Are tennis match odds a good bet? What ATP prices actually pay

Match odds are the main tennis market: one line, two players, pick the winner. The prices look plain, and on the favorite side they mostly are. The cost sits on the long side, and it gets steeper the longer the price gets.

Written by Jesse, NegativeEV. Last updated 27 August 2026.

Graded against 19 seasons of completed ATP matches - 36,345 of them, 2008 through 14 August 2026, at Bet365's closing price - the sides quoted at 5.00 or longer (about +400 in American odds) were priced as though they win 13.5% of the time. They won 10.5%, give or take 0.4. Backing every one of them would have returned about -26% per bet. That gap is not a bad stretch or an unlucky season: the sign held in all 19 seasons.

What does a long price in tennis really cost?

Roughly a quarter of the stake, on average, before a ball is struck. Taken together, those long sides carried prices implying they win about one time in seven. They won closer to one time in ten. Nothing on the slip shows the difference - the price implies one number, the record produced a smaller one, and the distance between them is the whole cost.

Does the same pattern show up at shorter prices?

Yes, and it runs in one direction the whole way down the ladder. Grouped by what the closing price implied, across the same 36,345 matches:

Price implies	91.3%	75.5%	60.5%	42.8%	28.0%	13.5%
Actually won	88.8%	72.8%	56.0%	40.7%	24.2%	10.5%

Every bucket came in under its own price, and the shortfall widens as the price lengthens. That is the longshot bias, one of the most durable findings in betting-market research, and tennis match odds show it about as cleanly as any market gets measured.

Is the favorite side the cheap side, then?

Cheaper, not cheap. Over the same 36,345 matches at closing prices, backing every favorite returned about -4.9% per bet and backing every underdog about -12.0%. Narrowing to recent seasons only, 2024 through mid-August 2026 and 6,442 matches, the same split held: about -3.4% on favorites and -9.9% on underdogs. Both cuts were negative in all 19 seasons. There is no side of this market that pays for itself - only a more expensive one and a less expensive one.

Why do tennis prices cost more than a baseball game line?

Two things stack. The first is the price band. A tennis draw produces a lot of lopsided matchups, so a large share of the board sits out at 4.00, 6.00, 10.00 and beyond - the longshot range, where the shortfall above is widest. A baseball moneyline rarely leaves the near-even band at all.

The second is the built-in margin. The two-sided hold on those Bet365 match-odds quotes averaged 6.2%, against about 3.9% on an MLB game line. A book sets each line to balance its own action and bank its margin, and a market of one-sided, mostly-on-the-favorite-or-the-story-pick action leaves room to carry a thicker one. Six percent held on every quote is most of what the graded results show.

Does that make every long price a bad bet?

No. Those figures are averages over tens of thousands of matches, and an average that negative still contains prices that are fair or better. The point is that the label does not sort them. At 6.00, the difference between an honest price and an expensive one is a couple of percentage points of win rate - invisible by eye, and invisible on the slip. Surface, opponent style, recent workload and a five-set format all move the real number, and none of them are readable from the odds.

What does this page not establish?

Whether any single match on the board is priced wrong. The figures here are averages over 36,345 matches, and an average does not sort one match from the next. Every price is one book's number at close, so the page does not say what the same match was worth earlier in the day or at another book. The window ends on 14 August 2026, so anything after it sits outside the measurement. This page measures match odds only. The 6.2% hold is the book's margin across both sides of a quote, which is a separate quantity from what a bet returned.

How can a tennis match bet be checked?

NegativeEV simulates each ATP match thousands of times, point by point, and counts how often the player actually wins across those simulations. That rate is compared with what the posted price implies, and the bet grades out as an expected value. Paste a bet slip or type the bet at negativeev.com and it comes back graded, free, with no account.

More on the number: https://negativeev.com/about/expected-value The measurement behind this page: https://negativeev.com/research/atp-match-odds