Stats glossary

Gross score on a hole minus that hole’s par. Negative is below par, zero is on par, positive is over par. The atomic unit every other stat is built from.

Formula

score_to_par = gross − par

Worked example

At the reference round (Wycombe Heights, 3 Sept 2023), James scored 2 on the par-3 hole 6: 2 − 3 = −1. That’s a birdie in the six-category classification.

Edge cases

• A missing hole score is skipped — not zeroed. Downstream averages treat the hole as not played by that player.
• Hole played by a guest outside the three-player group is ignored entirely (see ingestion rule 6).

Total gross for a round, expressed as a signed integer versus course par. The standard golf way to describe how the day went.

Formula

round_vs_par = Σ score_to_par(hole) = (Σ gross) − course_par

Worked example

Reference round: Matty 98 vs par 69 → +29. Matt 102 → +33. James 92 → +23. Written with a leading sign always — +23, E, −4.

Edge cases

• Rounds with any missing per-hole score are flagged as partial. Their vs-par is computed only over the holes actually played and excluded from stats that assume 18 holes (best / worst / personal course record / front-back split).
• If the group plays different tees at the same course, par comes from the round’s captured tee — it’s unambiguous round by round.

Maps each hole’s score-to-par into exactly one of six buckets. Shape carries the signal on the scorecard; colour reinforces.

Formula

See the table below.

Worked example

Reference round, hole 3 (par 5). Matt scored 7 → +2 = Double. Matty 5 → 0 = Par. James 7 → +2 = Double. Hole 6 (par 3), James scored 2 → −1 = Birdie.

Edge cases

• Albatross (≤ −3) buckets as Eagle — we don’t split the ≤ −2 tail.
• Hole-in-one on a par 3 is an Eagle by this classification. Whether to celebrate it as distinct is a presentation-layer concern, not a seventh category.

The total number of rounds in the database that the player participated in.

Formula

rounds_played(player) = |{ round : player ∈ round.players }|

Worked example

After the 11-round backfill, every player has rounds_played = 11. All rounds are 3-player rounds by design — Matt, Matty, James in every one.

Edge cases

• A round with fewer than 18 holes recorded still counts as one round played.
• Guest rows (e.g. Olly on scorecard 4) do not count — the stat is only computed for the three group players.

Three counts per player: rounds won (strictly lowest gross), tied for low, and lost (otherwise). The leaderboard ranks primarily by wins desc, mean-round-vs-par asc as tiebreak.

Formula

For each round the player was in: outcome = W if strictly lowest, T if tied for low, L otherwise. Aggregate across rounds.

Worked example

After 11 rounds: James 10·0·1 (dominant), Matt 1·0·10 (his one win at Hazlemere), Matty 0·0·11.

Edge cases

• Ties at the top of the leaderboard on both wins and mean-vs-par produce a shared rank (=1, =2, etc.). No further tiebreak.
• Only 3-player rounds are registered (CLAUDE.md §2). Two-player variants never reach the stats engine.

Arithmetic mean of the gross totals across every round the player has played. Context, not ranking — gross is course-difficulty-sensitive.

Formula

avg_gross(player) = (Σ round.gross[player]) / rounds_played(player)

Worked example

After 11 rounds: James 91.5 · Matt 101.8 · Matty 99.4.

Edge cases

• Partial rounds’ grosses are flagged and reported alongside the round count.
• rounds_played = 0 renders as an em-dash.

Arithmetic mean of round-vs-par values. The most meaningful single “how good is this player” number in the system — normalises across courses of different difficulty.

Formula

mean_vs_par(player) = (Σ round_vs_par(player)) / rounds_played(player)

Worked example

After 11 rounds: James +21.0, Matty +28.8, Matt +31.3. Signed, one decimal place. Tiebreak on the leaderboard if two players share a win count.

Edge cases

• Partial rounds contribute their actual vs-par (not a 18-hole projection) and get flagged.
• Don’t confuse with “average hole vs par” — that’s a per-hole mean (see Per-par averages).

The middle round. Robust to one truly terrible round dragging the mean upward.

