StatsNext

The StatsNext Evidence Engine

The Evidence Engine reads one exact selection (one player or team, one market, one line, Over or Under) and answers four questions from StatsNext’s existing research: what supports it, what argues against it, how the available price compares with the StatsNext estimate, and how reliable the reading is. It never produces a score or a pick, and it says so when the evidence is mixed or too thin.

What counts as evidence

Each finding comes from a research tool, read the same way for every player and market:

  • Results at the line: the last 10 games and this season, with the games behind each.
  • Opponent and venue: results against this opponent and at this venue.
  • Opportunity: the opportunity stat for the market (targets, carries, pass attempts, minutes, ice time), recent games against the season.
  • Matchup: a defense that allows a lot of the stat supports the over and one that allows little argues against it, and the other way round for the under.
  • Availability and role: a teammate’s absence or a role change, when the comparison is meaningful.

Over and Under are evaluated separately; switching sides moves each finding to the other column.

Overlapping facts are not separate confirmation

The last 10 games and the season share many of the same games, so when they agree they become one finding. When findings for one side rest on the same recent games, the reliability notes say so.

Yes-only markets

For a market priced only as yes (anytime touchdown, one or more home runs), a low hit rate is not automatically bad: a long-priced event can win 30% of the time and still be priced at 20%. StatsNext compares the yes rate with the price’s implied probability, and says so when the sample is too small to tell them apart.

Price interpretation

The price reading compares StatsNext Intelligence with the selection’s actual price and line: favorable relative to the estimate, unfavorable, or approximately aligned. When no sportsbook prices the line, or StatsNext has no model for the market, it says the price is not assessable. Line movement and sportsbook disagreement are shown as market context, already reflected in the price, and are never counted as support. The odds comparison guide explains the probabilities behind this.

Reliability

Reliability is stated in two parts that are never combined:

  • The evidence: leans toward the selection, leans against it, mixed, no clear evidence, or insufficient data, with notes on sample size, overlapping findings, uncertain injury designations and missing data.
  • The model’s input quality (Strong, Mixed or Limited): how much data stands behind the estimate.
Example from StatsNext, captured October 11, 2026 (Bills at Rams, October 12)

Dalton Kincaid receptions Over 4.5. Argues against: over in 3 of the last 10 games and 2 of the last 8 road games; the Rams had allowed tight ends 3 receptions per game, the 2nd fewest of 32 (league average 5.3). Supports: nothing verified. Reading: evidence leans against Over 4.5. Switching to Under 4.5 moved the same three facts to the supporting side.

What it never does

  • It never gives a score, a rank number or a pick.
  • It never counts the model estimate as supporting or contradicting evidence; the estimate is used only to read the price.
  • It never uses full-game statistics for a quarter or half market.
  • It never fills in missing statistics, such as targets in college football.
  • It never hides contradicting evidence to make a selection look better.

Where it appears

The full Evidence section sits near the top of Player Prop Research, for the side and line you choose, with links to the research behind each finding. In Explore, each Trends result carries a one-line Evidence summary, and the list can be opened on the strongest evidence first. The comparison with the StatsNext estimate is part of Full Research; until you open a player, the summary says it is compared there.

StatsNext is a research platform, not a sportsbook; it does not accept wagers or guarantee results (Research Disclaimer, Responsible Gambling).

What is free

The one-line Evidence summary on every Trends result is free; the full Evidence section, including the price comparison, is part of Full Research. Full Research for a player, team or game opens 0 times in total as a guest, 5 new subjects a day with a free account, and without limit with StatsNext Pro (US$19.99 a month or US$179.99 a year). See Pricing.

Continue in StatsNext

Trends

Patterns with an Evidence summary on every result.

Props

Open any prop for Player Prop Research and its Evidence section.

Explore

Every research tool in one place.

Research Disclaimer

What StatsNext estimates are and are not.

Free research pages for this topic

Built from StatsNext’s stored box scores and refreshed every morning: