US Open data shows why conventional sportsbook uptime can mislead

Traditional uptime data measures how much of a match a sportsbook is available for. Wagering-weighted uptime asks a different and commercially more important question: was the operator available during the periods of the match when betting activity was at its greatest?

Bettormetrics’ US Open analysis demonstrates how wagering activity can reveal whether sportsbooks are available when it matters most. With prediction market data being added alongside its existing exchange data, Bettormetrics can apply this commercially weighted view of performance across an expanding range of sports and markets, opening up a far richer understanding of betting patterns and operator performance, particularly in the US.

Why conventional uptime can mislead

Uptime is usually reported as a proportion of match duration: the share of time a market was live, regardless of how much money was moving through the wider market at any given moment.

It is a reasonable long-run efficiency measure, but it treats every stage of a match as equally important, even though wagering activity is not distributed evenly through an event. A sportsbook that is unavailable for the same length of time during a quiet first set and during a deciding tie-break receives the same duration-based penalty, despite the two periods having markedly different commercial significance.

A sportsbook can therefore report excellent uptime across the length of an event while being unavailable during the periods in which wagering is heaviest. A technically strong uptime percentage does not, on its own, show that an operator was available during the most commercially important moments.

How Bettormetrics measures it

Bettormetrics compared operator pricing and availability with exchange pricing and matched wagering activity across 39 in-play men’s singles matches at the US Open and 18 operators.

Performance was measured in two ways. By duration, each metric is the proportion of total match time for which it applied. By wagering, it is the proportion of exchange matched wagering that took place while that condition held. Wagering-weighted uptime, for example, is the share of exchange wagering transacted while an operator’s markets are live.

The data source matters. The US Open analysis uses exchange pricing and wagering data. Exchange wagering is not the same as sportsbook wagering and it cannot be used to directly calculate an operator’s actual lost revenue.

However, it does provide an independent, market-derived indication of when betting activity was concentrated, and therefore when sportsbook availability was very likely to matter most. Traditional betting exchanges and prediction markets are both exchange-based environments that produce market-derived pricing and wagering signals. Provided there is sufficient liquidity, the same methodology can therefore be applied to both.

For this analysis, “theoretical arbitrage” means that at least one sportsbook selection was priced so that a customer could theoretically back it profitably against the exchange. It does not mean that every opportunity was verified as executable.

LeoVegas: the clearest demonstration

LeoVegas posted a duration-weighted uptime of 97.8%, solidly in the upper half of the field and in line with Superbet, Paddy Power and Hard Rock. Its wagering-weighted uptime was 88.3%: the lowest of any operator in the sample and the only one below 90%. The 9.5-percentage-point gap between the two figures is more than double the next-largest gap in the field.

The Zverev vs Tabilo match makes the point particularly clear. LeoVegas was available for 98.5% of that match by duration, but only during periods accounting for 61.5% of exchange wagering activity. An operator that looked close to flawless on the conventional measure was therefore unavailable during periods that accounted for more than a third of the exchange betting in that fixture.

The pattern is consistent with LeoVegas’ downtime being concentrated in the moments of heaviest wagering, which is exactly what a duration-based measure cannot show.

Competitive pricing without disproportionate arbitrage exposure

Pinnacle carried the tightest average overround in the sample at 3.7%, comfortably the lowest of any operator tracked, yet its theoretical arbitrage exposure relative to exchange pricing was modest: 4.8% by duration and 6.4% by wagering, both well below the field median.

That combination is not the industry norm. Across the 18 operators, tighter margins were associated with more theoretical arbitrage, not less. A tighter book generally sits closer to the exchange’s fair price, and so slips into arbitrage more easily when that price moves. DraftKings, with the second-tightest overround in the sample at 4.5%, illustrates the pattern: 14.9% arbitrage by duration, roughly triple Pinnacle’s figure for a near-identical margin.

Pinnacle is the outlier. The tournament-wide relationship between margin and arbitrage would put a book priced as tightly as Pinnacle closer to 12%-13%, not under 5%. That is a materially lower arbitrage load than its pricing alone would predict, and is consistent with a suspension or repricing approach that works in step with sharp pricing.

