Honest scorecard
Tap a tool for the full record| Tool | Graded | Win rate | Result | Status | |
|---|---|---|---|---|---|
DFS Player Props In-sample. Includes off-season NBA/NHL. | 68,549 last graded today | 56.5% | +5.8% ROI ROI | Live · graded | |
Sportsbook Picks Best-available odds across books; single-book reality is lower. CLV not yet wired. | 15,932 last graded today | 51.1% | +0.7% ROI ROI (best odds) | Live · graded | |
Prediction Markets Model picks placed on prediction markets (Novig, Kalshi, Polymarket, ProphetX). Best-available odds across venues; single-venue reality is lower. | 13,170 last graded today | 55.4% | +6.6% ROI ROI (best odds) | Live · graded | |
First-Inning NRFI / YRFI Win-rate only — no payout in the ledger, so no ROI is claimed. | 276 18d ago | 57.2% | 57% win rate Win rate | Live · graded | |
Cross-Book Arbitrage Distinct opportunities (latest graded detection; re-detections deduped; palpable-error/>9% rows excluded). Net of each venue's published fee schedule at the detected price — modeled, not observed fills. Pays only if both legs fill. | 17,820 last graded today | — | +1.4% net ROI Net ROI after modeled venue fees | Live · graded | |
Prediction-Market Arbs Graded 224 of 570 surfaced (39.3%) · 224 of 228 due (98.25%) — settlement observed. Arbs surfaced across Kalshi/Polymarket, graded as they settle; return is per dollar locked. Grading observes settlement at detected size; fills were not observed. An arb does not lose — it pays or it voids — so there is no win rate to quote. | 224 last graded today | — | +2.7% net ROI Net ROI after modeled venue fees | Live · graded | |
Home Run Model Overconfident at long odds — not +EV as priced. Recalibrating. | 1,111 last graded yesterday | 15.9% | -13% ROI ROI | Not +EV | |
Pitcher Strikeouts Losing as displayed; claimed edge is anti-correlated with outcome. | 690 last graded today | 49.4% | -5% ROI ROI | Not +EV | |
Middles Distinct opportunities. The percentage is what a middle pays IF both legs fill at the quoted prices — 27.8% of them hit — but legs move before the second fills, so it is not a realised return. Discovery tool, net-negative as actually executable. | 2,141 last graded today | 27.8% | 2.5% on paper Theoretical (both legs must fill) | Not +EV | |
Prediction-Market Edge Paper-traded. Clean cohort only: priced >= 0.40 and locked pre-game (11886 graded). Realized -3.77% ROI — a 54.7% win rate at these prices needs 57.9% just to break even after fees. The lifetime number you may have seen was a longshot artifact from a price band the board no longer surfaces, plus rows locked in-play before the 2026-07-06 clock fix. No proven edge yet. | 11,886 last graded today | 54.7% | Experimental Experimental | Experimental |
Wins and losses both counted; picks past their settlement window without a grade are shown in mature_pending (target 0, monitored). Historical and in-sample; past results don't guarantee future performance. 21+.
How we grade
The tools shown here are available to members in real time. Older historical results and newer fully tracked results are labeled separately. First week of Pro is $1, then $149/mo only if you stay. Cancel in one click.
Frequently asked questions
Does Turtle +EV show its losing picks?
The page shows wins, losses, pushes and known pending grades. New Verified Results are separated from older Historical Results, and official corrections may update settled rows.
How many picks has Turtle +EV graded?
Thousands of graded predictions are included in the historical model and tool record. That total is not a count of unique customer recommendations, so records are shown separately by product.
Is the record audited?
The older record is reconstructed from Turtle’s prediction and grading history. New Verified Results require a posted pick, recorded price, model version and linked result.
Does a good record guarantee future profit?
No. Past results never guarantee future performance. The value is a transparent, auditable process and tools, not a promise. 21+.