Kalshi · Sports · Prediction markets

Research first. Execute second.

Edge-Radar compares Kalshi's price against an independent fair value built from sportsbook consensus and public data feeds. A bet fires only when the gap clears the exchange fee and every risk gate, and it is sized by fractional Kelly.

Dry run by default. Not financial advice: the repo documents its losing segments as carefully as its winning ones.

Source github.com / Edge-Radar →

Risk gates

Pipeline stages

Sport filters

Agent roles

How It Works

Pull Kalshi's market price. Pull an independent fair value from sportsbooks / data feeds. If the gap, net of the exchange fee, clears 21 risk gates, a Kelly-sized bet fires through the Kalshi trade API.

fair_value − market_ask = edge · when edge ≥ threshold & gates pass → bet

Reference Price

independent fair values, pulled from public APIs

The Odds API

GET

api.the-odds-api.com /v4/sports/…/odds

8–20 sportsbook lines → no-vig consensus

Sports stats

statsapi.mlb.com

api-web.nhle.com

site.api.espn.com

Crypto / weather / equity

api.coingecko.com

api.weather.gov

query1.finance.yahoo.com

Edge Engine

the seven-stage pipeline that finds & sizes the bet

Kalshi market data

GET

api.elections.kalshi.com /markets, /orderbook

live ask, bid, volume, expiry

Fair-value model

ticker → ML · SP · TO · FU · PR

de-vig, Poisson totals, Elo, hold-out

Edge calculation

fair − ask, with confidence tier

edge ≥ 3% + fee (per-sport floors)

21 risk gates

daily loss · per-event cap · series dedup

NO-favorite floor · min price · confidence

edge ceiling 50% · exposure caps · 10% hard stop

Kelly sizing

fractional Kelly · capped at MAX_BET_SIZE

half-Kelly on cheap NO bets

Execute & Track

wagers, settlement, calibration loop

Kalshi trade API

POST

api.elections.kalshi.com /portfolio/orders

RSA-PSS signed limit order

Position & fills

GET

/portfolio/positions · /fills

track exposure · trim stale orders

Settle & calibrate

27-field settlement record

Brier score feeds back into thresholds

Notify

email reports · daily digest

daily integration-drift check

Outcomes loop back: settled results re-calibrate thresholds & confidence tiers

DRY_RUN=true default · 21 gates · spec

Pipeline

Seven sequential stages. Scan → Size → Execute → Settle.

Fetch

Kalshi + Odds API

Categorize

ticker → model

Compare

fair vs ask

Cap

per-event dedup

Risk

21 gates

Execute

RSA-signed limit

Monitor

settle + calibrate

Risk gates · CLAUDE.md § Execution Gates Edge models · ML · SP · TO · FU · PR Settler · make settle

Coverage

What the scanner prices. Each segment carries its own edge floor.

Sports

NBA · NHL · MLB · NFL · NCAA · MLS · soccer · UFC · boxing · F1 · NASCAR · PGA · IPL · tennis · esports

Moneylines, spreads and totals

Prediction

Crypto (BTC, ETH, XRP, DOGE, SOL) · weather in 13 cities · S&P 500

Off by default behind ALLOW_PREDICTION_BETS

Futures

NFL · NBA · NHL · MLB championships · PGA majors

Priced from sportsbook outright odds

Quick Start

Python 3.11+. A preview risks nothing, and DRY_RUN=true is the default even with --execute.

git clone https://github.com/ai-automation-tools/Edge-Radar.git
cd Edge-Radar
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env            # add Kalshi + Odds API keys

python scripts/doctor.py        # validate the environment
python scripts/scan.py sports --filter nba --date today   # preview only