Find the mispriced markets

The Alpha Scanner runs every 30 minutes, comparing the AI’s probability estimate against the market price for every active market. When it finds a meaningful divergence, it calculates the expected value, Kelly fraction, and packages the full reasoning chain into an actionable edge card.

How it works

AI analyzes every market

The multi-agent system produces a probability estimate for each active market (tiered by volume/importance).

Edge calculated

For each market: edge = AI probability - market probability. A 72% AI estimate vs 45% market price = 27% edge.

EV and Kelly computed

Expected value per $1 bet and Kelly criterion optimal position size are calculated from the edge and odds.

Edges ranked and classified

Edges are ranked by conviction, edge size, and EV. Classified as high conviction, moderate, or speculative.

Edge card anatomy

Each edge card shows:

Component	Description
Edge badge	”27% EDGE” — color-coded by conviction level
Market question	The market being analyzed
The numbers	”AI says 72%” vs “Market says 45%” — visual comparison
EV display	”+`$0.60` EV per `$1`” with Kelly optimal sizing
Reasoning summary	2-3 sentence thesis (expandable to full reasoning chain)
Key factors	Bullet points of supporting evidence
Agent agreement	”4/4 agents agree” with confidence bar
Risk factors	Collapsed by default — the bear case
Trade button	”Buy YES @ 45¢” with pre-filled amount

Classification

Level	Criteria	Meaning
High conviction	Edge ≥ 25%, agreement ≥ 3/4, confidence ≥ 75	Strongest setups — agents agree, evidence is clear
Moderate	Edge ≥ 15%, agreement ≥ 2/4, confidence ≥ 50	Decent opportunity with some uncertainty
Speculative	Below moderate thresholds	Interesting but uncertain — smaller position sizes

The math

Expected Value
Kelly Criterion

EV = (AI probability × payout) - (1 - AI probability × cost)

Example: AI says 72%, market price 45¢

Cost per share: $0.45
Payout if right: $1.00
EV = (0.72 × $1.00) - (0.28 × $0.45) = $0.72 - $0.126 = +$0.594 per share

Kelly fraction = (edge / odds) = (AI prob - market prob) / (1 - market prob)

Example: AI 72%, Market 45%

Kelly = (0.72 - 0.45) / (1 - 0.45) = 0.27 / 0.55 = 49%
Capped by user’s max-per-trade setting (typically much lower)

Kelly tells you the mathematically optimal fraction of your bankroll to bet. In practice, half-Kelly or quarter-Kelly is safer.

Filtering edges

Filter	Options
Sort by	Confidence, Edge %, EV, Kelly fraction
Edge type	High conviction, Moderate, Speculative
Category	Politics, Crypto, Sports, Finance, etc.
Min confidence	Slider (0-100)
Min edge	Slider (0-50%)

Most profitable setup: High conviction edges in high-volume markets. The liquidity ensures clean execution, and agent agreement means the evidence is strong. Sort by EV, filter to 75+ confidence.

Edge ≠ guaranteed profit. The AI can systematically misjudge novel situations. Always read the reasoning chain and risk factors before sizing up. The edge card shows you everything you need to form your own view.

Documentation Index

​Find the mispriced markets

​How it works

​Edge card anatomy

​Classification

​The math

​Filtering edges

Find the mispriced markets

How it works

Edge card anatomy

Classification

The math

Filtering edges