Sign inStart free

Product

Contender keeps the input, evidence, estimate, limitations, and recorded outcome together so a polished number never outruns its source.

Start freeInspect the replay
InputEvidenceEstimateSnapshot

Historical replay

Historical example — replayed as of noon UTC on game day; not a live pick.

Historical exampleGame Mar 15, 2026 · as of 12:00 UTC that dayHistorical example — replayed as of noon UTC on game day; not a live pick.

Shai Gilgeous-Alexander

OKC vs MIN Timberwolves · Mar 15, 2026

Selection

Over 30.5

Points

Projection

36.0

+5.5 vs line

P(over)

86.6%

Heuristic, uncalibrated

Rating

5/5 · High

Last 10 games · PTS

ML + Rules (Random Forest)

010213141Jan 27 vs NOP · 36 min · 29 PTS · line 30.5 · loss29LJan 27Jan 29 @ MIN · 31 min · 30 PTS · line 30.5 · loss30LJan 29Feb 1 @ DEN · 33 min · 34 PTS · line 30.5 · win34WFeb 1Feb 3 vs ORL · 28 min · 20 PTS · line 30.5 · loss20LFeb 3Feb 27 vs DEN · 34 min · 36 PTS · line 30.5 · win36WFeb 27Mar 1 @ DAL · 33 min · 30 PTS · line 30.5 · loss30LMar 1Mar 4 @ NYK · 35 min · 26 PTS · line 30.5 · loss26LMar 4Mar 7 vs GSW · 35 min · 27 PTS · line 30.5 · loss27LMar 7Mar 9 vs DEN · 39 min · 35 PTS · line 30.5 · win35WMar 9Mar 12 vs BOS · 39 min · 35 PTS · line 30.5 · win35WMar 12Line 30.5Proj 36.0
Observed (PTS) Selected line ProjectionW / L / P = over result vs line

Over 30.5 hit rate by window

  • Last 5

    40%

    2–3 of 5

  • Last 10

    40%

    4–6 of 10

  • Last 15

    47%

    7–8 of 15

  • Last 20

    45%

    9–11 of 20

  • This season

    51%

    28–27 of 55

  • vs MIN

    71%

    5–2 of 7

hit_rate = wins/eligible; decisive_rate = wins/(wins+losses); push_rate = pushes/eligible; no_loss_rate = (wins+pushes)/eligible. Historical rates are not future win probabilities.

Why the model leans this way

  • Last 5 average is +0.1 vs line
  • Opponent allowance rank: 17
  • Projected minutes are +2.4 vs season average
Full data for these gamesShow
Recent games with minutes, PTS value, the 30.5 line and the selected-side result
DateOpponentMinPTSLineResult
Jan 27vs NOP362930.5loss
Jan 29@ MIN313030.5loss
Feb 1@ DEN333430.5win
Feb 3vs ORL282030.5loss
Feb 27vs DEN343630.5win
Mar 1@ DAL333030.5loss
Mar 4@ NYK352630.5loss
Mar 7vs GSW352730.5loss
Mar 9vs DEN393530.5win
Mar 12vs BOS393530.5win

normalized NBA store (stats.nba.com snapshots via Phase 3 ingestion). Seasons 2024-25, 2025-26; latest eligible game Mar 12, 2026. Not available for this example: legacy engine does not expose the H2H multiplier separately; no quote entered; NBA stats endpoints carry no prices; no active model artifact served (no_registry).

Actual 20 PTS · miss

Graded against the stored box score after the game.

Capability index

01

Projection

A forward estimate with its contributing methods and interval.

02

History

Selected-side outcomes across recent, season, and head-to-head windows.

03

Probability

An explicitly labeled estimate, separate from observed hit rates.

04

Matchup

Opponent allowance, rank, trend, venue, rest, and pace context.

05

Explanation

Readable reasons, risks, unavailable fields, and fallback behavior.

06

Provenance

Source, seasons, freshness, engine version, and immutable history.

History that does not drift

Saved results are immutable server snapshots. Reopening history does not silently recompute the old analysis against newer data.

Open the analyzer