Tipster Performance
Sporting Post · 6 sources 13544 settled picks · 13926 total ⚠ All-time HP picks vs shorter tipster window — not directly comparable ← Daily playsTop-pick (rank 1) performance — all SP bands sorted by Wilson LB
| Tipster | Picks | Wins | Win% | Wilson LB | ROI |
|---|---|---|---|---|---|
| ⭐ HorsePlay (model rank-1) | 11831 | 3964 | 33.5% | 0.328 | -6.6% |
| Morning Line | 758 | 234 | 30.9% | 0.2818 | -10.8% |
| Computaform | 741 | 217 | 29.3% | 0.2661 | -7.0% |
| Magic Tips | 750 | 204 | 27.2% | 0.2461 | -13.9% |
| Computaform Ratings | 750 | 188 | 25.1% | 0.2256 | -14.8% |
| Nadine Low Ah Kee | 16 | 6 | 37.5% | 0.2081 | -13.2% |
| Winning Form | 754 | 172 | 22.8% | 0.204 | -12.1% |
| Sporting Post | 753 | 170 | 22.6% | 0.2017 | -18.3% |
| Raceform | 1 | 0 | 0.0% | 0.0 | -100.0% |
Outsider detection — top-pick SP ≥ 13/1 sorted by Wilson LB
The key differentiator. Which sources can spot big-price winners, not just market leaders? Once ≥ 5 outsider picks per tipster, the consensus badge switches to outsider-specific weights.
| Tipster | Out picks | Wins | Win% | Placed | Place% | Wilson LB | ROI (win) | Weighting |
|---|---|---|---|---|---|---|---|---|
| ⭐ HorsePlay (Outsider Detective) | 52466 | 3174 | 6.0% | 11487 | 21.9% | 0.0588 | -39.5% | DQS clue-based selection |
| Winning Form | 62 | 4 | 6.5% | 17 | 27.4% | 0.0293 | 43.5% | ★ outsider weight active |
| Morning Line | 12 | 1 | 8.3% | 3 | 25.0% | 0.0188 | 41.7% | ★ outsider weight active |
| Sporting Post | 63 | 2 | 3.2% | 15 | 23.8% | 0.0106 | -39.7% | ★ outsider weight active |
| Computaform | 32 | 1 | 3.1% | 8 | 25.0% | 0.007 | -46.9% | ★ outsider weight active |
| Magic Tips | 32 | 1 | 3.1% | 4 | 12.5% | 0.007 | -46.9% | ★ outsider weight active |
| Computaform Ratings | 58 | 0 | 0.0% | 8 | 13.8% | 0.0 | -100.0% | ★ outsider weight active |
| Nadine Low Ah Kee | 1 | 0 | 0.0% | 0 | 0.0% | 0.0 | -100.0% | need 5 more |
| Raceform | 0 | 0 | – | – | – | – | – | need 5 more |
Wilson LB = confidence-adjusted outsider win rate. When ≥ 5 outsider picks settled,
this replaces the overall weight in the T: N/6 consensus for outsider-priced horses.
⭐ HorsePlay uses the Outsider Detective (DQS clue stack) for big-price picks —
not the ensemble ranking. The row above shows GOLDEN/HIGH/PLATINUM alerts.
Any-pick (ranks 1–3) performance — how often at least one pick lands · sorted by Win%
| Tipster | Total picks | Winners | Win% | Placed | Place% |
|---|---|---|---|---|---|
| ⭐ HorsePlay (model rank-1) | 35408 | 7809 | 22.1% | 19030 | 53.7% |
| Morning Line | 2251 | 493 | 21.9% | 1213 | 53.9% |
| Computaform | 2224 | 450 | 20.2% | 1150 | 51.7% |
| Magic Tips | 2251 | 454 | 20.2% | 1160 | 51.5% |
| Computaform Ratings | 2248 | 428 | 19.0% | 1094 | 48.7% |
| Nadine Low Ah Kee | 67 | 12 | 17.9% | 38 | 56.7% |
| Winning Form | 2251 | 400 | 17.8% | 1069 | 47.5% |
| Sporting Post | 2251 | 386 | 17.1% | 1034 | 45.9% |
| Raceform | 1 | 0 | 0.0% | 0 | 0.0% |
Settlement
Run after each race day to link picks to results:
docker exec horseplay-cogent-web-1 python manage.py settle_tipster_picks
Options: --date 2026-05-05 · --days 7 · --force (re-settle)