SURGETennis Intelligence
Methodology

How Tour Radar is built.

Tour Radar is a public proof surface. It turns charted matches into player fingerprints, not into a pretend all-knowing ranking system.

Source and permission

Source data from Jeff Sackmann's Match Charting Project, used with permission.

Match Charting Project data is used on this site with explicit permission.

Coverage rule

A player only gets a public page after reaching 20 or more charted matches since 2020. The current build covers 81 ATP players and 57 WTA players.

Opponent-strength weighting

Each match is weighted by an internal charted-match Elo built from the point files. Stronger opponents count more than softer ones when Tour Radar aggregates a player profile.

Empirical-Bayes shrinkage

Small samples are pulled back toward the tour-window baseline instead of being allowed to spike from a thin run of matches or pressure points.

Normalized signal scores

Percentiles and composite scores are calculated on the adjusted values, not on raw pooled percentages. The raw sample still stays visible on the player page.

Windows

Since 2020

Charted matches from 2020 through 2025.

2020-2025

Last 2 Seasons

Charted matches from 2024 and 2025.

2024-2025

Current Season

Charted matches from 2025.

2025

Metrics

Eight public scouting angles.

return-depth-pressure

Return Depth Pressure

How often a player sends returnable balls very deep.

Very deep returns divided by returnable returns from charted matches.

Displayed values are opponent-strength weighted and shrunk toward the active tour-window baseline before the percentile is assigned.

Threshold: 300+ returnable returns

forehand-intent

Forehand Intent

How often the forehand chooses line-changing or inside patterns.

Down-the-line, inside-out, and inside-in forehands divided by all charted forehand directions.

Displayed values are opponent-strength weighted and shrunk toward the active tour-window baseline before the percentile is assigned.

Threshold: 500+ charted forehand directions

serve-variety

Serve Variety

How evenly a player spreads first-serve targets across the box.

Normalized Shannon entropy across the six deuce/ad serve target buckets on charted serves.

Displayed values are opponent-strength weighted and shrunk toward the active tour-window baseline before the percentile is assigned.

Threshold: 300+ charted serves

net-pressure

Net Pressure

How often a player brings points forward.

Net points divided by total points played in charted matches.

Displayed values are opponent-strength weighted and shrunk toward the active tour-window baseline before the percentile is assigned.

Threshold: 30+ net points and 600+ total points

net-conversion

Net Conversion

How well a player finishes when the point moves forward.

Points won at net divided by charted net points.

Displayed values are opponent-strength weighted and shrunk toward the active tour-window baseline before the percentile is assigned.

Threshold: 30+ net points

break-point-resistance

Break-Point Resistance

How well a player holds up on serve at break point.

Break-point service points won divided by break points faced on serve.

Displayed values are opponent-strength weighted and shrunk toward the active tour-window baseline before the percentile is assigned.

Threshold: 20+ break points faced on serve

break-point-attack

Break-Point Attack

How often a player wins return points on break point.

Break-point return points won divided by break points created on return.

Displayed values are opponent-strength weighted and shrunk toward the active tour-window baseline before the percentile is assigned.

Threshold: 20+ break points on return

first-strike-compression

First-Strike Compression

How often service points are decided inside three shots.

Service points ending in one to three shots divided by all charted service points.

Displayed values are opponent-strength weighted and shrunk toward the active tour-window baseline before the percentile is assigned.

Threshold: 300+ charted service points

Caveats

01

Not every player is covered.

The charted sample is rich, but it is not complete. Tour Radar is deliberately honest about that.

02

Surfaces are uneven.

Hard-court coverage is stronger than grass, especially on the WTA side. Surface cards only appear when there are at least 10 charted matches in the selected window.

03

This is a charted-match Elo, not an official rating.

The weighting model is only built from the charted-match dataset. It is useful for strengthening the public signal, but it is not a replacement for full tour-wide opponent models.