We Scored the Same NBA Odds Two Ways and the Order Changed

We ran 1,046 NBA price comparisons and ended up with 1,558 best-price wins.

Then we changed how those wins were counted, and FanDuel and bet365 swapped places.

Every quote stayed the same. The difference was how we treated sportsbooks tied for the highest price.

That rule affected one in three comparisons in our sample. It was enough to change which of those two books looked better.

Before I trust a "best NBA odds" ranking, I want to know how it handles that.

WagerWise's analysis covered 9,076 eligible quotes from 19 bookmaker feeds across nine NBA games. Prices were reconstructed one hour before each game's scheduled start, with at least five feeds required in every comparison.

We matched the same game, market, selection and exact line. Lakers -4.5 was compared with Lakers -4.5. Over 228.5 was compared with over 228.5.

That produced 1,046 exact comparisons.

In 351 of them, more than one feed shared the highest price. That's 33.6% of the sample.

How you score those ties matters.

Take a hypothetical spread where two sportsbooks offer -110 and the others offer -115. Both -110 books offer the highest quote.

Give each a full win and that comparison produces two wins. Split the credit and each receives half.

The first method measures how often a book offered the highest price, including shared leads. The second measures its share of the leadership.

Both are useful. They should be labelled clearly because they can produce different orders.

FanDuel came out ahead when every tied leader received a full win. Bet365 came out ahead when the credit was shared.

Across the whole sample, counting every joint leader produced 1,558 best-price wins from 1,046 comparisons. Splitting ties produced exactly 1,046 credits.

Those extra 512 win labels came from shared leads. They weren't extra opportunities.

A reader looking for somewhere to place a bet may care most about access to the highest price. An analyst allocating leadership may prefer fractional credit. The table needs to say which question it answers.

Then it needs to show the denominator.

FanDuel appeared in 1,018 comparisons. Bet365 appeared in 375. Those percentages describe different sets of opportunities, even before we consider the scoring rule.

Pinnacle provides an even clearer example. Its fractional best-price rate was 44.68%, compared with FanDuel's 21.01%.

That looks convincing until you add coverage.

Pinnacle appeared in 180 comparisons, covering 17.2% of the eligible groups. FanDuel covered 97.3%.

Pinnacle performed strongly within its recorded sample. We have no price to assess in the groups where it was absent.

I'd want that distinction beside the percentage. A 44.68% rate across 180 opportunities and a 21.01% rate across 1,018 opportunities tell us different things.

Even the choice between a count and a rate changes the order.

FanDuel accumulated 213.85 fractional credits, slightly more than BetMGM's 210.70. Divide by each feed's opportunity count and BetMGM moves ahead, at 27.22% against 21.01%.

Neither result requires different prices. It requires a different question.

Game totals shared the highest price in 259 of 576 comparisons, or 44.97%. Point spreads tied in 86 of 452, or 19.03%.

Totals therefore had roughly 2.36 times the tie frequency of spreads in this sample.

A leaderboard containing mostly totals encounters a different pattern from one containing mostly spreads. Combining them into one percentage carries that market mix into the result.

Our sample included alternate lines. Spreads and totals accounted for 1,028 of the 1,046 groups. Just 18 were moneyline comparisons.

That composition belongs in the description. Readers shouldn't have to discover that an "NBA odds ranking" contains very little moneyline data.

For anyone building a comparison with an NBA odds API, keeping the exact selection, line and timestamp attached to each quote makes these checks possible. Without that context, the table can look precise while comparing different bets.

There's also a limit to what the highest price tells you.

A sportsbook can lead every other recorded quote and still offer a price below your estimate of fair odds. Evaluating a positive EV betting edge also requires a credible estimate of the selection's probability. Our analysis measured displayed prices, with no betting returns modelled.

The nine games were played on February 24, 2026, US time. They form a small historical case study using international feeds, rather than a season-wide ranking of sportsbooks available to one bettor. The primary comparison required source and recorded state timestamps within five minutes of the checkpoint.

Within that scope, the finding is clear: changing how ties counted reversed FanDuel and bet365's order.

Before I accept a sportsbook ranking, I want its opportunity counts, coverage, market mix and tie rule.

Those are part of the result. Put them beside it.