FPL Bookies XI - Odds-Blended Best XI for GW36
Bookmaker market data (SBOBET totals + Asian handicap) flows into the xPts engine for every defender, GK, and clean-sheet pick. GW36 2025/26.
What the Bookies XI is built from
Two bookmaker markets per fixture, refreshed up to hourly during deadline windows.
- Totals (over/under goals) - implies the bookmaker's expected total goals for the fixture
- Asian handicap - implies the expected goal differential between home and away
- Poisson solver - converts those two market lines into per-team lambdas (λ_home, λ_away)
- Derived probabilities - clean sheet (e^(-λ_opp)), match win/draw/loss, BTTS, over 2.5
- xPts blend - 65% market / 35% stats weighted average for defensive-side projection
- Captain Accuracy 2.0 - dual-tracks blended vs unblended picks for a public scoreboard
Where to see it live
| Surface | What it shows | Where |
|---|---|---|
| Odds page | Per-fixture lambda bars, CS pips, W/D/L %, scoreline grid, 7-day time-series | https://www.fpldave.com/odds |
| Fixtures > Bookies tab | Same odds grid embedded inside the Fixtures page sub-nav | https://www.fpldave.com/fixtures |
| Sharp Money strip on Home | Top 3 biggest 24h CS-probability movers | https://www.fpldave.com/ |
| Captain Accuracy 2.0 receipt | Blended pick hit-rate vs pure-stats baseline (after 5+ GWs) | Home > Now scoreboard |
| Engine internals | Captain Pick, Transfer Recommender, AI Coach, FDR Projections, planner xPts | Every existing analytics surface, silently upgraded |
The math, briefly
Given the bookmaker's totals line T and home-perspective Asian handicap H, we solve:
λ_home = (T - H) / 2 λ_away = (T + H) / 2 P(home CS) = e^(-λ_away) P(away CS) = e^(-λ_home) P(home win) = Σ over (i > j) of P(home=i) · P(away=j)
Inside the xPts engine, opponent-goals projections are blended with the stats-derived value at 65% market weight (configurable, calibrated against historical captain accuracy). Player attacking xG share stays stats-driven - bookmaker totals are team-level signal, not player-level.
Why bookmaker odds add information
Three things sharp markets price in that pure-stats models miss:
- Team news - injuries, suspensions, rotation rumours flagged hours before official confirmation
- Lineup leaks - probable XIs often appear in betting markets 24-48h before press conferences
- Sharp money flow - movement from people who know things you don't (insiders, professional bettors)
None of which the FPL form / xG dataset will catch in time. Bookmaker odds are how markets aggregate dispersed private information into a single number.
The FAQ
What is FPL Dave's Bookies XI?
The Bookies XI is FPL Dave's odds-blended best XI — each player's expected points are computed using a Poisson model fed by SBOBET's over/under totals and Asian handicap markets, not just season-form stats. This captures team-news, lineup leaks, and sharp-money signals the pure-stats model misses, especially for defensive picks where clean-sheet probability matters most.
How does FPL Dave use bookmaker odds?
The Odds API streams SBOBET's totals + handicap lines for every Premier League fixture every 6 hours (more often near deadlines). We solve a Poisson regression for each fixture's home/away expected goals (λ_home, λ_away), then blend those values 65% market / 35% stats inside the xPts engine. The blend affects clean-sheet probability, GK saves, and goals-conceded penalties; the attacking side (player xG share) stays stats-driven because totals speak to team-level xG, not player distribution.
Which bookmaker do you use and why SBOBET?
Primary source is SBOBET via The Odds API. SBOBET is the canonical reference for sharp betting markets (matches the same source FPL Joe uses) and sits inside The Odds API's "au" region. The endpoint preference cascades sbobet → pinnacle → marathonbet → any bookmaker offering both totals + spreads, so even when SBOBET's line is missing for a fixture we fall back to the next-sharpest book.
How does this compare to pure-stats projections?
Captain Accuracy 2.0 tracks both picks per gameweek — the blended pick (with bookmaker odds) and the unblended pick (pure-stats Poisson). After 5+ GWs of data the Home > Now dashboard publishes the A/B receipt, e.g. "Bookies-blended pick beat the popular captain 64% of GWs vs 51% for pure stats." The dual track means we hold ourselves to a public scoreboard rather than just claiming better picks.
When is the data refreshed?
A cron-job.org schedule fires every hour against /api/odds?action=snapshot. The endpoint self-throttles: 6-hourly baseline cadence outside hot windows, hourly within T-24h of a deadline and through T+3h after kickoff (when team news settles). Total Odds API usage stays around 400 credits/month, inside the 500-credit free-tier cap.
What is the 'Sharp Money' strip on the Home page?
Live widget on Home > Now showing the biggest 24h clean-sheet probability movers. "Saliba CS odds shortened 22% in 24h" tells you sharp money sees a clean sheet — a leading indicator of mass transfers in. Drill through to the /odds page for the full fixture grid + per-fixture time-series.
Is this gambling content?
No. FPL Dave displays odds as inputs to an analytics model — there's no affiliate link, no "place this bet" button, no sportsbook redirect. The page includes a BeGambleAware footer link as a courtesy. If you want to actually place bets, that's outside our service.
How is the Bookies XI different from the AI Coach captain pick?
The Captain Pick and the AI Coach already use the bookmaker-blended xPts engine internally (the blend ships engine-wide via the useOddsBlend hook). The Bookies XI page is the public-facing explainer — what powers those picks, where the data comes from, and how transparently we score our accuracy. Same engine, different surface.
Odds data is for informational purposes only. 18+ · BeGambleAware · @FPLDaveHQ