Tuesday Tool Series #8: Middle Finder

Tuesday tool series, article 8 of 19. The tool: the Middle Finder at theoddsgap.com/middles

You like the under in Ravens at Falcons on Sunday night. Your app has the total at 45.5, so you bet the under 45.5 at -110 and move on.

Here is what that app does not show you. On the Monday before Week 5, most books had that game at 45.5. Two of the major US books did not. Caesars hung 45. DraftKings hung 46.5. Same game, same moment, a point and a half apart, both at -110.

Most bettors never see that, because they only see their own book. Nobody is hiding it. But a point and a half between two books is more than a slightly better price. Put the over at the low number and the under at the high number, and there is a result where both tickets cash. That is called a middle, and finding them is what this tool does.

What the tool does

A middle is two bets on opposite sides of the same line, placed at different numbers. Over 45 at one book, under 46.5 at another. If the final total lands between them, both bets win. If it lands anywhere else, one wins and one loses, and you pay the margin on the losing side.

That shape is the whole appeal. Small cost when it misses, big payout when it hits.

The Middle Finder scans every sportsbook we carry for those disagreements. It reads the main line and every alternate line on spreads, totals and player props, pairs the lowest over against the highest under (and the shortest spread against the longest on the other side), sizes the two stakes, and shows you three numbers for each one: what you collect if both win, what it costs if only one wins, and the exact results that land in the window.

At 2:41 PM on Monday it had 370 middles on the board across every sport, 72 of them in the NFL, sorted best payoff to risk first.

A real example

Back to Ravens at Falcons, Sunday, October 11, 8:20 PM ET.

At 2:47 PM, here is where the main total sat:

• Caesars: 45, over -110, under -110
• FanDuel, BetMGM, BetRivers, Hard Rock Bet, theScore Bet and most of the board: 45.5
• DraftKings: 46.5, over -110, under -110

The Middle Finder pairs the two ends. Over 45 at Caesars, under 46.5 at DraftKings, $100 split evenly, $50 on each side because the prices match. Here is every way it can end:

• The total lands on 46. Both bets win. Each $50 at -110 profits $45.45, so you collect $90.91.
• It lands on 45. The over pushes and refunds your $50. The under wins $45.45. You are up $45.45.
• It lands on 47 or more, or 44 or fewer. One side wins $45.45 and the other loses $50. You are down $4.55.

That is the trade. Risk $4.55 to win $90.91, with a half-win on 45 as a bonus. The card counts only the numbers where both bets cash, so it calls this a 1.5-point window that hits on 46 and leaves the push out.

Is that worth it? That depends entirely on how often an NFL total lands on 46, which the tool does not claim to know. What we can give you is the break-even: ignoring the push on 45, a $4.55 risk against a $90.91 payout pays for itself if the total lands on exactly 46 a little under 5% of the time, about one game in 21. Whether a given number clears that bar is the question you are answering when you take a middle.

Props are where middles get wider, because books disagree more on players than on games. Here is one of Monday night's NFL prop middles.

Tyler Shough, pass plus rush yards, Falcons at Saints. FanDuel had the over at 277.5 for -113. DraftKings had the under at 282.5 for -114. Anything from 278 to 282 wins both, a 5-yard window. On $100, the finder puts $49.90 on the over and $50.10 on the under, so either single win pays the same. Both win: +$88.11. One wins: -$5.95.

That game was played Monday night, so treat it as an illustration of the shape, not a bet you can still make.

How to use it

Pick the market. Choose a sport, then All, Spreads, Totals or Props.

Set the books to the ones you have. The Books filter removes any book you cannot bet. A middle at a book you do not have an account with is not a middle.

Set a minimum window and a stake. The window filter hides anything narrower than the number you type. The stake is your total across both legs, and the finder splits it so the two single-win outcomes cost the same.

Read the sentence at the top of the card. It tells you, in plain words, what result wins both: "Total lands on 46" or "Winning margin of 9." Then the three numbers underneath: hit profit, miss cost, and the stake they are measured on. You can edit either leg's stake and the numbers update.

Bet the flagged leg first. One leg on every card carries a Bet This Leg First tag. That is the price furthest above the average for its own line, which makes it the one most likely to disappear. Lock it, then place the other side. If the second price is gone by the time you get there, you are holding an ordinary bet at a good number, which is a fine place to be.

Free and Live. A visitor on the free hourly board sees one middle from the latest hourly snapshot. The full list, refreshed every minute, is on Live Beta, which is free with an email address.

The math

For two prices in decimal odds, d1 and d2, add their implied probabilities: 1/d1 + 1/d2. Call that number S. On a normal two-sided line, S is a little over 1, and the amount over 1 is the margin you are paying.

The finder splits your stake in proportion to each leg's implied probability, so either side winning alone returns the same amount. With a total stake T:

• If only one side wins, you get back T divided by S. Because S is above 1, that is a little less than T. That gap is the miss cost.
• If both win, you get back twice that, minus your stake.

For Ravens at Falcons, both legs are -110, which is 1.909 in decimal. S is 1.0476. On $100, one side winning returns $95.45, a $4.55 miss cost. Both winning returns $190.91, a $90.91 profit.

If S ever comes in under 1, the two prices pay more than they cost even when only one side wins, and the middle is also an arbitrage. Those rank at the top of the list automatically.

A few rules keep the list honest. Game-line legs priced worse than -350 are dropped as noise. The prediction markets and exchanges are left out, because they generally hold their spreads and totals at the consensus number rather than posting the alternate lines a middle needs. The five offshore books are off by default, as they are everywhere on the site. The Middle Calculator runs the same math on any two prices you type in, and the methodology page covers how every price on the site is read.

What it does not do

This is the section nobody else writes, so here it is.

It does not tell you how often a window hits. The card shows you the payout and the cost. It does not estimate the chance the result lands in the gap, and that chance is the whole answer to whether a middle is a good bet. A 1-point window on a number that rarely comes up can be a bad trade at almost any price.

It is not free money. Most middles miss. When they miss, you pay. The asymmetry makes them attractive, not guaranteed, and the only middle that cannot lose is one that is also an arbitrage.

The prices move. Every number in this article is a snapshot from Monday afternoon. The best middles are also the ones most likely to be gone first, because two books that disagree tend to stop disagreeing. Check both prices before you bet either leg.

Books limit accounts that do this a lot. Betting both sides of a line at two different books is a pattern books notice. That is their call to make, and it is worth knowing before you build a habit around it.

Props have their own rules. Books do not all settle a player prop the same way if the player leaves early or does not play. Check both books' rules before you rely on a prop middle.

It does not pick a side. Nothing here tells you whether a game goes over or under. It tells you where two books disagree enough that you might not have to choose.

The point

Most of the time, two books a point and a half apart is just a better price waiting for whoever checks. Once in a while, it is a bet where both sides win. The Middle Finder is how you see the second kind without opening every app you have.

Free, no signup for the hourly board. Also on iPhone: The Odds Gap on the App Store.

Every price above was read off the live Middle Finder at 2:41 PM ET on Monday, October 5, 2026, and the Ravens at Falcons total was checked against every book's main line at 2:47 PM ET. Prices move, so always verify both prices before placing either bet. Nothing here is a pick. Please gamble responsibly.