Across 238,716 player-matches from 2018 to 2026, players returning from a layoff win less often than their ELO expects, on a monotone curve running from 0.9 percentage points below expectation after 8 to 14 days out to 9.8 points below after more than a year. The deficit fades over four to six matches back.
Previews of a returning player always mention rust and never say how much. We measured it on every player-match in our database from 2018 to 2026, 238,716 observations, and added the curve to our model.
Rating systems treat inactivity as uncertainty, not as a deficit. Glickman's Glicko system adds a Rating Deviation per player, and "time passing without competing in rated games always increases a player's RD" [1], [2]. A wide RD on either side pulls the predicted probability toward 50%: a statement about what is known, not about how well the player will play.
Sipko's Imperial College thesis, a widely cited feature-engineering treatment, includes FATIGUE (games played in the past three days, discounted at 0.75 per day; our measurement of that window is the fatigue study) and RETIRED, a flag for a player's first match since retiring from one [3]. Its only published magnitude for RETIRED is a raw split on 2004-2010 ATP data, 48.5% won and 51.5% lost, not a residual against a rating. Sipko considered time since retirement as a severity measure and dropped it because "a player that has not competed for longer has also had more time to recover, so the relationship is unclear" [3]. We found no published layoff curve after that. ELO-based methods are among the most accurate published tennis forecasters [4], so a systematic error in ELO is an error in a strong baseline.
Each observation is a player-match residual: won (0 or 1) minus the ELO-expected win probability, in percentage points. The as-of rating is the stored post-match rating minus the delta applied, so nothing after the match is read.
The headline keeps one observation per match, the rustier side, and drops matches where both players sit in the same layoff bucket: the two residuals in a match are exact negatives and would cancel. The layoff clock counts every match in the database, including exhibitions and juniors, because any time on court counts. Walkovers are skipped.
| Layoff | n | Observed minus expected (pp) | 95% CI | Implied ELO overstatement |
|---|---|---|---|---|
| 8-14 days | 24,088 | −0.89 | [−1.46, −0.31] | +6 |
| 15-28 days | 21,993 | −2.41 | [−3.01, −1.81] | +17 |
| 29-56 days | 13,556 | −4.20 | [−4.95, −3.44] | +30 |
| 57-90 days | 6,163 | −4.00 | [−5.11, −2.89] | +29 |
| 91-180 days | 4,591 | −4.20 | [−5.47, −2.93] | +31 |
| 181-365 days | 2,518 | −4.88 | [−6.62, −3.13] | +36 |
| 366+ days | 768 | −9.75 | [−12.85, −6.64] | +73 |
The right-hand column inverts the logistic at each bucket's mean expected probability: how many rating points the player's ELO was overstated by. Two weeks off is worth about 6 points, a month about 30, a year 73, roughly the gap between a solid Challenger regular and a player two rungs above. The curve replicates on ATP and WTA separately, on the post-2023 window, and on the surface rating book.
A residual method can create effects out of rating miscalibration, since returning players skew toward one side of the probability range. We ran the same measurement on players with no layoff (0 to 7 days off, against an opponent who had played within 14 days): +0.06pp, CI [−0.09, +0.22], null. The shipped curve is therefore exactly zero below 8 days.
For returns of 91 days or more, by match since coming back:
| Match back | 1st | 2nd | 3rd | 4th to 6th | 7th to 12th | 13th to 30th |
|---|---|---|---|---|---|---|
| Residual (pp) | −3.11 | −2.77 | −2.34 | −0.91 | null | +0.54 |
It re-converges over four to six matches: a stale rating plus a short-lived rust component, not a permanently weaker player. The positive last cell is one bucket of eight with no multiplicity correction, where about 0.4 false positives at p<0.05 are expected, so it is not encoded. The curve never becomes a bonus.
A player whose previous match ended in their own retirement underperforms by a further −1.85pp, CI [−2.86, −0.84], n=8,284, across layoff bands, additive to the layoff penalty. When the opponent retired, the cell is null: −0.59pp, CI [−1.53, +0.35]. The effect follows the player who broke down, not the short scoreline. Only the own-retirement side is encoded.
Split by whether the rustier side was the favourite, 119,692 observations since 2020:
| Layoff | Rusty favourite (pp) | Rusty underdog (pp) |
|---|---|---|
| 8-14 days | +0.37 | +0.51 |
| 15-28 days | −1.90 | +0.02 |
| 29-56 days | −4.63 | −0.99 |
| 57-90 days | −7.38 | −2.54 |
| 91-180 days | −6.71 | −3.03 |
| 181-365 days | −9.01 | −3.92 |
| 366+ days | −14.33 | −2.51 |
| Pooled | −3.93 [−4.34, −3.52] | −1.48 [−1.81, −1.16] |
Symmetric Glicko attenuation would shrink the rating gap for both sides and push a rusty underdog's residual positive. Rusty underdogs are negative in every bucket from 15-28 days onward, so the effect is directional: the rusty side underperforms whether favoured or not.
