Skip to main content
Blaze Sports Intel
ScoresCollege BaseballIntelAgentPricing
College Baseball scores updated live every 30 seconds
HomeScoresBaseballAnalytics
  1. Home
  2. ›Models
  3. ›HAV-F
Live — v1.0

HAV-F Methodology

Hitting. At-Bat Quality. Velocity. Fielding. Four measurable dimensions of player performance, compressed into a single composite score that tells you what batting average and ERA alone never will.

HAV-F is BSI's proprietary player evaluation framework for college baseball scouting analytics. Every player is scored 0-100 on each component via percentile rank against their cohort, then the four components are weighted into a composite. No black boxes — every weight and every input is documented below.

Why This Exists

MLB has Statcast — Hawk-Eye cameras in all 30 ballparks tracking every pitch, swing, and throw at 300 frames per second. FanGraphs, Baseball Savant, and Baseball Reference turn that data into consumer-grade analytics. College baseball has nothing equivalent.

TrackMan and Yakkertech are installed at many D1 programs, but that data stays behind closed doors. The public gets box scores and conference stats pages. If you want to evaluate a college player's actual profile — not just his batting average — you're on your own.

HAV-F fills that gap. It takes the publicly available statistical output and builds a structured evaluation framework around it. The limitation is real: without pitch-tracking data, HAV-F uses power metrics as a proxy for bat speed and exit velocity (the V component). The advantage is also real: HAV-F provides a consistent, documented, comparable player evaluation that doesn't exist anywhere else for college baseball.

How It Works

HAV-F scores are percentile-based, not raw. A player with an 80 H-Score isn't “80% good at hitting” — he's better than 80% of the players in his cohort at hitting. This means scores automatically adjust as the cohort changes. A player who looks elite in a weak conference will score lower when ranked against the full D1 population.

The computation runs in three steps:

1. Build the percentile table

For each stat (AVG, OBP, SLG, etc.), sort all players in the cohort from lowest to highest. This sorted distribution becomes the lookup table for percentile ranking.

2. Rank each player

For each stat, binary search the sorted distribution to find where the player falls. The position in the distribution, adjusted for ties, becomes the percentile rank (0-100). A player at the median gets a 50.

3. Weight and composite

Sub-stat percentiles are weighted within each component (e.g., AVG 25% + OBP 25% + SLG 20% + wOBA 20% + ISO 10% = H-Score). The four component scores are then weighted into the final composite: H 30% + A 25% + V 25% + F 20%.

Composite Formula
HAV-F = (H × 0.30) + (A × 0.25) + (V × 0.25) + (F × 0.20)

where H, A, V, F ∈ [0, 100]
and   HAV-F ∈ [0, 100]

The Four Components

H
Hitting30% of composite

Pure offensive production. How well does this player hit, measured against every other player in the cohort?

Sub-stat weights
AVG25%

Batting average — contact rate and ball-in-play quality

OBP25%

On-base percentage — reaching base by any means

SLG20%

Slugging percentage — total bases per at-bat

wOBA20%

Weighted on-base average — assigns run values to each outcome

ISO10%

Isolated power — extra-base hit ability (SLG minus AVG)

A
At-Bat Quality25% of composite

Plate discipline and process. A player who walks, avoids strikeouts, and drives the ball hard is winning at-bats even when the box score says otherwise.

Sub-stat weights
BB%30%

Walk rate — patience and pitch recognition

K% (inv)30%

Inverted strikeout rate — fewer Ks means higher score

BABIP20%

Batting average on balls in play — quality of contact

HR%20%

Home run rate — damage per plate appearance

V
Velocity25% of composite

Power proxy. Without Hawk-Eye data at the college level, BSI uses power metrics as a proxy for bat speed and exit velocity — the physical tools that translate to the next level.

Sub-stat weights
ISO40%

Isolated power — the strongest proxy for raw bat speed

SLG35%

Slugging — total bases reflect hard-hit frequency

HR%25%

Home run rate — over-the-fence power

F
Fielding20% of composite

Defensive value. Fielding data at the college level is thin — BSI uses what exists and defaults to league-average (50) when data is unavailable rather than penalizing or rewarding blindly.

Sub-stat weights
FPCT60%

Fielding percentage — errors relative to chances

RF40%

Range factor — putouts and assists per game

Interpreting Scores

The composite tells you where a player sits overall. The component breakdown tells you why. A 70 composite could be a well-rounded-sm player (65/70/72/68) or a specialist (90/80/60/30). The shape of the radar chart matters as much as the number.

