A prediction market with a verifiable reputation layer. Every bet weighs money × reputation, and reputation only comes from being right when the crowd was wrong.
Prediction markets are failing in four specific ways. Each one has a direct answer here.
In Polymarket's $7M Ukraine minerals market, one UMA token whale reportedly cast about 25% of dispute votes and forced a premature “Yes”. Source ↗
KleosResolvers are gated by earned reputation, not purchasable tokens.
A Wall Street Journal investigation reportedly found nearly 20% of disputed Polymarket outcomes involved oracle voters with stakes in the market they judged. Source ↗
KleosAnyone with a position is excluded from resolving that market.
One trader spread more than $45 million of Trump bets across several Polymarket accounts. Source ↗
KleosA reputation-weighted price shown next to the money price, with a 10% per-address cap.
In a federal criminal case, a soldier is charged with using classified information to bet about $33,000 on the Maduro operation and cashing out about $400,000. Source ↗
KleosReputation vests over time, gains are capped per market, anomalies are reviewed.
Small crowds of proven forecasters outperform larger crowds in both markets and polls. Atanasov, Witkowski, Mellers & Tetlock, International Journal of Forecasting, 2025 ↗
Across 5,456 Polymarket and Kalshi markets, bigger bettors were not more accurate; in some markets whales had negative edge. Daleep et al., 2026 ↗
Every bet can carry a call: a public one-line reason, written before the outcome and scored after it. Follow the traders who call it early, and see who was right, not who was loudest.
Kleos is the Greek word for the renown a person earns through their deeds, carried by what others say about them. Here that renown is your track record: public, non-transferable and impossible to buy.
Feeds reward the loudest voice and markets reward the biggest wallet. Kleos shifts attention to people who called it before the crowd did, and shows their read next to the money's read.
Every reputation change is an on-chain event with the math attached. Look up any trader, retrace every point they earned or lost, and verify it on Monadscan. No private scores, no black box.
Agreeing with the crowd earns almost nothing. Being early and right earns a lot. Being loud and wrong costs more than it pays. The incentives favor independent thinking over hype.
The wreath fills in dot by dot as your tier rises. It shows up next to your address everywhere: bets, leaderboards, profiles.
Move the sliders. These are the same formulas the contract runs.
M = min(1 + α·√R, 5) α = 0.51 + 0.5·√16 = 3.00. The square root means diminishing returns: 1× → 3× takes R = 16, but 5× takes R = 64. Nobody becomes king of the market.
E = m · M P_money = YES$ / total$ P_rep = YES_E / total_EEach sharp: $80 × 3.00 = $240 of weight. The whale: $2,000 × 1 = $2,000. Same market, two readings. The gap is what Kleos adds: what people who are usually right think, not just who has the most money.
payout = m + (m / Σwinning$) · (L − 3%) bonus = pool · ΔR·M / Σ(ΔR·M)Principal: $80 + ($80 / $1,600) × $1,940 = $177. Of the losing side, 2% ($40) becomes a bonus for winners who gained reputation, weighted by ΔR × M: $2 each. Money pays money, so a newcomer on the same side gets the same principal per dollar. Reputation steers the signal and earns the bonus, and 1% is protocol revenue. Nobody prints money: everything paid out was bet.
Right: ΔR = +k·s·(1 − p) Wrong: ΔR = −λ·k·s·p λ = 1.5+1 · 1.00 · (1 − 0.30) = +0.70You earn reputation for how much you surprised the market. Agreeing with everyone and being right is worth almost nothing. Following the herd and being wrong hurts a lot. Stakes under $50 count partially, so nobody farms reputation with dust bets.
E[ΔR] = q·(1 − p) − λ·(1 − q)·pWith λ = 1 the expected gain is exactly q − p: you only win reputation by knowing something the market doesn't. With λ = 1.5 you need a real edge, not a coin flip. Lying or guessing has negative expected value.
boost stops at 10% of ΣE (pools ≥ $5,000)Once a market is deep, no address can gain extra weight beyond 10% of the reputation-weighted pool, not even the best sharp. The cap trims the boost, never the money you put in.
Votes are hidden until the reveal, but a juror can prove their vote afterward, so open bribes are possible. Random draws and reputation at stake make it costly, not impossible. Jury draws use block randomness, which block producers can nudge.
Thin markets move sharply with each bet. LMSR is the planned upgrade for thin markets.
The anomaly flag is a heuristic, not proof of insider trading.