How the Scout thinks

One model behind every number. Here it is, opened up.

◎ Start here — five steps

  1. Find a steady player. Sort the board by Last 10.
  2. Read the bars. Green cleared the line; red or grey did not.
  3. Check the opponent. H2H lists past meetings, with the sample size.
  4. Check the matchup. The ring grades tonight's opponent on that stat: A+ = easiest, F = toughest.
  5. Check playing time. A player who might sit is a pass.

◈ What PropScope is

A research tool for player props — US leagues, esports and tennis — across the biggest US apps.

One model gives every line a projection, a chance and a grade.

  • Call of Duty — kills on maps 1–3 of a CDL series, or on one map.
  • League of Legends — kills, assists and creep score on maps 1–2.
  • Tennis — aces, double faults, games and breaks, ATP and WTA.

Demo build: synthetic players, lines and votes. Not affiliated with any app or league.

◇ Goblins & Demons

Goblin 25.5— lower line: easier to clear, pays less.
Demon 33.5— higher line: harder to clear, pays more.

Both are Over only. Every offered line is priced at its own model chance.

O Leans and confidence

OVER LEANP(Over) ≥ 57%
UNDER LEANP(Over) ≤ 43%
◆ NO CLEAR EDGEin between: a coin flip
✕ AVOIDplaying time unsure, or under 6 games
  • High — 12+ points from 50%, steady minutes.
  • Medium — 6+ points from 50%.
  • Low — everything else, and every Avoid.

♥ Responsible play

Estimates, never guarantees. A 65% call still loses 35 times in 100.

Set limits first. Never stake more than you can afford to lose.

18+ (21+ in some places). Help: 1-800-GAMBLER (US), GamCare (UK), Gambling Help Online (AU).

≡ Every market, in plain words

What each line asks, and tonight's typical number.

    A Glossary

    Every word on our pages, one line each.

      ƒ The Scout — the projection model

      One formula, then a probability. A worked example inside.
      projected value = season rate per full game × (expected playing time ÷ a full game) × opponent factor × venue factor × form factor
      • Season rate per full game — per 90 minutes in soccer, per 48 in the NBA.
      • Expected playing time ÷ a full game — scales the rate to tonight's workload.
      • Opponent factor — what this opponent allows, vs a league average of 1.00. For saves and similar work, it is how much work they create.
      • Venue factor — a small home boost (1–6%), mirrored on the road. None on neutral ground (esports, tennis).
      • Form factor — the last 5 games, blended in at 25%.

      A full game, per sport:

      SportA full gamePlaying time measured as
      Soccer90 minutesminutes played
      NBA48 minutesminutes played
      NFLevery offensive snapsnap share
      MLBa starter's 9 innings; a hitter's 4 or 5 plate appearancesinnings pitched, or plate appearances
      NHL60 minutes of ice timeice time
      WNBA40 minutesminutes played
      Call of Dutymaps 1–3 of the seriesmaps in the line
      League of Legendsmaps 1–2 of the seriesmaps in the line
      Tennisone set, times the sets expectedsets

      Count stats use a Poisson curve; volume stats a Normal one.

      Every chance is capped at 2–98%. The model is never certain.

      Worked example — from today's data:

      ◧ Hit rates & the blended score

      Every rate shows its sample. Missed games never count.

      Every hit rate shows its sample: "4 of 5" is not "40 of 50".

      Games a player missed are left out. They show as DNP — never a miss.

      blended score = 0.6 × model probability + 0.4 × historical hit rate  ·  history = 0.5 × L10 + 0.3 × L5 + 0.2 × venue split

      Both halves face the model's side. When they disagree, the score drops.

      Σ Fantasy scoring — app by app

      The scoring tables this build prices against.

      Soccer stat lines follow Opta's event definitions.

      Scoring charts change. Check your app before relying on a weight.

      ⇄ From demo to real data

      Static files today; a licensed data feed next.

      Today: static files, computed in your browser. No server, no tracking.

      Real lines will come from a licensed feed, never scraping.

      • Real rosters and fixtures.
      • Auto-graded pick history, wins and losses alike.
      • Closing-line tracking.
      • Line-move alerts on saved props.
      • Real accounts.

      ? Frequently asked questions

      Is this betting advice?

      No. It is a research tool: estimates, not recommendations. Your call, always.

      What does "1 = toughest" mean?

      Rank 1 allows the least of that stat. The highest number allows the most.

      Why does the model disagree with the crowd?

      The crowd reads stories; the model reads data. We show both.

      Where do the lines come from?

      Here they are synthetic. A real version uses a licensed feed.

      Can I trust the hit rates?

      Each shows its sample. Missed games are left out. Thin samples get ✕ Avoid.

      What is free?

      Everything except Auto-Build and Line Gaps. No price for signed-out visitors.

      Open today's board 🎯