A constant ELO subtraction does not fit either. At the observed mean rating gaps (+129 for favourites, −163 for underdogs), a flat penalty predicts a ratio of about 1.08 between the two pooled residuals; the observed ratio is about 2.7. We cannot derive that from first principles, so the shipped curve is the pooled measurement and the split is recorded, not fitted.
The curve went through our backtest gate as variant "pen": 2023-01-01 cutoff, full-history rating state, paired bootstrap against the unmodified baseline. In cells with n of at least 500 it is positive in 13 of 15, significant in 10, positive on all three surfaces, and it wins at penalty scales 0.5, 1.0 and 1.5. That first validation window sat inside the 2018-2026 fit window, so we reran it out of sample: curve refit on data through 2024 only, gate on a 2025 cutoff. Result: 14 of 15 cells positive, 9 significant, all three surfaces positive, and the refit curve within a few ELO points of the full-sample one in every bucket.
The improvement is small: about 0.0003 to 0.0012 Brier on a base of about 0.22, or 0.15% to 0.5% relative.
A variant adding K-factor inflation on top, so a stale rating also re-converges faster, scored higher on hard and clay, in several cells nearly twice as high, but was negative in all three grass cells, and K inflation alone was the weakest of the three variants. Our gate requires the same sign on at least two of three surfaces, and a variant that helps on two surfaces and hurts on the third is one we do not understand. We shipped the plain penalty, as with our surface-ELO change.
Method: 238,716 player-match observations, matches 2018 through 2026, our own ELO system (main tour through Challenger and ITF), pre-match ratings only: the as-of rating is the stored post-match rating minus the update just applied, so nothing after the match is ever read. Each observation scores won (0/1) minus the ELO-expected win probability, in percentage points; the headline view keeps the rustier side of each match and drops matches where both players sit in the same layoff bucket. Layoff is calendar days since the player's previous played match, cross-surface by design, counted over every match in the database including exhibitions and juniors; walkovers skipped. The shipped penalty is keyed on the gap that STARTED the return, not the gap since the immediately preceding match: once a 200-day returnee plays once their current gap is three days, but their second match back is still −2.77pp, so a current-gap penalty would collapse to zero exactly where the effect is largest. Confidence intervals are match-clustered; the decay split is measured on returns of 91 days or more and applied only to that population. The favourite/underdog split is 119,692 observations since 2020. The backtest gate is variant pen on a 2023-01-01 cutoff with full-history rating state and paired bootstrap intervals against the unmodified baseline; the first gate window sat inside the fit window; the out-of-sample check (curve refit on pre-2025 data, gate scored 2025 onward) confirms it: 14 of 15 cells positive, 9 significant, all three surfaces positive.
References:
[1] M. E. Glickman, "The Glicko system," Harvard University. https://www.glicko.net/glicko/glicko.pdf (accessed Aug. 12, 2026).
[2] M. E. Glickman, "Parameter estimation in large dynamic paired comparison experiments," Journal of the Royal Statistical Society: Series C (Applied Statistics), vol. 48, no. 3, pp. 377–394, 1999.
[3] M. Sipko, "Machine learning for the prediction of professional tennis matches," MEng Computing final year project, Imperial College London, London, U.K., Jun. 2015. https://www.doc.ic.ac.uk/teaching/distinguished-projects/2015/m.sipko.pdf (accessed Aug. 12, 2026).
[4] S. A. Kovalchik, "Searching for the GOAT of tennis win prediction," Journal of Quantitative Analysis in Sports, vol. 12, no. 3, pp. 127–138, 2016, doi: 10.1515/jqas-2015-0059.
On 238,716 player-matches from 2018 to 2026, a returning player wins less than their ELO rating expects on a monotone curve: about 0.9 percentage points below expectation after 8-14 days out, 4.2 points after one to two months, and 9.8 points after more than a year, corresponding to a rating overstated by roughly 6, 30 and 73 ELO points. The deficit fades over roughly four to six matches back, which is the signature of a stale rating plus a short-lived rust component.
No, and that was the surprise. Glicko treats inactivity as uncertainty, which pulls predictions toward 50% and therefore predicts a rusty UNDERDOG should overperform their rating. In our data rusty underdogs underperform in every layoff bucket from 15-28 days onward (pooled -1.48pp, favourites -3.93pp). The effect is directional, the rusty side plays worse whether favoured or not, so we encode a penalty rather than an uncertainty term.
Unproven, and we say so in the article. The curve is measured against our own ELO rating, not against betting markets, so it shows that ratings overstate returning players, not that the market does. Encoding it improved our model's Brier score by roughly 0.15 to 0.5 percent relative, a real but small accuracy gain, confirmed out-of-sample by refitting the curve on pre-2025 data and validating on 2025 onward. Until it is tested against closing prices it is model calibration, not a betting edge.
See today's picks: published before the match, graded in public →