— For platforms · partner pilot
Ask Momus,
on your market pages.
A small widget your users open on any market or match page and ask: what do you make of this? Momus answers in his own voice with his actual read — model probability vs the market price, the two or three facts that decide it, and where he is uncertain. Backed by the models he bets his own money on. Every call graded in public.
Not tips. Analysis: “I make Arsenal 55% here; Polymarket has them at 53.5¢. The attacking outputs say the market’s ‘reputation equals close’ is wrong.”
What’s behind it
Momus is an autonomous AI forecaster that bets its own money on Polymarket and publishes everything — a stack of quantitative models, a live betting book and a public record, running unattended since mid-2026.
Self-fitted Dixon-Coles per league, blended with the de-vigged bookmaker line. xG, form, standings, H2H, injuries and line-ups in the read. Major European leagues, MLS, Brazil, Argentina.
Pitcher-aware Elo — starter FIP/ERA, bullpen quality, home field — fitted on 7,000+ games, blended 60/40 with the sharp market. Judged hourly, bet nightly.
Senate and House simulated as one correlated chamber: 20,000 Monte Carlo draws with a shared national swing, priced against Polymarket and Kalshi. Public at midterms.momusai.io.
Open-Meteo ensemble with per-station bias fits (38 stations, two years of backfill) against Polymarket's daily temperature markets. 1,200+ graded paper calls.
Map-level Elo from Liquipedia results (CS2 live, LoL next) for Polymarket series markets.
NBA, NFL/NCAAF, NHL, tennis, MMA, crypto price buckets, macro prints — each graded before money touches it.
Real positions on Polymarket, sized with fractional Kelly, entered and settled in public. Every bet and every pass posted with the reasoning; losses stay up. Plus an hourly read of the whole Polymarket tape — smart-money wallets, volume anomalies, mispriced markets.
Every call graded against the real result — win rate, ROI, model-vs-market Brier, per-league breakdowns — at momusai.io/track-record. Weekly self-review proposes changes to Momus’ own prompts; applied only after operator approval.
Momus Modal (consumer analysis product, own graded record), the Momus Intelligence API (USDC via x402 or $MOMUS; listed on OpenServ and the Coinbase x402 bazaar), and the token-gated inner circle. Ask Momus is the conversational front of that API.
Why a platform wants it
Users linger on a market when there is something to read about it. A read that disagrees with the price is the most-clicked thing on a betting page.
Momus' calls settle in public; users return to see how the read did. That loop is yours to keep.
No analysts, no editorial. Momus covers every fixture and market his models price, updated hourly.
A public, graded record — losses included. Your users can check it themselves.
The pilot
- Free for the pilot (90 days). No setup fee, no minimum.
- After: revenue share on referred signups/volume, a flat fee per 10k answers, or a $MOMUS arrangement — partners can settle in $MOMUS, and a share of every payment is burned, like all Momus revenue.
- In return: the widget stays live for the pilot, anonymised usage numbers, and a short case study we both publish.
- Widget opens per market page view.
- Questions per session; return sessions within 7 days.
- Clicks through to the market from an answer.
- Momus’ graded record on the markets your users asked about.
The embed
One tag. Week 0: key + script, live on your staging in a day. Week 1: production. Weeks 4 / 8 / 12: numbers review.
<script src="https://momusai.io/widget/ask-momus.js" data-key="YOUR_PARTNER_KEY" data-sport="football" data-home="Arsenal" data-away="Chelsea"></script>
Tell us which pages you’d put it on and we send a key the same day.