The goalie season forecast
One number per goalie, made at season’s end and governing every shot of the next. It starts an unseen goalie below average, and runs generous on seasons cut short: a limit of the question, not a bug.
What the number is
One row per goalie per evidence season N. The row is the forecast made at the end of N, and it governs every shot of N+1 (forecast_for_season = season + 10001, stored rather than re-derived). It is an information-weighted average of the goalie's whole history, each season discounted geometrically by how long ago it was, pulled toward an anchor by a prior. The anchor depends on how many shots he has faced in his career: a goalie nobody has seen starts below average, at −1.40 goals saved per 1,000 shots, and the anchor rises with the volume teams have chosen to give him. A goalie with thin evidence is mostly prior, which is the point and, as the light-workload section shows, also the whole problem.
Until 2026-10-01 the anchor was one constant, set so that the goalies who went on to face 800 shots came out level. That tuned it on the survivors. Out of sample, the forecast ran 4.28 goals per 1,000 too generous on goalies with under 500 career shots, and 0.79 too generous across every goalie it covered. With the anchor fitted on every shot, whether or not the goalie kept the net, those read 1.04 and 0.01, and the first is well inside its error.
A goalie who misses a season keeps his forecast. For the three seasons after his last one, the stage carries his history forward, discounted once more for every season he sits out, so the prior counts for more the longer he is away. Only a goalie with no NHL shots in the last four seasons is priced as an unknown.
A positive number is the better goalie. It is published raw, with no display flip, and points the same way as the talent table it sits under.
It is a rate, with no volume in it. The forecast is a talent per shot faced, and nothing in the table multiplies it by a projected workload. How many shots a goalie faces depends on whether he keeps the net, and that depends on how he plays, which is the thing being forecast. Baking in a projected volume would quietly use the answer. So the consumer supplies volume: multiply the rate by the shots you expect him to face, and state your assumption about his role out loud. gsax_per_shot is the same talent as goals saved above expected per Fenwick faced at league-average shot quality. Neither column is a season total.
The formula, the units and the checks are in the appendix.
The age term
Goalies decline, and the shape of the decline is now a smooth curve rather than a straight line with a break at 30. It rises from −0.037 logits at 20 to a peak of +0.031 at 24, holds roughly level from 28 to 31 (−0.001 to −0.006), then declines to −0.040 at 35 and −0.094 at 40. It is a natural cubic spline with knots at 21, 24, 27, 31, and 35, linear outside that range, ages clipped to 20–40, and centred so a goalie of average age gets roughly zero. There is no break at 30 or anywhere else. In the table the term is applied to the output — ageing is a fact about the next season, not about what the goalie did — and stored separately (age_adjustment and forecast_logit_no_age) so any consumer can switch it off. A goalie with no birth date gets no age term.
Those are the curve's values at full strength, and since 2026-10-01 nobody gets it at full strength. A goalie gets the curve in proportion to how much of his forecast is his own record. The curve is an expected change for a goalie whose level has been measured, and its bump in the early twenties was learned from goalies who were already starters at that age; teams start the good ones young. Handing the same bump to a 24-year-old nobody has seen credits him with a selection he has not passed. Nico Daws, at 24.8 with a prior share of 80.7%, now gets +0.0055 of the curve's +0.0285.
Retired 2026-09-28. The previous version of this term was linear past 30 — AGE_BETA · max(0, age − 30) with AGE_BETA = −0.01054 — chosen because the pre-registered primary estimator cleared its threshold at that size. The spline replaces it: it scores the same out of sample and does not assert a break the data never showed.
Chosen out of sample, and it is a tie on prediction
The curve was fitted by weighted MAE against goalies who returned the following season and judged out of sample, walk-forward, with pre-season precision weights only — the published forecast is for goalies known to be playing, so its target is conditional on survival. Draft pedigree as the shrinkage anchor, a team-change term, a "late ager" interaction, a sliding delta curve, a truncated-imputation aging curve, and a joint performance-and-survival model were all tested and rejected; none beat the shipped curve; most tie it within about one standard error and a few do worse. The anchor is fitted separately, by likelihood on every shot with no volume gate.
Walk-forward MAE among returners: 0.12996 for the new estimator against 0.12972 for the retired one — a difference of +0.00024 with a goalie-clustered standard error of 0.00084. That is a tie. The spline was not chosen for lower error; it was chosen because its shape is defensible where the old break at 30 was not, and it costs nothing on the honest metric. (That comparison is of the curve and the decay; the anchor's own check is the ungated one above.)
