TrapMetrics
Try the demoAccess →View access options
HomeThe greyhound System Builder
System Builder

Your idea.
A repeatable set of rules.

Turn a research question into rules and test them on five years of the official record, with a holdout, a walk-forward and a since-saved record that hindsight cannot touch.

Try the rule filter Read about evidence
1.51m

settled runs

The official record since June 2021

64

rule-building elements

In 11 groups, all knowable before the off

11

runner form measures

Ranked within every field

7

race analysis views

From the form to the market

Run count from the System Builder record as of 11 September 2026; the other figures are counts from the app. Sources & definitions ↗

System Builder / Behind the scenes

A backtest is only
the beginning.

64 elements across race conditions, the dog, connections and the model’s view. Test your rules, then inspect how they hold up across different periods.

  • Holdout & walk-forwardSee the development period separately from the test, then inspect each half-year.
  • Trim & FrontierFind which rules earn their place and how changing a threshold changes the trade-off.
  • Since-saved trackingFollow the record from the day you saved the system.
Inside System Builder
Actual System Builder element configurationActual System Builder element configuration
From the app / Element configuration · Enlarge screenshot ↗
Actual walk-forward table with bets, strike rate, ROI, confidence intervals, A/E and subsequent resultsActual walk-forward table with bets, strike rate, ROI, confidence intervals, A/E and subsequent results
From the app / One real system, losing periods included; not a typical-return claim · Enlarge screenshot ↗
01 / Define your rule

Find a starting point.

A filter over the twelve runners in the two demo races. The real Builder tests a rule on 1.5 million runs, not twelve.

02 / Review the matches

Matching runners

RunnerTrackRating
Twelve runners say nothing about a rule. That is what the holdout and walk-forward are for.
From an idea to useful evidence

Define it. Test it. Challenge it.

02

Inspect the full sample

Keep the denominator, dates and exclusions beside any result. A handful of winners is not the whole record.

03

Check a fresh period

Use data that did not inform the rule. Tuning against one sample makes chance patterns look convincing, which is why there is no Goal Seek.

Why there is no Goal Seek →
An actual look inside

Build. Test. Refine. Follow.

System Builder start screen with 64 elements and starter systemsSystem Builder start screen with 64 elements and starter systems
From the app / Starter systems include examples chosen with hindsight and labelled so; they are not recommendations · Enlarge screenshot ↗
Your next step

Give your next idea a place to develop.

Research access opens the model era, Trim, Frontier and staking in the Builder, and follows each saved system from the day you save it.

View access optionsExplore Portfolios

Register before accounts open and Research is £19.99 a month, locked in.