How to Encode the Saturday Set Method Into a Model

Why is the Saturday Set method well-suited to a model?

Unlike vague handicapping intuition, the Saturday Set method decomposes into discrete, encodable rules. Pace mapping produces a tempo classification (soft, average, above average). Tempo mismatch is a comparison between a horse's win tempo and today's expected tempo. Race strength is a numerical comparison. Barrier-to-pattern matching is a logical compatibility check. Each of these becomes a feature or a filter in a model — and because they're structural, they generalise across meetings and jurisdictions.

How do I encode the pace map as a feature?

Tag each runner with an early-speed rating and a preferred running style (leader, on-pace, midfield, off-pace). Sum the genuine early-speed runners to classify race tempo: several leaders drawn to press implies above-average tempo. Then derive each horse's likely position. This tempo classification becomes the master filter — downstream rules for tempo mismatch and ruffy qualification depend on it, so validate it first against how races actually unfold.

How do I turn race strength into a quantitative rule?

Assign every race a numerical strength and store each horse's strength-at-win and strength-at-placing history. Build a feature counting how many times a horse has run at or above today's strength and its results in those runs. Encode the elimination rule: repeated place-misses at or above today's strength downgrade the horse sharply; no runs at this level flag it as an unknown-risk with higher variance. This directly models the class-query logic that catches overpriced rising favourites.

How do I model the barrier-to-pattern compatibility check?

Create a compatibility score from two inputs: the horse's required running position and the consequences of its barrier given the track's bias. A pace-reliant horse drawn wide scores poorly because it must expend early energy. An off-pace horse drawn inside on a dead-inside track scores poorly due to being trapped. Feed a live or historical inside-bias signal into this score so the model penalises structurally compromised draws automatically.

How do I combine signals to surface value against the market?

The Saturday Set horse is the convergence of class superiority and positional advantage plus the right preparation stage and neutral-or-positive track. Model it as a weighted composite: ratings rank, tempo-fit, race-strength clearance, barrier compatibility, and preparation flag. Then compare the model's implied probability against market prices. Where the model rates a horse well above its market price — often a good runner eliminated by the crowd for a contextual reason — you've found value. Ruffies emerge specifically in high-tempo races where an off-pace, able horse is drifting; encode them as a separate long-shot each-way signal gated on the tempo classification.

How should I handle track conditions in the model?

Encode the Heavy 10 Rule explicitly: apply significant condition-based adjustments only when the track is Heavy 10 and race-day rain is confirmed. Otherwise, keep condition adjustments minimal. This prevents your model from overfitting to weather forecasts — a documented human failing the method warns against — and keeps form-based signals dominant across Soft and Heavy tracks that aren't actively deteriorating.

Next step: Prototype the pace-map classifier first, validate it against how tempos actually played out over a sample of past cards, then layer in the race strength and barrier compatibility rules. Backtest the composite Saturday Set signal and the ruffy signal separately against closing prices to measure real value capture.