Systematic prediction-markets engine
WAYL ENGN is an always-on engine that finds mispriced sports outcomes in prediction markets (Polymarket / Kalshi-class venues). Nearly three months of transparent, timestamped, auto-graded forward-testing: positive ROI every month, with the All-Singles win rate climbing 60% → 71% across the first three months, and September the strongest, led by the Top-Single at 79%.
Live Test #1 · Jul 18 → Oct 13 · interactive
A live test, not a backtest: every pick was locked before the game, timestamped, and auto-graded in real time. Live Test #2 (Oct 14 → Dec 31) is running now and will be published here the same way, win or lose.
Every pick from Jul 18 to Oct 13 on one live curve — hover any game to see it, switch between All-Singles, Top-Single and the Engine Parlay, and set your own starting bankroll and flat bet size. The numbers update instantly. The games and entry prices are locked (it already happened); only your sizing changes.
The trajectory
The flagship product, every daily pick bet as a flat single, climbed its win rate each of the first three months (60% → 70% → 71%), with positive ROI every month. Critically, the selection model was held stable through this window (a deliberate no-curve-fit data-collection period), so the climb reflects a maturing system on more data, not a curve fit to the results.
All-Singles = every daily engine pick bet as a flat single. Win rate (bar) with flat-stake ROI beneath. First three full months shown; October is in progress (see the full ledger below). 50% line = a coin flip, shown for reference only; profitability depends on entry price, not a fixed threshold.
The September step-change
The clearest sign of a system maturing rather than a hot streak: measured against the first two months (Jul–Aug), September didn't just hold the edge, it accelerated. The engine's highest-conviction product nearly 5×'d its yield (profit per $1 staked, the fairest way to compare periods of different lengths).
The single highest-conviction pick each day, the sharpest expression of the edge.
Every daily pick, bet flat, improvement on the largest, highest-volume sample.
The vig-heavy product (books tax every leg), and even this is now climbing. The singles still carry the edge.
"Before" = July (from the 18th) + August combined; "September" = the full month. September is the strongest month on record, led by the Top-Single.
Full transparency
Three ways the same engine can be traded, shown side by side, including the one that doesn't flatter us. Books charge vig on every parlay leg, so the parlay is deliberately hard mode; the singles are where the edge lives.
| Period | All-Singles | ROI | Top-Single | ROI | Parlay |
|---|---|---|---|---|---|
| July 2026 (from 18th) | 15–10 · 60% | +45%+9.1% yield | 8–5 · 62% | +56%+21.6% yield | 5–8 · 38% |
| August 2026 | 54–23 · 70% | +159%+10.3% yield | 17–12 · 59% | −23%−4.0% yield | 10–19 · 34% |
| September 2026 | 55–22 · 71% | +98%+6.4% yield | 22–6 · 79% | +106%+18.9% yield | 10–19 · 34% |
| October 2026 (to 13th) | 21–10 · 68% | +78%+12.5% yield | 7–6 · 54% | +11%+4.1% yield | 5–8 · 38% |
| Cumulative | 145–65 · 69.0% | +380%+9.1% yield | 54–29 · 65.1% | +149%+9.0% yield | 30–54 · 36% |
ROI = how much a flat-20%-a-pick bankroll grew that period (same % at any size: $1, $10K or $100K); the small grey figure is yield (profit per $1 staked, the sizing-independent number). Cumulative is the full-run $100K journey. Top-Single = betting only the engine's single highest-conviction pick each day, higher variance, smaller sample (note August). All picks are timestamped at entry price and settled automatically. The complete leg-by-leg ledger, dates, markets, entry prices, results, is provided under NDA.
The complete pick ledger
All 212 Live Test #1 picks, grouped by the day they were placed, with entry price, result, and game time (UTC). No cherry-picking, no survivorship, the full record, so the only open question is what to do with it.
Read it as a build story. The first two months were openly trial-and-error, the engine finding its footing (July even drew the live bankroll down). By September, structure emerged: data-integrity hardening, the upset-risk discovery, and a settled selection process, and the numbers stepped up across the board. This is an edge at its inflection point. The raw material is here and it is real; the ceiling is what a team with capital, venue access and engineering can build it into. Here is what it is, and what you can make it become.
How to read this. All-Singles = betting every one of the engine’s daily picks as a flat single. Top-Single = betting only the engine’s single highest-conviction pick that day. Every pick is generated at a fixed time each morning (11:00 ET), hours before the games begin. The times below are game start times in UTC, so the gap between a pick’s generation and its game is the proof: every call was locked in well in advance. No lookahead, no hindsight.
Times in UTC. Nothing here is selected or smoothed: this is the engine’s complete daily pick record, every selection timestamped at entry and settled automatically. The flat-stake ROI shown per month is the performance measure. By design, WAYL ENGN is a pure signal, built to be run at scale by a capitalized partner, not the founder.
Where the edge concentrates
Across 210 graded singles, performance isn't uniform, it clusters. That's a feature, not noise: it tells a partner exactly where to weight capital and where the model earns its keep.
Win rate by sport across all tracked singles (main verticals shown). Soccer is the highest-volume vertical; NFL the highest win rate on a smaller sample. Profitability depends on entry price, not win rate alone.
What changed, and when
Full transparency isn't only the results, it's the changes behind them. Here's the honest record of what moved and when.
The engine goes always-on. Every daily pick recorded and settled in real time from this date, the start of the track record above.
Selection logic deliberately held stable through the data-collection window, no tuning to chase results. The August → September improvement is a maturing system on more data, not a curve fit.
Fixed a soccer settlement-resolver edge case and void handling; grading now re-runs on every refresh. Records became fully accurate in the highest-volume vertical.
Identified that losses cluster on a specific, measurable pre-bet condition. A flag was built; out-of-sample validation is in progress. Early read: separating flagged vs. clean picks sharply improves win rate.
Auto-filter on the validated upset-risk signal, real-time data feed, and vertical expansion. There is identified edge still on the table, disclosed under NDA.
Why now
This isn’t a bet on the future, it’s happening now, and the biggest names on Wall Street are moving first.
The funds entering prediction markets aren't new, many are 5+ year-old crypto/quant shops adding this asset class right now. It's an active land-grab for who owns the best signal first.
The known players (e.g. BlueWalker-class funds) compete on mispricing/arbitrage. WAYL ENGN is a different, complementary edge, a directional selection signal, that a desk can run alongside, not against, its existing book.
This is already built and running 24/7 across multiple markets, with a graded track record. A partner supplies capital, venue access, and scale; the engine supplies the signal.
The dataset is early, and that's the opportunity. You're seeing an edge as it forms, before it's obvious or crowded, at a valuation that reflects that.
What it is
At the category level: WAYL ENGN reads where informed capital concentrates across prediction-market venues and converts it into ranked, daily, directional selections, fully automated, always-on, and self-grading. The features, signal construction, and selection logic are proprietary and disclosed only under NDA. The results above are the public tearsheet; the mechanism is the asset.
Everything above is public. Under NDA, a serious counterparty gets the full picture to run their own diligence:
Ways to engage
Open to any of the following, whichever fits your mandate.
Full purchase of the engine and its IP.
Deploy it under license or a revenue-share; ownership stays with the founder.
Fund a vehicle that runs the engine as its edge.
Run the signal alongside your existing prediction-market desk.