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 logbookEvery so often the market hands you a read that the form book simply cannot — today is one of those days, and I'm following the money. The engine scanned all 50 races declared for 23 July 2026, working through 442 runners in three filter stages. First, odds: any runner priced odds-on or beyond 49/1 in decimal terms was cut — keeping only the 2.0–50.0 band where there is genuine signal without the chaos of complete outsiders. That trimmed the field to 374 runners. Then field size: races with fewer than six or more than fourteen declared runners were removed — small fields are noisy, big fields are chaos — leaving 260 runners across the qualified races. Feature weights ran as follows: market_move at 1.6 (the heaviest), speed at 1.5, form at 1.2, class_change at 1.0, trainer_strike and going both at 0.8, and draw reserved for v2. Engine conviction came back at MEDIUM.
The day opened with 50 races and 442 runners. The odds-band filter dropped 68 runners — anything odds-on or priced beyond 50.0 decimal — leaving 374. The field-size filter then removed runners in races outside the 6–14 bracket, cutting a further 114 and landing on 260 qualified runners. From that pool the engine ranked every horse by weighted composite and surfaced six names worth a proper look.
The top six by composite were L B Double U (6.90/10), Rogue Dynasty (6.77/10), Timely Affair (6.76/10), Ruby Sands (6.74/10), Little Mester (6.74/10), and Arry Up (6.69/10). The scores are honest: these are tightly bunched, separated by fractions, and I'm not going to pretend the gap between first and sixth is wide — it is not. What separates L B Double U at the top is the combination of the single biggest market_move score on the sheet and the widest estimated edge in the field: a model-implied win probability of 35.1% against a market-implied 8.1%, an edge of +286.3%. That edge figure is roughly ten times the edge on Rogue Dynasty. At 10/1, you are being offered a price that the engine says is dramatically wrong. That is the feature that tips the balance.
The read I keep coming back to with L B Double U is the market_move score: +2.00 on a weight of 1.6 — the maximum the engine awards, and it awarded it on the basis that a strict majority of bookmakers shortened overnight with none drifting. At 10/1 in an eight-horse novice at Southwell, that kind of uniform steam is not punters having a flutter; that is informed money, and it registers a market_move contribution of +1.37 to the composite — the joint-highest on the entire shortlist alongside Ruby Sands. The model's estimated win probability is 35.1% against a market-implied 8.1%: an edge of +286.3%, which is not a rounding error — it is the engine telling me the market has mispriced this horse by a significant margin.
On the speed side, L B Double U carries an SR of 141 against a field mean of 128.2, a z-score of +0.86 and a weighted contribution of +0.55 to the composite. That is a solid read — not the sharpest speed number in the shortlist (both Timely Affair and Little Mester post higher z-scores), but it is meaningful in a Class 5 restricted novice where the field mean is modest. The horse has class on raw figures.
I have to be straight about what I do not have here. Form is unavailable — no form string on file. Going is unavailable — no career runs on good or adjacent ground to draw on. Class_change is unavailable — no prior recognised class to compare. Trainer Tom Dascombe is running at 11% over the last 30 days, marginally negative in the model. Those gaps are real. The entire case rests on the speed figure and, overwhelmingly, on the market. On a day where the shortlist is tightly packed and the scores are marginal from second downwards, I'm prepared to follow a +286.3% edge at 10/1 — but I'm doing so with my eyes open.
The risks here are material and I'm not going to dress them up. Form is completely absent from the data — there is no form string on file for L B Double U, which means the engine is working blind on recent racecourse behaviour. Going is equally absent — no career runs on good or adjacent ground, so if the surface rides in a way that does not suit, there is no historical evidence either way. The trainer strike rate of 11% is marginally below par. This is a restricted novice stakes at Southwell — by definition, lightly raced or unexposed horses, which adds variance. Rogue Dynasty and Timely Affair both bring fuller evidence bases: form strings, going records, and composite scores within a whisker of the selection at 6.77 and 6.76 respectively. If the steam here proves to be stable-yard confidence rather than market intelligence, and the horse runs green or encounters unsuitable ground conditions, the 10/1 will look the right price. Engine conviction is MEDIUM — I'm backing the signal, not overclaiming the case.
L B Double U · 10/1 · 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.
