Post-Race Forensics: Greyhound Racing Edge

Post-Race Forensics

Transforming retrospective analysis into a forward-looking signal generation system. Focusing on lay strategies, market inefficiencies, and systematic execution.

Target Strategy Lay-First / Inefficiency
Core Metric Outcome Plausibility
Execution API / Automated

Market Reality: Causes vs. Outcomes

Greyhound betting markets structurally over-reward visible outcomes (finishing position) and under-price the causal factors (mechanics, interference, split times). The market forms narratives based on emotional resolution immediately after a race.

The Edge: By decoupling the result from the performance, we identify "False Winners" (candidates to Lay next time) and "Good Losers" (candidates to protect/Back).

Cognitive Gap Analysis

  • ✖ Public Market: Remembers the result. Forgets the structure.
  • ✔ Forensic Model: Discards the result. Stores the structure.

Fig 1. Divergence between Market Price (Outcome Driven) and True Probability (Cause Driven).

Forensic Timelines

Strict temporal discipline is required to avoid hindsight bias. Analysis must be frozen before the next market forms.

Variables vs. Constants

To build accurate models, we must rigorously distinguish between dynamic inputs (recalculated per race) and fixed anchors.

Variables (High Signal)

Dynamic

These are recalculated after every performance. They represent the current form and situational outcomes.

Performance Residuals

Difference between expected and actual split times, adjusted for interference.

Structural Advantage

Was the path clear? Did the trap draw force a collision? (Contextual luck).

Market Behaviour

Late drift or steam in the previous race indicating insider confidence/lack thereof.

Constants (Anchors)

Immutable

Fixed anchors that must NOT be reinterpreted based on a single result. Changing these leads to overfitting.

Track Context

Track geometry, surface type, and standard bias for specific trap numbers.

Baseline Genetic Profile

A sprinter does not become a stayer overnight. Breeding and history are fixed.

Economic Incentives

Prize money distribution and kennel rank hierarchies remain constant constraints.

Outcome Plausibility Matrix

This model compares the Quality of Performance (Forensic Score) against the Official Result.

Implausible Win (Lay Target) Dog won, but forensics show low performance quality (lucky gaps, interference behind). Market will over-shorten next time.
Good Loss (Value Target) Dog lost, but forensics show high performance quality (blocked, bad draw, fast split). Market will drift price.

Operational Integration

Consistent edge requires an automated pipeline. Manual re-analysis at the track is prone to cognitive failure.

API
1

Ingest

Raw Race Data
Video Parsing

2

Compute

Forensic Metrics
(Cooling Phase)

3

Label & Store

Versioned JSON
"Implausible Win"

4

Execute

Next Race Entry
Auto-Lay Trigger