engineering
I Built a Sports Prediction Bot That Bets Against the Market — Here's the Architecture
Mar 4, 20268m
Concept artwork · not a product screenshot.
NBA / MLB
Sports
Six factors
Model
Polymarket
Markets
Outcome logs
Evaluation
// What It Does
Shiva explores NBA and MLB prediction markets by comparing a model estimate with the market price. Its documented inputs include sports statistics, schedules, injury context and market data.
The six-factor design considers home advantage, rest, recent form, injuries, head-to-head history and media sentiment. An edge estimate then feeds sizing and risk checks. A difference between model and market probabilities is an estimate, not proof of a profitable opportunity.
Recording evaluated markets and resolved outcomes supports calibration and model review. This case study describes NBA and MLB coverage; it does not claim verified NFL support, daily model updates or a measured profitability record.
// Illustrative Interface
Sample data · not connected to a live service. Displayed trades and P&L are illustrative, not verified results.
SHIVA · SPORTS EDGE
min edge ≥5% · 2026-03-04MATCHUP
SPORT
MODEL
MARKET
EDGE
SIGNAL
Lakers vs Celtics
NBA
62%
$0.52
+10%
BET YES
Chiefs vs Ravens
NFL
55%
$0.58
-3%
SKIP
Thunder vs Nuggets
NBA
71%
$0.60
+11%
BET YES
Yankees vs Red Sox
MLB
48%
$0.53
-5%
SKIP
adaptive ML · updates post-game● 2 signals today
// Architecture
- 01 →Sports data
- 02 →Probability estimate
- 03 →Edge & risk checks
- 04 ✓Outcome evaluation
Sports statistics and market prices feed a probability estimate. Estimated edge and position-sizing checks inform execution; recorded outcomes support model evaluation.
// Related Reading