Formula

Standard median over { round_vs_par(player) : round }. Even counts use the mean of the two middle values. One decimal place.

Worked example

If mean is meaningfully higher than median, blow-ups are inflating the mean — the median is the steadier read on “how they usually play.”

Edge cases

• With a single round, median = mean (both equal that round).
• rounds_played = 0 renders as an em-dash.

Per player, the round with the lowest (best) and highest (worst) round-vs-par. Vs-par is the headline; gross + course + date is the secondary context.

Formula

best_round = argmin round_vs_par(player, r). worst_round = argmax. Ties break to the more recent date, then to lower gross.

Worked example

James’s best: +15 at Huntswood on 8 Mar 2026. Worst: +29 at Wentworth (Edinburgh), 2 Oct 2023. Matt’s best: +23 at Hazlemere (the day he won the round). Worst: +43 at Wentworth.

Edge cases

• Partial rounds are excluded — their vs-par isn’t comparable.
• With exactly one round, best = worst — the profile collapses to a single “only round so far” treatment.

The lowest round-vs-par a player has shot at a given (course, tees) combination. Different tees at the same course are separate records.

Formula

pcr(player, course, tees) = min round_vs_par over rounds at that (course, tees). Ties resolve to the earliest round that set the record.

Worked example

At Old Thorns (yellow, 2 rounds), James’s PCR is 94 (+22) on 3 Dec 2023 — lower than his 97 the day before.

Edge cases

• First round at a course seeds the record automatically; the UI labels this rather than showing a “record broken” framing.
• Ties resolve to the earliest round — later ties are record-equalled, not record-broken.
• Partial rounds excluded.

Count and rate of holes a player has made in each of the six score-to-par categories. Both counts and rates are always shown — rates normalise across players with different round counts.

Formula

For each category c: count(player, c) = number of holes in c; rate = count / total holes played. Rate to one decimal.

Worked example

Reference round, James (18 holes): 0 eagle / 2 birdie / 3 par / 4 bogey / 6 double / 3 triple+. Across the 11-round backfill his lifetime bogey rate is just under 40%.

Edge cases

• Categories with zero count still render (0 · 0.0%) — consistency of the six-column shape matters.
• Missing holes contribute nothing to counts or denominators.

Match-play record of every hole the player has played vs the other two group players in the same round. W = strictly lowest, H = tied for low, L = otherwise. Shown as counts and rates.

Formula

For each hole: low = min(gross[q] for q in group). If player == low and only one at low → W; if player == low and multiple tied → H; otherwise → L.

Worked example

Reference round: James 5·4·9, Matt 3·4·11, Matty 4·4·10. The three counts sum to 18 per player; across all three players, W + H + L = 54 player-hole outcomes.

Edge cases

• A hole where the player has no recorded score but others do contributes nothing to that player’s W/H/L.
• Only 3-player rounds — the stat has no 2-player semantics.

Pairwise match-play record A vs B across every hole both played together. Shown overall and split by par type (3 / 4 / 5).

Formula

For each hole: if gross[A] < gross[B] then A_win += 1. If equal → halved. If greater → A_loss. Aggregate overall and filtered by hole.par ∈ {3, 4, 5}.

Worked example

Reference round, Matt vs James: 6 won · 4 halved · 8 lost overall (Matt lost the round by 10 strokes but won 6 individual holes — two separate stories).

Edge cases

• An opponent the player has never played with renders as an em-dash, not 0–0–0 — the “no data yet” signal is different from “all drawn”.
• A vs B and B vs A are mirror images: A.W = B.L, halves are shared.

For each par type (3, 4, 5), the mean of score-to-par across all holes of that par the player has played. Shown with the sample size (number of holes).

Formula

avg_par(player, p) = mean(score_to_par) restricted to holes where hole.par == p

Worked example

Reference round alone, James on par 3s (5 holes): scores 4, 2, 3, 4, 6 → diffs +1, −1, 0, +1, +3 → sum +4 → avg +0.8.

