alongside the arbs, with the Middles Only filter isolating pairs whose numbers leave a gap. Open one and the page names the exact winning results and the return on each branch, from the same engine that powers the widget above. The Arbitrage Finder walkthrough covers the filters in detail.

The workflow that follows is short. Read the winning numbers, judge how often they land, compare that to the break even the hold implies, and take the ones with real room. Skip the rest.

The Risks Worth Naming

A middle is a small, defined cost most of the time and a large payoff rarely, which makes it honest but lumpy:

• Most results miss. A middle that clears its bar comfortably still loses on the majority of bets. The gap is a rare event by construction. Size the bets so a long run of misses is boring rather than alarming, which is what bankroll management is for.
• One leg may not fill. If the second bet is rejected, limited, or repriced, you hold a naked position at a number you chose for a different reason. Place the harder side first.
• A void or a rule difference breaks the pair. Two books settling a market differently, or a leg voided for a scratched player, can leave the two tickets no longer covering each other.
• Both bets shrink your limits. Two sides of the same game at two books is a visible pattern. Avoiding betting limits covers what the account side of this looks like.

None of that makes middling a bad bet. It makes it a bet with a threshold, and the threshold is knowable before you place it. A middle is never truly without risk, and any individual bet can still win or lose.

Frequently Asked Questions

The Short Version

A middle is two bets on opposite sides at different numbers, and it wins twice when the result lands in the gap. The bar it has to clear is the hold on the two prices, which means the gap's width decides your chance of clearing that bar while the price decides where the bar sits. Half point middles that only push double the bar, so read the numbers before assuming a gap is a gap.

Take a pair you are looking at, drop the two numbers and two prices into the widget above, and compare the break even it prints against your own read of how often that result lands. Then go find the next one on the Arbitrage Finder.