Eight independent detectors, applied to every market with sufficient pricing data. No black boxes, no subjective calls — just signals that can be measured, audited, and graded against the closing line.
Every market with sufficient pricing data is continuously scored by the same eight detectors. Each detector is a binary signal — it either fires or it does not — based on deterministic, quantitative criteria. There is no committee, no override, and no editorial judgment layered on top. The same inputs produce the same output, every time.
A pick only reaches the dashboard when enough detectors agree:
This multi-detector architecture is deliberately conservative. A single detector firing on its own — even one as informative as edge magnitude — is not enough to commit capital. Agreement across multiple independent views is what separates a real edge from a lucky tick.
Each detector measures one specific, narrow property of a market. The signal is the intersection.
Measures the bid-ask spread on the market at scan time. A tight spread indicates price discovery has converged and liquidity providers are actively competing.
Wide spreads signal either one-sided flow or an illiquid market where any fill price is suspect — we penalize these markets because any realized price is likely worse than the quoted mid.
Confirms that enough contracts have changed hands for the current price to be trusted. Markets with only a handful of trades are effectively priced by a single opinion.
The detector fires when cumulative volume crosses a per-category threshold calibrated against historical outcomes — high enough that the quoted price reflects competitive trading.
Flags a sudden surge in trading activity relative to the market's recent baseline. Volume spikes typically precede or accompany the arrival of new information.
When a spike fires in combination with the edge magnitude detector, it usually means the market is actively repricing — and we want to be on the right side of that move before it completes.
Quantifies the gap between the current market price and our fair-value estimate, derived from cross-platform consensus and model-based priors.
The detector fires when the gap clears ≥3¢ raw edge OR ≥5% of contract price, whichever is more conservative — the threshold where post-fee, post-slippage returns remain positive under reasonable assumptions about execution quality.
Examines the order book beyond the top-of-book quote. A $100 quoted price with only $20 of size behind it is functionally a $100 market for the first trader and a $105 market for everyone after.
The detector fires when at least one standard clip size can be executed at or near the displayed price — the edge has to be realizable, not theoretical.
Gates signals by time-to-expiry. A 50bp edge on a market that resolves in six hours is very different from the same edge on a market that resolves in six weeks.
The detector fires when there is enough runway for the edge to play out — typically a minimum of several hours — and penalizes markets inside the final-minutes window where settlement risk dominates.
Compares the market under review to every other platform offering the same or an equivalent outcome. When prices diverge meaningfully across platforms, at least one of them is wrong.
This is the core cross-platform signal. A 53¢ Kalshi contract trading alongside a 47¢ Polymarket contract for the same event is a discrepancy — and a candidate for both directional and arbitrage plays.
Looks for prices that have diverged from their recent moving average without a clear catalyst. Prediction markets, like every other market, exhibit short-term reversion to the mean.
The detector fires when the current price sits more than a calibrated number of standard deviations away from its rolling baseline — conditions under which a snap-back is statistically favored.
Every pick we publish is logged at its entry price and graded against the closing line once the market settles. Closing Line Value — the difference between the price we entered at and the price the market ultimately converged to — is the single strongest predictor of long-run profitability in sharp trading, precisely because it does not require a large outcome sample to be statistically meaningful.
A short-term win rate can flatter a poor process. Hit-rate variance is high enough over any 100-pick window to make bad strategies look fine and good strategies look broken. CLV strips that variance out: if we consistently enter at prices better than the closing line, the P&L will follow, and if we don't, we can identify and correct the leak long before it shows up in the ledger.
Our public track record page surfaces both: per-pick CLV at settlement and the cumulative realized P&L. Both update automatically as markets resolve.
Our scanner ingests prices from 25 platforms — prediction markets (Kalshi, Polymarket, and peers) alongside regulated sportsbooks (DraftKings, FanDuel, BetMGM, Pinnacle, ESPN BET, Fanatics, and 11 more). Each feed is normalized into a common market schema so the same detector stack can evaluate a political contract, an NFL spread, a crypto price target, and a CPI print on equal footing.
Prices arrive via sub-second Server-Sent Event streams where the platform supports them, with a polling backstop for the handful that do not. The entire grid re-scores on every meaningful price move.
Running alongside the detector grid is a dedicated arbitrage scanner. It enumerates every pair of quotes on equivalent outcomes across the 25 platforms and computes whether a risk-free spread exists once fees, slippage, and realistic fill sizes are accounted for.
Every arbitrage opportunity is validated through the same math before it reaches the dashboard — if the spread does not hold up under execution-realistic assumptions, it never ships. The arbs you see are the arbs you can actually take.
Every card on the dashboard surfaces its full detector state, so you can always see why a market did — or did not — make the cut.
Launch DashboardSharp prediction market analyst
I compare odds across DraftKings, FanDuel, BetMGM, Kalshi, and 25 platforms total to find you +EV plays and arbitrage opportunities.