Edge cases

• Par types with zero samples render as an em-dash.
• No par-6 handling — no par-6 holes appear in the current course set.

Mean score-to-par bucketed by the hole’s stroke index: Hard (SI 1–6), Mid (SI 7–12), Easy (SI 13–18). Captures whether a player disproportionately bleeds on the hard holes.

Formula

Three buckets over six stroke indexes each. avg = mean(score_to_par) within the bucket; samples = 6 holes × rounds played.

Worked example

Across 11 rounds, each bucket has 66 samples per player. Hard-bucket averages run 0.5–1 stroke higher than Easy-bucket averages — the stroke index does its job.

Edge cases

• Courses with non-standard SI assignments (ties, missing values) still bucket cleanly by SI number.
• Empty buckets (zero rounds played) render as em-dash.

Mean score-to-par on the front 9 vs the back 9. Makes late-round fatigue or front-nine slow starts visible. Delta (back − front) is shown below.

Formula

avg_front = mean(score_to_par for hole index 0–8). avg_back = mean(score_to_par for hole index 9–17). delta = avg_back − avg_front.

Worked example

Across 11 rounds, each side has 99 samples per player — the same rounds, split in half. A positive delta means the player plays worse on the back nine on average.

Edge cases

• Partial rounds missing holes in either nine still contribute their played holes; the denominators narrow accordingly.
• Unused holes on 9-hole variants would break this stat — the data model doesn’t accept them.

After a bogey or worse, what share of the next holes score par or better. Resilience after a bad hole — does the damage stop at one?

Formula

Trigger holes = any hole scored bogey+ (+1 or worse). Hit = the next hole (same round) scored par or better. Rate = hits / triggers.

Worked example

Across 11 rounds James’s bounce-back rate is around the mid 30s. Translated: after a bogey or worse, he plays the next hole to par or better about one in three times.

Edge cases

• Triggers that sit on the last hole of a round are skipped — no “next hole” within the round.
• “Next” does not cross round boundaries — this is within-round resilience only.
• rounds_played = 0 or no bogeys yet renders as em-dash with caption “Not enough bogeys yet”.

After holding par or better on a par-4 or par-5, what share of the next holes score double bogey or worse. Damage after a steady hole — the blow-up we all fear.

Formula

Trigger holes = par-4 or par-5 holes scored par or better. Hit = the next hole scored double+ (+2 or worse). Rate = hits / triggers.

Worked example

Typical values in the backfill run low-teens percent. A high rate means the player rarely consolidates a good hole; a low rate means the good hole often leads to another one.

Edge cases

• Last hole of a round is skipped as a trigger.
• Limited to par-4+ triggers deliberately — par-3 par is noise rather than signal.
• No triggers yet renders as em-dash with caption “Not enough par-or-better yet”.

Mean vs par across the last 5 rounds, compared to lifetime. Captures whether the player is trending up or down recently.

Formula

Last-5 rounds by round_date (most recent first, reversed for the sparkline). recent_mean = mean(round_vs_par) over those. Delta = recent_mean − lifetime_mean.

Worked example

If a player’s recent 5 rounds average +24.0 while their lifetime mean is +28.5, they’re playing 4.5 strokes better per round lately. Down is an improving trend.

Edge cases

• With fewer than 5 rounds, the stat uses however many exist (and the sparkline has fewer points).
• With zero rounds, renders as em-dash.

For a given opponent pair (A, B), A’s biggest win margin and A’s narrowest loss margin across every round both played. Margin is the gross differential.

Formula

margin = gross[B] − gross[A]. Biggest win = max positive margin (A beat B by the most). Narrowest loss = the loss margin closest to zero (least negative).

Worked example

James vs Matt across 11 rounds: biggest win +17 (Huntswood, 8 Mar 2026 — James 85, Matt 110). Narrowest loss −2 (Hazlemere — Matt 93, James 95, Matt’s only win).

Edge cases

• An opponent against whom the player has never lost renders Narrowest loss as em-dash.
• An opponent never beaten renders Biggest win as em-dash.
• Ties on margin resolve to the earliest round by date.