InplayRadar
Personal sports intelligence

Read the match
as it happens.

A quantitative framework for live football. Two proprietary metrics - rolling expected goals (rxG) and the d_index momentum series - feed a family of specialist models that evaluate every chance the moment it occurs.

85.4%ROBIN.AI · 1X2 accuracy
30sd_index sampling interval
2Proprietary metrics
Live
Matchday feed
00'
Northbridge
rxG 0.00
0:0
Crosshaven
rxG 0.00
HomePressureAway
54%46%
Momentum indexd_index +8
Matrix50
Reading the match…
rxG +0.62 / overd_index +37ROBIN.AI · 85.4%Matrix 68MACY.AI · 80.1%rxG modellemePERSIE.AI · 67.1%d_index Δ30sBergkamp.ai · 0.74rxG +0.62 / overd_index +37ROBIN.AI · 85.4%Matrix 68MACY.AI · 80.1%rxG modellemePERSIE.AI · 67.1%d_index Δ30sBergkamp.ai · 0.74
Proprietary mathematical metrics

A formal model of the live match

InplayRadar reduces ninety minutes of football to three proprietary quantities. Each is a defined mathematical metric, recomputed continuously from event-level data rather than estimated from the scoreline.

rxG

Rolling Expected Goals

A proprietary rolling estimator that integrates the expected-goals value of every chance as it occurs, weighting each attempt by its modelled conversion probability. It states the statistical quality of play, independent of whether a shot was finished.

2.84cumulative xG
d_index

d_index Momentum

A proprietary pressure metric sampled on a fixed 30-second interval. The series is signed: positive values quantify home territorial dominance, negative values the away side, giving a continuous mathematical trace of where control sits.

+42home dominance
Matrix

Matrix Score

A composite evaluation index that combines the live rxG differential with the d_index trace into a single bounded score from 0 to 99, expressing which side mathematically controls the match at any instant.

68
control index

The rxG estimator and the d_index series are the foundation of every downstream model. Their full derivation, parameters, and worked match studies are documented in the rxG modelleme reference, alongside the broader canlı futbol analitiği platform.

The model ensemble

Specialist models over a shared feed

Each model is trained on a distinct question, yet all read the same rxG estimator and d_index series. Reported accuracy figures are measured on out-of-sample matches. Select a model to inspect what it observes.

Together the ensemble functions as a continuous momentum takip robotu, reading each fixture in real time from the same underlying metrics.

Methodology, in your hands

Test the metrics against real conditions

Two interactive instruments. Define thresholds on rxG and the d_index and measure how a rule would have classified past matches, or step through a simulated fixture and watch the metrics recompute on the 30-second clock.

Custom strategy builder

Backtest · 12 matches

Set the conditions you want a signal to fire on, then see how often the model's call matched the eventual result across a sample of past matches.

Base model
Signals
6
of 12
Accuracy
100%
6 correct
Coverage
50%
of matches
Matched
NBR v CRHASH v MRNHBR v PNEGRA v MILDRY v WNDCLF v BAY

Live match simulator

A simulated fixture, scored exactly as the live engine reads a real one. Watch rxG and momentum react minute by minute.

Northbridge
rxG 0.00
0:0
00'
Crosshaven
rxG 0.00
HomePressureAway
54%46%
d_index logLevel edge
Matrix
50
d_index
+8
Total rxG
0.00

These instruments mirror the live engine on simulated data. The production models, full match archive, and backtesting workbench run on the InplayRadar stratejiler workspace.

Read the match as a system, not a scoreline.

The live rxG estimator, the d_index series, the full model ensemble, and the backtesting workbench run inside the platform. The methodology is open to read in full.