FanDuel and Superbet provide further evidence that operators can price competitively without carrying the arbitrage exposure their margins alone would suggest. Superbet’s 5.3% overround came with 7.9% arbitrage by duration, roughly 1.8 points below the tournament-wide trendline for a book of that margin. FanDuel’s 4.9% overround came with the same 7.9%, around 2.6 points below trend.

Both also posted strong wagering-weighted uptime, at 97.5% for Superbet and 98.7% for FanDuel, so their disciplined exposure did not come at the expense of availability when wagering was heaviest. Neither is the standout on any single measure, but the combination makes both solid all-round performers.

The big players: Bet365 and DraftKings

The standout single-fixture figure in the dataset came from Mariano Navone’s five-set upset of Novak Djokovic. In that match, Bet365 was in theoretical arbitrage against exchange pricing on at least one selection for 56.5% of match duration. That is more than double the next-highest operator in that match, Ladbrokes at 27.1%, and roughly 3.5x Bet365’s own tournament-average arbitrage by duration of 16%, itself the highest average in the field.

Measured by wagering, the picture was less extreme: 16.7% of exchange wagering in that match took place while Bet365 was in arbitrage, against a tournament average of 11.1%. The gap between the duration and wagering figures is consistent with much of the arbitrage falling in quieter periods of the match. Bet365’s overround in the fixture, 5.1%, was close to its tournament average of 4.8%, so the exposure does not appear to reflect unusually wide or tight pricing in that match.

Elsewhere, DraftKings recorded the second-highest theoretical arbitrage exposure in the field by duration and the fifth-highest by wagering. As FanDuel’s key sportsbook rival in the US, that leaves it with significant ground to make up in its tennis offering.

The wider opportunity: prediction market data

The US Open findings show what a wagering-weighted view adds to sportsbook benchmarking. LeoVegas and Superbet posted similar duration-weighted uptime, yet their wagering-weighted figures were 88.3% and 97.5% respectively.

Bettormetrics will shortly be able to combine exchange data and prediction-market data. That will provide a broader set of independent pricing and wagering signals, and allow commercially weighted performance analysis to be applied much more widely wherever liquidity is sufficient, particularly across the NFL, NBA and other US sports.

A sportsbook’s headline uptime can look excellent while concealing unavailability during the periods when wagering activity is greatest. Bettormetrics can already reveal this using exchange data, and integrating prediction-market data will extend that capability across more sports and markets, opening up a new level of commercially meaningful sportsbook performance insight.

Why traditional uptime metrics can miss the mark

Sabin Brooks, CEO of Bettormetrics, says the findings highlight the limitations of conventional uptime metrics and the broader commercial value of measuring sportsbook performance against actual wagering activity.

“What the US Open clearly shows is that traditional uptime alone is no longer enough,” he explains. “It tells an operator how often it was available but not whether it was available when it mattered most.

“For example, LeoVegas and Superbet recorded almost identical uptime by duration yet finished more than nine percentage points apart when performance was weighted by wagering activity. That is a commercially significant difference which conventional reporting misses and senior management may never see.

“This analysis is one example of the wider value Bettormetrics brings to the industry. Across core markets and player props, we connect pricing, availability and wagering signals with confidence ratings that help operators reliably identify and prioritise the issues and opportunities with the greatest commercial impact.

“With prediction-market data now being integrated alongside our existing exchange data, we will extend that insight across far more sports and markets, particularly in the US. This has the potential to change how the industry measures, understands and improves sportsbook performance.”

Methodology note

Uptime, green, amber and arbitrage rates are calculated both by duration (share of total match time) and by wagering (share of exchange matched wagering activity transacted under that condition). The wagering weighting is based on exchange activity, which is not the same as sportsbook wagering. Arbitrage = at least one selection priced such that a customer could theoretically back it at a profit against exchange pricing (theoretical, not verified as executable). Amber = at least one selection priced between the exchange’s back and lay prices. Green = at least one selection available and priced favourably to the operator relative to exchange pricing. Unavailable = the operator’s market was not live.