Carlos Alcaraz and Aryna Sabalenka: A Surface Elo Deep Dive

Carlos Alcaraz and Aryna Sabalenka currently sit at No. 2 in our overall Elo rankings on their respective tours. Both trail the tour’s No. 1 by a meaningful but not enormous margin overall. Split their ratings by surface, though, and the two stories diverge: one player’s weakest surface number still leads the whole tour, and the other’s rating on a single surface drops enough to fall out of the top five. This article walks through both, using our surface-specific Elo model and the training run dated September 27, 2026 (03:10 UTC), built on 34,943 processed matches. Ratings are recalculated weekly from our growing match database, so treat every number below as a snapshot of that date, not a permanent figure.

Why the overall number alone understates this

An overall Elo rating blends a player’s level across every surface they’ve played into a single figure. That’s a reasonable summary of general form, but it can hide a lot about any one surface specifically. We’ve covered the mechanics of this in general before; this piece applies it to two specific players whose surface splits turn out to tell very different stories from their overall rank.

Carlos Alcaraz: his weakest surface rating still leads the tour

Surface	Alcaraz Elo	Alcaraz rank	Sinner Elo	Sinner rank
Overall	2024	2	2132	1
Hard	1947	2	1951	1
Clay	1954	2	2000	1
Grass	1928	1	1891	2

Carlos Alcaraz trails Jannik Sinner in the overall Elo table (2024 to 2132) and on both hard (1947 to 1951, a 4-point gap) and clay (1954 to 2000, a 46-point gap). On grass, the order flips: Alcaraz’s 1928 is ahead of Sinner’s 1891, a 37-point edge, and it’s enough to make Alcaraz the No. 1 grass-Elo player on the ATP tour right now.

What makes this worth a second look is that grass is also Alcaraz’s own lowest surface rating. His gap from his own overall number is largest on grass (2024 minus 1928, a 96-point drop), slightly larger than his hard gap (77 points) and his clay gap (70 points). So Alcaraz doesn’t lead the grass list because grass is relatively his best surface by his own standard. He leads it because Sinner’s grass number falls even further below Sinner’s own overall rating: Sinner’s grass Elo (1891) sits 241 points under his overall figure (2132), by far the widest surface-to-overall gap of any player in this article. Clay pulls Sinner’s overall rating up a long way (his clay number, 2000, leads the entire ATP clay list by 46 points over Alcaraz); grass does the opposite.

Sample sizes here are solid for both players: Alcaraz’s hard, clay, and grass ratings are each built on his full 102 tracked matches, and so are Sinner’s. The split isn’t an artifact of one player having played less tennis on a given surface than the other; the hard-to-reach exception is Alcaraz’s own grass log, covered below.

Alcaraz’s last 10 logged matches

Our database currently shows Alcaraz’s 10 most recent logged matches running from April 10 to September 26, 2026: an 8-2 record, with the two losses coming against Ben Shelton (September 8) and Jannik Sinner (April 12, on clay). The most recent entry, a win over Taylor Fritz on September 26, is tagged “i.hard” (indoor hard) in our log, even though our model still rates it under the single combined “hard” surface alongside outdoor hard matches. We don’t currently split hard courts into indoor and outdoor ratings, a limitation worth keeping in mind as the indoor swing of the season picks up: a “hard” rating from this point forward is still a blend of both.

That log also explains the grass number’s smaller sample: Alcaraz’s grass rating is built on just 12 tracked matches, his thinnest surface sample of the three (hard: 56 matches tracked; clay: 23). A top-10 overall grass rating on a 12-match sample is still a real signal, not an error, but it’s the thinnest-sampled number in this article and should be read that way.