What the 800-shot check sees, and what it cannot
Out of sample and gated at 800 shots, the 34-and-up bucket was over-docked by 2.8% while the curve applied in full, and reads 1.0% now that it is scaled. The same gated check reads the goalies with under 1,000 career shots 3.2% over-docked. That is not a new miss but the old one's mirror: the gate keeps only the low-volume goalies whose seasons went well enough to keep the net, which is the group the old anchor was tuned on. The ungated check, which counts every goalie, is the one that speaks for them. The decay is fixed at 0.85 per season of evidence, not fitted: a walk-forward profile from 0.7 to 1.0 was flat, and a freely fitted rate wandered between runs.
Why the age term is in the forecast, not in the walker
The obvious place for an aging effect is the talent walker itself, at the season boundary, where it already applies a goalie aging curve. That placement question predates the spline and was settled against the retired linear term: fit through 2020-21 and score what follows, and the walker's 34+ bucket already read 0.9988 on goals against over expected, and adding drift moved it away from 1.000. It has not been re-run against the spline, but the question the check answers — is there something broken at the walker's boundary right now — has not changed with the curve's shape. So talent_ekf.py and aging_curves.json are still not touched, and this stage still triggers no refit and no cascade.
Precedent, for comparison of approach
Evolving-Hockey's public GSAx, as Luke and Josh Younggren describe it, carries no aging adjustment on the goalie side at all, and McCurdy's Magnus does not either. That is a precedent for leaving an age term out, not a test of leaving it in.
The light-workload miss
The forecast is level overall and about nine per cent generous on the thinnest third of the workload: predicted goals against divided by actual reads 0.911 in the light tercile, 0.976 in the middle and 1.023 at the heavy end. That gap is the most important honest thing on this page.
These ratios were measured against the estimator retired 2026-09-28 and have not been reproduced on the decay-spline estimator shipped since. The mechanism below — a light season is light because it went badly, not because of anything the estimator did — is about workload and selection, not about which estimator produced the number, so it is kept; the exact figures should be treated as approximate until reproduced.
It is censoring, not the estimator. Classify each light season by why it was light. A season that was light by design, a backup on one roster from the first tenth of the schedule to the last, misses by 2.2%. A season that was cut short, by a demotion, a trade, an injury or a hook, misses by 9.9%.
- Light by design (a backup all season)2.2%
- Cut short (demoted, traded, hurt or pulled)9.9%
Those seasons are short because they went badly, and a forecast made before the season cannot condition on a season that never happened. The model is not so much wrong about those goalies as being asked about a set of seasons selected on their own failure.
A goalie who was pulled, demoted, traded or hurt faced fewer shots because his season went badly, and a forecast made before the season cannot know that. Expect the published number to run about 9% light on the thinnest third of workloads.
The evidence that it is censoring
It is not the estimator. Simulate the goals as draws from the model's own probabilities and recut the terciles each draw, and the light bucket comes back at 1.001. The observed 0.911 sits more than nine standard errors from that. It is not a season that starts fine and goes wrong either: on the very first appearance of the season the light tercile already reads 0.934 against 1.052 for the heavy one. Two thirds of the gap is there before a puck is faced.
Read forward instead of backward it is starker still: a goalie who continues in the league misses by 3.3%, one who is demoted by 9.4%, and a goalie whose light season was his last allowed 17% more goals than the forecast said.
The only covariate that closes the gap is a leak. Offset the forecast by the volume the goalie actually faced in the season being predicted and the tercile span collapses from 0.112 to 0.024, so 79% of the gradient is repaired, while the stayer correlation rises 67%. Improving level and prediction at the same time is what a leak looks like in this frame; nothing honest did both. It is the one quantity that would fix the light tercile and it is the outcome itself, so it is kept as a yardstick and will never ship.
What was tried and rejected, and what survives censoring
The honest candidates were run and rejected, and are recorded so nobody re-derives them. Prior workload and a short-season hazard model were both tested walk-forward; the hazard model is the better of the two and closes 0.017 of the 0.112 span for 14% of the stayer correlation. Inverse-probability weighting is not a candidate at all: reweighting changes which seasons the level check averages, not what the forecast says about any one of them.
What survives censoring is small: among by-design backups alone, the three terciles read 0.978 / 0.975 / 1.025. That five-point gradient is ordinary shrinkage, since workload is assigned on talent and a Marcel regresses everyone toward the middle, and it is a fifth the size of the headline miss. A weaker prior recovers part of it at a cost to the calibration slope, which is not a trade worth making now.
The tercile figures come from the study run on the forecast before the age term, where the gated overall level is 1.005; the shipped table, with the term, reads 1.0097 on the same gate. The direction and the diagnosis are unchanged.
