One horse. One bet. Every day. Sly Man’s read on the day’s card — the shortlist, the price, and the case for the pick.
Browse the logbookMullins and the money rarely lie in the same direction by accident — when they do, you listen. Today's scan covered all 48 declared races and 516 runners on the card, running them through a four-stage filter pipeline before a weighted composite score did the rest. The odds band held runners priced between 2.0 and 50.0 decimal — no odds-on, no 50/1 lottery tickets. Field size was capped at 6–14 runners, cutting out the noise of small fields and the chaos of big ones. Two runners were dropped at the lay-of-day collision stage, leaving 367 runners in the live pool. Feature weights in the engine run heaviest on market movement (1.6) and speed (1.5), followed by form (1.2), then class change and going (both 0.8), trainer strike rate (0.8), and draw (0.5 — reserved for v2, unavailable today).
The day opened with 48 races and 516 runners. The odds band trimmed that to 463, field-size filtering brought it to 369, and the lay-of-day collision check removed two more, leaving 367 runners in valid contention. From those, the engine ranked every survivor by weighted composite and handed me the top six to look at.
The top six by composite were Thread Of Gold (7.40/10), Quest For Stars (6.98/10), Nibras Gold (6.96/10), Angels' Share (6.91/10), Pace Attack (6.87/10), and Cranachan (6.81/10). The gap between first and second is not enormous — 7.40 versus 6.98 — so I want to be straight about that: Thread Of Gold leads but it is not a runaway. What tipped the balance was the convergence of a clean market steam and a strong form string at a price that the model reads as significantly mispriced. Quest For Stars has the better raw form figures and a hotter trainer, but the market gave us nothing — no movement either way — and at 9/2 the edge is thinner in value terms despite the cleaner form. That market silence at the top of a Group 1 field is itself a signal worth heeding.
The read I keep coming back to on Thread Of Gold is the market move. The engine weights market movement at 1.6 — the highest of any feature — and the score here is the maximum the model awards: +2.00, flagged as a steamer where a strict majority of bookies shortened and none drifted. At 6/1 in a nine-runner Group 3 field, that kind of coordinated move is not casual. The model's estimated win probability sits at 37.5% against a market-implied 12.8%, a raw edge of +162.6%. I treat that number with appropriate scepticism — models do not win races — but when the bookmakers are simultaneously shortening a horse the model already rates as underpriced, the two things are pointing the same way, and that matters.
Beyond the money, the feature picture holds up. Speed scores at +0.81 on a Saturday Rating of 112 against a field mean of 100.5 — that is a meaningful gap at this level, not a flattering z-score dragged up by a weak field. Form is rated +2.15 on a recent line of '17-311', with the decay-weighted average on the latest win coming in at +1.43. The 7 and the 1 in that sequence are recent enough to matter, and the retrieval from a mid-run position is encouraging. Going is +1.10 — 2-from-3 with three placed on good or adjacent ground, which is exactly what Leopardstown is showing today. Trainer W P Mullins is running at 21% over the past 30 days from 14 wins from 66 runners — the engine calls that hot, and on a Saturday Group 3 at his home track, there is no reason to argue.
Class change is unavailable — no prior races with a recognised class to compare against, so that feature contributes nothing to the score. Draw is also unavailable, reserved for v2. Two missing features in a composite of 7.40/10 means the score is built on five pillars, not seven. That is an honest limitation. But the five features that do fire are all pointing the same direction, and the pillar that carries the most weight — market movement — is as clean as the model ever records.
The composite lead over Quest For Stars is only 0.42 points, and Quest For Stars has a form line of '111' that is simply cleaner — three consecutive wins, going 3-from-3 on today's ground, and a trainer in Charlie Appleby running at 31% over the last 30 days. If the market's silence on Quest For Stars reflects inside knowledge rather than indifference, that horse could make the model look slow. The class change feature is unavailable for Thread Of Gold, which means we are flying partially blind on how it handles a step up to Group 3 company — that gap in the data is real. The going is forecast Good at Leopardstown but any overnight rain softening it beyond Good to Firm / Good would need checking against the going score, which is built on good or adjacent form only. Finally, this is a market-move led case more than a speed-led or form-led one; if the steam was tactical rather than informed, 6/1 in a competitive Group 3 is a price that does not forgive being wrong.
Thread Of Gold · 6/1 · 1 unit · MEDIUM
The Fox’s Wire · Bet of the Day
The algorithm pulled five contenders from today’s card. One earned the ticket — the rest are what it beat. When each race settles, the finishes fill in so you can judge the shortlist, not just the selection.
Position only — payout-neutral. A process check, not a P/L. Click any horse to read its profile.