Scouts already think this way — they evaluate tools separately before forming an overall grade. HAV-F formalizes that process against a statistical cohort rather than relying on subjective comparisons.

80-100
Elite

Top of the cohort. Draft-board material.

60-79
Above Average

Strong performer with standout dimensions.

40-59
Average

Solid contributor. The shape of the radar matters more than the composite here.

20-39
Below Average

Weaknesses outweigh strengths against this cohort.

0-19
Developmental

Early-career or limited data. Not a death sentence — check the breakdown.

Data Sources & Limitations

What Feeds HAV-F

Player batting statistics from the Highlightly Pro API, supplemented by ESPN's college baseball endpoints where Highlightly coverage is thin. Advanced metrics (wOBA, ISO, BB%, K%, BABIP, HR%) are derived from basic batting lines using FanGraphs linear weights.

Fielding data comes from the same sources but is significantly less reliable at the college level. Many programs don't report range factor, and fielding percentage alone rewards immobility (a player who never reaches the ball never makes an error). HAV-F acknowledges this by defaulting to a neutral 50 when fielding data is unavailable.

Known Limitations

No pitch-tracking data

Exit velocity, launch angle, and bat speed are unavailable for college baseball. The V component uses power metrics (ISO, SLG, HR%) as a proxy. This correlates with — but does not measure — actual bat speed.

Cohort-dependent scoring

Scores are relative to whoever is in the cohort. A 90 in a 30-player sample means something different than a 90 in a 300-player sample. Always check cohort size.

Batting only

HAV-F v1.0 evaluates position players. Pitchers are not scored. A pitching-specific framework (K/9, BB/9, FIP, WHIP, velocity) is planned for v2.0.

Early-season noise

Small sample sizes in the first 2-3 weeks of the season produce volatile scores. A player who goes 5-for-8 in opening weekend will look elite until the sample stabilizes.

Fielding data quality

Fielding percentage and range factor are blunt instruments. The F component is weighted lowest (20%) for this reason, and defaults to 50 when data is missing.

Technical Implementation

The computation engine lives in lib/analytics/havf.ts — pure math with no external dependencies. It takes an array of player stat objects, builds a percentile table from the cohort, and returns scored results.

Percentile ranking uses binary search for O(log n) lookup per stat per player. Ties are handled with midpoint averaging — if three players share the same OBP, they all receive the percentile at the middle of the tie range.

Computed scores are persisted to Cloudflare D1 (the havf_scores table) with raw input stats preserved for audit trail. The leaderboard API reads from D1 with KV caching (5-minute TTL) for production performance.

D1 Schema Highlights
havf_scores
├── player_id       TEXT (unique per league+season)
├── h_score         REAL [0-100]
├── a_score         REAL [0-100]
├── v_score         REAL [0-100]
├── f_score         REAL [0-100]
├── havf_composite  REAL [0-100]
├── raw_avg, raw_obp, raw_slg, raw_woba, raw_iso
├── raw_bb_pct, raw_k_pct, raw_babip, raw_hr_rate
├── data_source     TEXT
└── computed_at     TEXT (ISO 8601)

Explore the Data

Live

HAV-F Leaderboard

Top 50 college baseball players ranked by HAV-F composite with radar charts and component breakdowns.

15 Profiles

Player Pages

Individual player profiles with HAV-F scores, stat tables, and scouting analytics for top prospects.

Cite this page
Austin Humphrey. (2026, February 24). HAV-F: College Baseball Player Evaluation Methodology. Blaze Sports Intel. https://blazesportsintel.com/models/havf
← All ModelsHAV-F Leaderboard →
BSI
BSIBlaze Sports Intel

Born to Blaze the Path Beaten Less

The only free, park-adjusted sabermetrics platform for D1 college baseball. Plus live scores across MLB, NFL, NBA, and NCAA.

Start Here

  • Live Scores
  • College Baseball
  • BSI Savant
  • Editorial
  • Intelligence

Sports

  • College Baseball
  • MLB
  • NFL
  • NBA
  • College Football

Tools

  • Transfer Portal
  • NIL Valuation
  • Models
  • Glossary
  • Pricing

Ecosystem

  • BSI Labs ↗
  • Arcade ↗
  • BlazeCraft ↗
  • Austin Humphrey ↗

Company

  • About
  • Methodology
  • Standards
  • Data Sources
  • Contact
  • Status

Legal

  • Privacy Policy
  • Terms of Service

© 2026 Blaze Intelligence LLC

Built on Cloudflare · Austin, TX