Each model's prediction for the target minute. Weight = how much the blend trusts it, based on its average miss at this distance.
Kalshi odds by price bucket
Chance BTC settles in each $100 range
How this prediction has moved
Every re-evaluation for the selected minute
PredictionBTC price
!
What the research says about betting propsand how the picks below use it
Cheap longshots lose on Kalshi. Across Kalshi markets, 5¢ contracts lost about 60% of what was put in; people taking the posted price lost ~32% on average versus ~10% for people resting limit orders. So YES picks under 15¢ are ranked lower, and the ticket defaults to "wait at my price". Source: Bürgi, Deng & Whelan (CEPR summary).
Unders have historically hit more often. 61% of season-long NFL player props went under in 2021–22 (4for4): injuries, benchings and blowouts only push numbers down. It swings, though: one tracker's rushing unders went 111–64 one season and 12–24 the next (Stealing Lines). On Kalshi, buying NO on "k+" is the under; those picks are tagged.
Props are priced less carefully than game lines. Books hold roughly 4–6% on star props and 6–10%+ on role players, and sharp money pays them less attention (Wizard of Odds). That's where edges exist, and also why the hold is higher.
How the score works. Each prop starts from the player's own rate (this season + last season, pulled toward league average when data is thin), then adjusts for the opposing pitcher or defense, platoon (lefty/righty), ballpark, temperature and wind, home/road, day/night, weekday, month, recent form, game script, rest and injury reports. Small splits like "Tuesdays in October" are shrunk hard, so they barely move the number. That stats estimate is then blended with Kalshi's own price (30–70% weight on the stats, by confidence), because the market often knows lineup and injury news first. Score 50 = fairly priced; 60+ Good, 70+ Strong. Click any row to see every factor.
Risk level (top of the page). LOW shows exactly the picks this tab always showed (score 60+). MEDIUM adds solid-edge picks scoring 55+ with at least a 35% hit chance. HIGH and EXTREME add cheaper, less likely picks, but only when the edge beats the longshot penalty (cheap contracts are overpriced on average) and the payout is at least 1.15× / 1.5× what the price implies. Each extra pick is tagged with its tier. Game-time injury risks are never picks at any level.
Injuries and lineups. Players on the injured list, sent down, ruled Out or Doubtful are removed. Questionable players, starters who aren't confirmed, and players who haven't played in several days are held below 50 so they're never picks. Until a team posts its lineup (usually 2–4 hours before first pitch), MLB hitter scores are halved toward 50, since he could get the day off. Always check news before betting.
Where the data comes from. Bets and prices: Kalshi only. MLB stats, lineups, probable pitchers and game-day weather: MLB's official Stats API. NFL stats, schedules and injury reports: nflverse. NHL schedule, live scores, standings and game logs: the NHL's public API. Forecasts: Open-Meteo. No sportsbook odds are used. NBA and WNBA aren't scored yet (no free stats feed).
Loading props…
Kalshi lists props as ladders ("Player: 50+ rushing yards"). YES pays $1 if the player reaches the number; NO pays $1 if he doesn't. Hit chance = our stats estimate blended with Kalshi's price. Edge = hit chance minus the price you'd pay, in cents per $1 contract. Only two-sided markets with a spread of 6¢ or less are shown, for games in the next 3 days. NHL props are scored from NHL game logs; NBA and WNBA props are listed without a score (no free stats feed).
Loading games…
Prices are Kalshi's game-winner markets (a YES contract pays $1 if that team wins). "Win chance" is our own model (MLB: runs scored and allowed plus both probable starters; NFL: points for and against, home field and rest; NHL: goals for and against, home ice; NBA/WNBA: Kalshi price only) blended with Kalshi's price. No sportsbook odds are used. During a game there's no model; the market has the score. A "value" tag needs at least a 3¢ gap to the ask with a tight spread.
!
Loading games…
Refreshes every 15 seconds. Scores and game state come from each league's own feed where available; the odds bar is Kalshi's mid price (YES chance to win). Click GAME on any pick in the Sports tab to jump here; hover it for a mini gamecast.
Connect your Kalshi account
Create an API key on Kalshi (Account & security → API Keys) and save the private key file.
Put it in the dashboard's keys folder as kalshi_key.pem.
In keys/kalshi.env, paste the Key ID after KALSHI_KEY_ID= and set KALSHI_ENV=demo or live.
Double-click Start Dashboard again.
Cash—
Positions—
Open P&L—
Resting orders—
Buy or sell from the BTC or Sports tabs; everything you hold shows here.
Positions
BTC and sports, marked at the price you could sell for right now
BTC auto-trader
Idle
TRADING LOCKED
Bot P&L (only its own trades)
$0.00
Risk budget
Time left—
Open—
Trades · won—
Avg per trade—
Settings
Then it holds what's open until it settles
Most it can lose if every open position lost
Most a session can lose; $10 max
Strategy (fixed): buy only the favored side at 70–95¢, with a resting bid (no fee), only when our BTC model puts it at 80%+ and at least 3¢ above the price; one position per settlement hour; hold to the $1 payout; sell early only if the model drops it below 45%.
Why this strategy: a study of Kalshi's full trade history (Bürgi, Deng & Whelan) found contracts priced 70–99¢ earned small positive returns on average (about +2–3%), cheap contracts lost 20–60%, and traders resting limit orders did far better than those paying the asking price (−10% vs −31%), with the pattern strongest in crypto markets. So the bot only rests bids on favored sides, never pays the fee, and collects at settlement instead of flipping for a cent. It's a small edge with real swings: one loss costs about as much as several wins. Every closed trade is saved to data/trades.csv. It never touches contracts you already held.
For information and entertainment only: not financial, investment or betting advice. Forecasts and picks can be wrong. Market prices from Kalshi; stats from MLB, nflverse and the NHL; weather from Open-Meteo. Bet responsibly, 21+ where legal.
—
Live price streams from Coinbase. Kalshi's hourly markets settle on the CF Benchmarks BRTI (60-second average), so expect small differences from Coinbase.
Over minutes, Bitcoin is close to a random walk: the blend's edge comes from Kalshi's crowd pricing and short-term order flow, and the models table shows honestly whether each one beats simply guessing "no change". Not financial advice.