Methodology

Every Diamond DFS projection is a transparent, weighted sum of public inputs. Nothing is a black box — you can tune every weight, see every input, and rebuild the projection from scratch.

Inputs

• Hitter rate stats — singles, doubles, triples, HR, BB, HBP, SB, R, and RBI per plate appearance. Season totals blended with last-14-day form (~35% weight on L14) and Empirical-Bayes shrunk toward league average for small samples. Source: MLB Stats API.
• Expected PA — lineup-slot table (slot 1 ≈ 4.6 PA, slot 9 ≈ 3.7 PA) scaled by the team's Vegas implied total. Pulled from official starting lineups on game day.
• Pitcher rate stats & role — K/IP, BB/IP, H/IP, ER/IP, HBP/IP from season totals. Recent gamelog usage classifies role (SP, Opener, Bulk, RP) and sets projected IP. Source: MLB Stats API.
• vs SP composite — pitcher dominance (K rate, last-30d WHIP, BAA, home/away, rest, quality starts) blended with opposing team vulnerability (team K%, on-base quality, BA L14, R/G L14, lineup strength). Hitter side adds platoon hand-OPS split and a K-rate mismatch adjustment.
• Park factor — multi-year runs, HR, and hits factors per venue. Coors boosts hitter ceilings; Petco suppresses them. See the full parks table.
• Weather — temperature, wind speed and direction relative to home plate, plus dome, turf, and day/night flags. Wind blowing out adds materially to HR expectation; domes are flat-flagged. Source: OpenWeather.
• Vegas implied team total — derived from moneyline and over/under, refreshed when lines move. Drives the team-total factor and feeds expected PA + pitcher win probability. Source: The Odds API.

Projection formula (simplified)

hitter_points =
    Σ over events {singles, doubles, triples, HR, BB, HBP, SB, R, RBI} of
      rate_per_PA(event)
    × expected_PA(lineup_slot, team_total)
    × event_multiplier(vsSP, recentTrend, parkFactor,
                       weather, teamTotal, environmentalSplits)
    × DK_points(event)

pitcher_points =
    Σ over events {K, ER, H, BB, IP} of
      rate_per_IP(event)
    × projected_IP(role: SP | Opener | Bulk | RP)
    × event_multiplier(vsSP, recentTrend, parkFactor,
                       weather, teamTotal)
    × DK_points(event)
  + win_probability(team_total, opp_total, isHome) × W_bonus

Weights

The six factors (vs SP, recent trend, park, weather, team total, environmental splits) are controlled by integer sliders that sum to a budget. Each factor's effective impact is its slider value divided by the total — moving one slider up automatically pulls impact away from the others. Defaults give every factor a roughly equal share; users can rebalance to lean harder on weather, downweight park, or change how aggressively the SP matchup fires.

NFL projections

Football scores differently than baseball, so the NFL model uses six football-native factors in the same slider machinery. See the NFL DFS pages for the weekly output.

• Opportunity — snap share, target share, carries, and red-zone touches. Volume, not efficiency, is the largest driver of NFL fantasy scoring.
• Matchup — the opposing defense's fantasy points allowed per game to that position, adjusted for pace and pass-rate tendency.
• Game environment — Vegas implied team total and spread. The spread sets game script: favorites run, underdogs throw.
• Recent form — the last three weeks against the season baseline, shrunk toward the baseline for small samples.
• Weather — wind, precipitation, and temperature for outdoor stadiums only; domes and closed roofs play neutral.
• Efficiency — yards per target and per carry with touchdown rate regressed toward stable rates.

Playing time is modeled explicitly: depth-chart rank, injury designation, and salary rank combine into an expected share of a full starter's workload, so a backup is promoted automatically when the man ahead of him is ruled out. Projections are then scored with DraftKings full-PPR or FanDuel scoring, including each site's bonuses, and with the 1.5x Captain or 2x MVP multiplier in single-game formats.

Optimizer

Given projections, the optimizer solves the salary-cap problem with integer linear programming to maximize total projected points under DraftKings' roster constraints ($50,000 cap, 2 SP, 1 C, 1 1B, 1 2B, 1 3B, 1 SS, 3 OF). Locks, excludes, and stack rules are first-class constraints. A single solve runs in <100ms; multi-lineup portfolios run in 1–3s.

Data refresh cadence

• DraftKings player pool & salaries: every 15 minutes
• Probable pitchers & starting lineups: hourly until first pitch
• Weather: hourly, plus 90 minutes before each first pitch
• Vegas lines / implied totals: on line move (push from The Odds API)
• Season-to-date rate stats: nightly

Frequently asked