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 logbookOn a day where three Brighton runners and a pair of Kempton runners all steamed into the shortlist, the one horse with no form string, no going data, and no class history still topped the board — and that tells you something about where the real signal was coming from. Today's scan covered the full Saturday card across 33 races and 331 runners, applying a sequential filter pipeline before any composite scoring began. Odds-on shots were stripped out; only runners priced between 2.0 and 50.0 decimal survived. Fields smaller than 6 or larger than 14 were removed as too noisy or too chaotic to read with confidence. Four runners were then dropped for colliding with the Lay of the Day list, leaving 203 runners to score. Feature weights ran at: market_move 1.6, speed 1.5, form 1.2, class_change 1.0, trainer_strike and going both 0.8, draw 0.5 — draw data unavailable for v2 build, noted throughout.
The day opened with 33 races and 331 runners declared. The odds filter cut that to 261, the field-size gate to 207, and the Lay of the Day collision check removed a further 4, leaving 203 runners eligible for composite scoring. From that pool of 203, six runners surfaced as the meaningful shortlist — three from Brighton, two from Kempton AW, one from Yarmouth.
The top six by composite were: Montalbano at 7.21/10, Highland Harvey at 7.03/10, Silca Bay at 6.79/10, Sea Suite at 6.65/10, Angel Gabriel at 6.57/10, and Havana Mojito at 6.55/10. The gap between first and second — 0.18 points — is not wide; I want to be straight about that. What tipped the balance for Montalbano over Highland Harvey was a combination of the market_move contribution (+2.00, the ceiling figure, identical across the top runners) arriving alongside a cleaner speed z-score and, critically, a market-implied probability that the model's estimate genuinely eclipses — a +58.6% edge versus Highland Harvey's edge of -24.9% (that horse is market-favourite; the model does not see the same value). Havana Mojito fell away despite a strong speed reading because of a 0/19 cold trainer strike rate — that drag was decisive.
The read I keep coming back to with Montalbano is the market move, and I mean that precisely: the data shows a strict majority of bookmakers shortened, none drifted. That is the cleanest possible steam signal the engine can return — +2.00 on a 1.6-weighted feature is the maximum contribution available, and it is carrying this score. At 10/3 in a novice stakes, a horse steaming that hard is telling you something the tissue price did not. The model's estimated win probability sits at 36.6% against a market-implied 20.6% — that is a +58.6% edge. I am not in the business of pretending edges that size come for free; they can reflect a market that has not yet corrected, or they can reflect data the model is reading cleanly that the public is ignoring. Here, I think it is the former — the steam is the correction beginning. The speed figure adds meaningful weight: an SR of 152 against a field mean of 140.6, a z-score of +1.06, is a solid read in a Class 3 novice. Andrew Balding's 30-day strike rate sits at 18% across 22/124 — on par, not hot, but not a drag either. The composite of 7.21/10 is a strong score by the engine's own scale.
There are three honest flags to carry into this bet. First, form is unavailable — there is no form string on file for Montalbano, which in a novice context can mean a lightly raced or unraced horse; the model cannot penalise what it cannot see, but neither can it validate it. Second, going data is absent — no career runs on good to soft or adjacent surfaces are recorded, so how this horse handles Kempton AW conditions on an August evening is genuinely unknown. Third, the shortlist gap to Highland Harvey is only 0.18 points, and while Harvey's negative edge versus the market cools my interest in it, a composite of 7.03/10 with a hot trainer (25% strike rate), genuine form, and a confirmed speed edge is not a runner I am dismissing lightly. Engine conviction is marked MEDIUM — that is the right call.
Montalbano · 10/3 · 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.
