What Did Arsenal See in David Raya?

In the summer of 2023, Arsenal signed David Raya from Brentford. At first on loan, then permanently for around £27 million. The reaction from fans was mostly confusion.

Aaron Ramsdale had just finished two very good seasons as Arsenal's number one. The team was winning. The clean sheets were piling up. He was a fan favourite with a personality that lit up the dressing room. Why would you mess with that?

We pulled the goalkeeper data from every Premier League season since 2013-14 and looked at the one stat that strips away everything else: save percentage. The answer to "what did Arsenal see?" becomes pretty obvious once you look at it.

The Stats That Made Ramsdale Look Great

Give credit where it's due. Ramsdale's two full seasons at Arsenal were genuinely impressive by the numbers that most people look at.

In 2021-22, he won 21 of his 34 starts, kept 12 clean sheets, and conceded just 39 goals. The next season was even better. 26 wins from 38 starts, 14 clean sheets, and a title challenge that went down to the final weeks. Across those two seasons combined, Arsenal won 65% of the games Ramsdale started and kept a clean sheet in 36% of them.

If you're a fan watching from the stands, those numbers look brilliant. You're winning most weeks. The net is staying empty more than a third of the time. Why would you change anything?

The Stat That Told a Different Story

Wins and clean sheets tell you how good the team is. Save percentage tells you how good the keeper is. It strips away everything around the goalkeeper and just asks one question: when a shot on target came at you, did you stop it?

So we ranked every Premier League goalkeeper by save percentage in 2022-23 (minimum 15 starts). Here's where Ramsdale landed.

Tenth. Out of 22 regular starters. Behind keepers at Fulham, Bournemouth, and Brentford.

The previous season told a similar story. Ramsdale's save percentage was 70.3%, which put him 8th in the league. Decent. But not the kind of number that matches a keeper being called one of the best in the country.

We took it a step further with a metric called Goals Prevented Above Average (GPAA). It works out how many goals a league-average keeper would have conceded from the same shots on target, then compares that to what actually happened. Positive means you saved more than average. Negative means you didn't.

Across two full seasons at Arsenal, Ramsdale's total GPAA was +2.2. That means over 72 starts, he saved roughly two more goals than a completely average Premier League keeper would have. Two goals. In two years.

Not bad. Not elite either.

What Raya Was Doing at Brentford

Now look at what David Raya was doing at the exact same time, at a club with a fraction of Arsenal's defensive talent.

Look at 2022-23 specifically. Raya was facing 5.2 shots on target per 90 minutes at Brentford. That's nearly 50% more than Ramsdale was dealing with behind Arsenal's defence. And he was saving 78.2% of them. That's 9 percentage points higher than Ramsdale.

His GPAA that season was +17.7. That means Raya prevented almost 18 more goals than a league-average keeper would have in the same position. It was the best single-season GPAA by any Premier League goalkeeper in our entire dataset going back to 2013. Better than De Gea's legendary 2017-18 (GPAA +17.4). Better than anyone.

If you're an analytics department at a club trying to win the league, and you see a goalkeeper preventing 18 extra goals per season at Brentford while your current keeper is preventing one, you're making that phone call.

The Defence Effect

This goes beyond just Raya and Ramsdale. Great defences make goalkeepers look better than they are. Worse defences reveal what a keeper can actually do.

Arsenal's defence in 2022-23 was one of the best in the league. Saliba, Gabriel, Ben White. They limited opponents to just 3.6 shots on target per 90. Ramsdale wasn't being tested very often. And when a goalkeeper faces fewer shots, their clean sheet numbers go up and their goals against numbers stay low, even if they're just average at the actual shot-stopping part.

Raya saw the same effect in reverse. At Brentford, he was under siege. 5.2 shots on target per game. Every time he pulled off a great save, there was another one coming 15 minutes later. That workload is what let his ability show up in the data.

When Raya moved to Arsenal, something predictable happened. His GPAA dropped. In 2023-24 it was actually slightly negative (-1.6). Not because he got worse. Arsenal's defence was so good that he barely faced any shots. His shots on target per 90 went from 5.2 at Brentford to just 2.2 at Arsenal.

By 2024-25, with more shots coming his way (3.2 per 90), Raya's GPAA bounced back to +4.5 with a 71.7% save rate. The underlying quality was always there. He just needed the chances to show it.

This is the same reason Ederson, despite eight seasons at Manchester City with the most wins and clean sheets of any keeper in the league, has a career GPAA of -2.7. He's below average at actual shot-stopping. But when you face just 2.5 shots on target per 90 behind City's defence, it barely matters. He wins because his team wins, not because he's preventing goals that other keepers wouldn't.

The Verdict

Arsenal's analytics team almost certainly saw some version of what we've laid out here. The specific metric might not have been GPAA, but the signal was the same. When you strip away the defence, strip away the wins and the clean sheets, and look purely at shot-stopping ability, Raya was operating at a completely different level to Ramsdale.

Over four qualified Premier League seasons, Raya's career save percentage is 73.8%. That puts him 5th in career GPAA among all Premier League keepers since 2013, behind only Nick Pope, Hugo Lloris, Alisson, and David de Gea. Ramsdale, across five seasons at Bournemouth, Sheffield United, Arsenal, and Southampton, sits at 67.8%.

None of this means Ramsdale was a bad keeper. He did his job at Arsenal and did it well enough to help challenge for a title. But "well enough" is the point. Arsenal weren't looking for well enough. They were looking for a keeper who could add something that the defence alone couldn't provide.

The numbers say they found one. And the decision that confused so many fans at the time was probably the most data-literate goalkeeper signing in Premier League history.