Aryna Sabalenka: a podium overall number, a top-of-the-table hard rating, and a clay drop

Surface	Sabalenka Elo	Sabalenka rank	Context
Overall	2015	3	behind Rybakina (2040) and Gauff (2025)
Hard	2018	2	behind Rybakina’s 2033
Clay	1868	7	147 points below her own overall
Grass	1882	2	behind Swiatek’s 1904

Aryna Sabalenka sits third in our overall WTA Elo table at 2015, behind Elena Rybakina (2040) and Cori Gauff (2025). Split by surface, her hard-court rating (2018) is actually 3 points above her own overall figure, placing her second on the WTA hard list, just behind Rybakina’s 2033. Among every player compared in this article, including Alcaraz, Sinner, Djokovic, Swiatek, Rybakina, Gauff, and Jessica Pegula, Sabalenka’s hard rating is the only surface number that sits above that same player’s overall rating. Every other surface figure we checked sits below its player’s overall number, sometimes by a wide margin.

Clay is the opposite story. Sabalenka’s clay Elo, 1868, is 147 points below her own overall rating, the largest surface-to-overall gap of any player discussed here, including Sinner’s 241-point grass drop relative to his own much higher overall figure (Sinner’s gap is larger in absolute points, but his overall rating is also unusually high; Sabalenka’s clay gap is the larger one relative to where she sits in the standings, dropping her from third overall to seventh on the surface-specific list). Her clay rating also sits behind Rybakina’s clay number (1873) and well behind Iga Swiatek’s overall grass leader status, discussed next.

On grass, Sabalenka’s 1882 places second, 22 points behind Swiatek’s tour-leading 1904. Swiatek’s own overall rating (1973) sits fifth, another case of a player whose best surface-relative performance doesn’t match her overall rank.

Two names are worth flagging by their absence rather than a number: neither Rybakina nor Gauff appears in our WTA grass top 20 at all. That is not a claim about a thin grass sample for either player; it simply means neither rating currently clears the 20th-place cutoff on that list, which sits at 1743 (held by Mirra Andreeva) for this training run.

Sabalenka’s last 10 logged matches

The 10 most recent matches in Sabalenka’s log run from August 16 to September 12, 2026: an 8-2 record, with losses to Sara Bejlek (August 19) and Rybakina (September 12). Every one of those 10 matches is tagged “hard” in our data, consistent with a hard-court-heavy stretch of the calendar. As of this writing (October 1), her log’s most recent entry is 19 days old. That is simply when she last played a match that has been logged in our database by this training run, not a sign that anything is wrong with the data or that she has stopped playing; a player’s log can sit a few weeks behind the present date purely because no new match has been recorded since.

Turning these gaps into probabilities

None of the numbers above are forecasts for a specific match; they’re ratings that convert into a win probability only once you plug in two specific players. The clay gap between Sinner and Alcaraz (46 points) and the grass gap between Swiatek and Sabalenka (22 points) are both modest, not decisive, edges by Elo standards: in our other analysis of this model, a 100-point gap works out to roughly a 60-65% win probability for the higher-rated player, and gaps well under that are correspondingly closer to a coin flip. If you want the exact number for any two ratings rather than estimating by eye, the Elo probability calculator does that conversion directly.

What these ratings don’t do

A surface Elo number is a model output, built from logged match results, not a prediction of any single upcoming match. It doesn’t know about a recent injury, a coaching change, or how a specific player performs at one particular tournament’s conditions versus another. It also doesn’t currently separate indoor hard courts from outdoor ones, which matters for Alcaraz specifically given his most recent logged result (discussed above) falls in that indoor stretch. And as with any Elo-style system, a rating built on a smaller match sample, like Alcaraz’s 12-match grass number, carries more uncertainty than one built on 50-plus matches, even when it clears whatever cutoff a leaderboard uses for inclusion.

We also don’t use this article, or any of our editorial content, to make picks for specific upcoming matches. If you want to see how a model like this one performs when it’s actually used to find value, that record is tracked separately and kept current at /results/ and /track-record/, rather than quoted here where the numbers would be stale within days.

Where to check the live numbers

Every figure in this article is a snapshot from the September 27, 2026 training run. Because the model retrains weekly from a growing match database, the live numbers can differ by the time you’re reading this. Check the full Elo leaderboard for the current ratings across all three surfaces and both tours, and see why surface-specific ratings matter for the mechanics behind the model used throughout this piece.