Where it comes from, and what checks it
The stage is scraper/compute_goalie_forecast.py and it writes one table, goalie_season_forecast. Nothing upstream reads it; the team projections, the lineup card and the goalie card do. It runs in the cascade immediately after compute_net_rating, so every stage in one run shares one set of season boundaries and one xG vintage, and it costs about thirty seconds.
It is a re-derivation, and the re-derivation is checked. On the same 1,531 goalie-seasons, the production stage and the research estimator agree to a maximum absolute difference of 1.4 × 10⁻⁴ logits, median 1.4 × 10⁻⁵, which is 1.2% of one materiality unit. The stage raises and writes nothing when it produces fewer than 500 rows, when any forecast is non-finite, or when the gated or the ungated level ratio is missing or outside [0.95, 1.05]. The level check is read twice: gated at 800 shots faced, the floor this project uses everywhere else, and ungated, by career volume and by age, because the gate selects on the outcome.
Full methods: backend/docs/goalie-forecast.md; the age term in research/goalie_context/step1_age.md; the censoring study in research/goalie_context/lightload_level.md.
One standing caveat. The forecast consumes the talent walker's own output, shooter-adjusted xG, as its evidence. Refit the xG model, or change the walker's priors, variances, aging or centring, and every number on this page moves, because the evidence underneath it moved. Nothing upstream depends on the stage, but plenty cascades into it: it is rerun after any cascade that touched compute_xg, and a forecast read against a stale xG is a stale forecast.
The estimator and the checks
Reference material. Each drawer opens on its own.
The estimator and its units
The estimator is the goalie's whole history on the production talent walker's own per-season evidence, each season weighted by its Fisher information and discounted geometrically by how long ago it was, all of it pulled toward an anchor set by career volume, plus a smooth age effect scaled by the data's share:
mu(V) = MU_ENTRANT + MU_VOL_SLOPE·ln(1 + V/1000)
pred_no_age = (Σ_k d^k·I_(N-k)·o_(N-k) + P0·mu(V))
--------------------------------------
Σ_k d^k·I_(N-k) + P0
var = 1 / (Σ_k d^k·I_(N-k) + P0)
s = P0 / (Σ_k d^k·I_(N-k) + P0)
pred = pred_no_age + (1 − s)·f(age)o is a season's own one-step estimate of the goalie term and I its information, summed over every season of the goalie's history with k counted in seasons elapsed — a season he missed discounts his older evidence just like a played one. The decay d is fixed at 0.85, not fitted: a walk-forward profile was flat from 0.7 to 1.0, and a freely fitted rate wandered between runs. The prior precision is P0 = 329.0099, toward an anchor that depends on V, the goalie's career shots: MU_ENTRANT = −0.022844 and MU_VOL_SLOPE = 0.009883, both fitted by shot log-likelihood on every shot with no volume gate. The anchor at zero is also the forecast for a goalie with no row at all. It replaced a single anchor of mu0 = 0.010673, which had been re-set to put the 800-shot gated level at 1.000. f is the age spline described above, and s is the prior's share. Each season's evidence is centred on that season's information-weighted league mean, because the walker re-centres the league at every season boundary; an uncentred cross-season comparison would read a frame move as talent.
The sign. forecast_logit is g in logit p(goal) = pre − g, where pre is the model's logit with the goalie term removed. Subtracting g lowers the goal probability, so a positive number is the better goalie. Scale: one per cent of xG(sv) is 0.01141 logits, and one tier is 0.08926.
The evidence gate and the stage's guards
The evidence gate is events.is_fenwick_for_xg, a known goalie, and a final game from the model's first season on, regular season and playoffs, which is what the walker itself consumes. A narrower regular-season mask exists and is used only for the prediction correlations, never for the evidence. The stage reads events, games and players.birth_date, and nothing else.
The research estimator reads cached frames that do not exist in production; everything the production stage does is rebuilt from the database. On identical inputs the two agree to float noise, pinned by a test at 1e-9. The residual is a vintage difference: the research cache's shooter-adjusted xG came from a different compute_xg run than the values in events today.
The stage raises and writes nothing when it produces zero rows or fewer than 500, when any forecast is non-finite, when the gated or the ungated level ratio is missing or outside [0.95, 1.05], or when the row count after the write does not match what was computed. A full corpus is about 2,245 rows, gap rows included. It also asserts that every input was knowable before the season being forecast: no row carries a lag or a career term it could not have earned, and the season key steps forward rather than back.
The ungated check is pooled across seasons, not a per-season target: a per-season ungated target drifts with the league's exit rate and would read as a model regression when nothing had regressed. Beside the gated check, the excluded seasons are reported rather than hidden.