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 logbookWhen the scores cluster this tight, the market stops being background noise and becomes the deciding vote. Today the engine ran across 60 races and 574 runners declared on 17 October 2026, applying a filter pipeline designed to strip out odds-on shots, chaotic big fields, and anything the Lay of the Day flag has already marked. The odds band was set at 2.0–50.0 decimal (evens to 49/1, no odds-on), field size capped at 6–14 runners, and the feature weights loaded with market_move carrying the heaviest hand at 1.6, speed close behind at 1.5, and draw reserved for a v2 build — unavailable today. Seven features in total, six live.
The pipeline opened on 574 runners. The odds-band filter cut that to 487, removing all odds-on runners and anything priced beyond 50.0 decimal. Field-size trimming brought it to 350, keeping only races where the draw can't hide a poor horse and small fields won't make a nonsense of form reads. A single Lay of the Day collision dropped one further runner, leaving 349 eligible runners entering the scoring stage.
Five horses came out of the scoring with composites in a 7.09–7.85 band, and the honest read is that the gap between them is narrow. Duchessofaquitaine leads on composite at 7.85/10 — driven hard by a maximum market_move score and a speed figure of SR 152 against a field mean of 131.3 — but form and going are both unavailable, which softens the confidence picture. Emerald Enigma posts 7.67/10 with the cleanest feature set: speed, form ('1-131'), going, and market_move all scoring positively. Arapaho Gold sits at 7.36/10 with a strong going record (4-from-5) and sharp recent form ('11113'), but trainer Michael Dods is running at 1% over the last 30 days — a drag that pulled the composite down. Trean and Carried With Love both land at 7.09–7.10, with Ride Like A Girl matching that score at the bottom of the shortlist. What tipped the balance to Trean is not the composite — it isn't the top score — it's the margin between first and second: the lead of 0.18 between Duchessofaquitaine and the pack falls well short of the 0.4 threshold the engine requires to call a clear leader. When the scores cannot separate the field, the protocol is to follow the shortest price on the shortlist: the horse the market itself rates most likely to win.
Trean goes to post at 6/5 (decimal 2.2), the shortest price on the entire shortlist, running in the Mongey Communications Killavullan Stakes (Group 3) at Leopardstown at 16:15. The market is stating a 40.6% implied win probability, and while the engine's own estimate of 36.1% shows a slight negative edge of -20.6% relative to our model — I'm not going to pretend otherwise — the protocol here is clear: when composite scores are level, the shortest-priced runner carries the crowd's money, and that crowd includes bookmakers who price Group 3 races with more precision than maiden handicaps. The selection is chosen by tiebreak, not by edge, and I'm saying that plainly. The composite of 7.10/10 is built on a speed figure of SR 121 against a field mean of 92.0 (z = +1.75), recent form of '301341' with the latest win carrying a decay-weighted score of +1.01, and a market_move score of +2.00 — a maximum steamer reading, strict majority of bookies shortened and none drifted. Trainer P. Twomey has gone 8-from-30 over the last 30 days, a 27% strike rate that the engine flags as hot.
The going score is the weakest plank here: Trean has run three times on good or adjacent ground without a win, with just two places — a +0.10 going score that is as close to neutral as it gets. If the Leopardstown ground is riding quicker than the published good, that record doesn't improve. The negative model edge (-20.6%) is also unusual — most selections today carry a positive edge; this one doesn't, because the market prices Trean more favourably than the model does, and that cuts both ways. The form string '301341' is honest but lumpy; the 3 and the 0 in recent starts are there in plain sight. Finally, the value case today — if you're looking for it separately — sits with Duchessofaquitaine at 9/2, which carries the session's top composite of 7.85/10 and a model edge of +118.2%; that horse is not today's bet but it is worth knowing about.
Trean · 6/5 · 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.
Value pick 4 of 24 won · +6.83 units at advised prices · since 11 October 2026. Kept separate from the bet’s record above.
