Cross-Market Arb Bots Surge; PMXT Hits 1,900 Stars
GitHub activity on 2026-06-23 reveals a concentrated wave of Polymarket tooling centered on cross-market arbitrage, particularly between Polymarket and Kalshi BTC short-interval markets. The unified prediction market API PMXT has reached 1,913 stars and 225 forks, signaling it has become infrastructure-grade tooling that serious operators should be tracking.
GitHub Pulse — 2026-06-23
Lead Story: PMXT Is Now Infrastructure
The most significant signal in today's feed is not a new repo — it's the maturity of an existing one. **pmxt-dev/pmxt** (1,913 stars, 225 forks, 9.2MB) bills itself as "CCXT for prediction markets," providing a unified API layer across Polymarket, Kalshi, and additional venues. At this star count, PMXT has crossed the threshold from hobbyist library to de facto standard. Any operator building execution infrastructure in 2026 who is not evaluating PMXT as a dependency is doing redundant work. The repo was updated today, indicating active maintenance. Watch for forks and downstream bots that consume PMXT as a substrate — several in today's feed likely do.
Dominant Theme: Polymarket/Kalshi BTC Cross-Venue Arb
Three independent repos published or updated today target the same narrow opportunity: price discrepancies between Polymarket and Kalshi on short-interval BTC direction markets.
- **Juanp2389/Kalshi-trade-bot** explicitly describes "spotting price gaps" between the two venues on BTC 15-minute markets and executing paired trades in TypeScript.
- **Corettafinnougricspeaking368/Polymarket-Arbitrage-Trading-Bot-Spreadmaker** targets the same 15-minute crypto markets with hedged two-sided spread trades and a configurable entry cost ceiling.
- **kenfri13/polymarket-arbitrage-trading-bot** automates 15-minute Up/Down market arb via the Gamma API with hedge thresholds and risk management controls.
The convergence of three independent builders on the same structure in the same week is a meaningful signal. Either this arb is newly viable due to recent liquidity changes, or tutorial content is driving a wave of copycat implementations. Either way, if this edge exists, crowding will compress it quickly. Operators already positioned here should monitor spread tightening.
Notable Individual Repos
**nayrbryanGaming/arbiter** stands out for its discipline. The description explicitly states it "refuses to trade without a backtested out-of-sample edge" and defaults to paper trading. It hunts no-arbitrage violations between related election markets and sizes with fractional Kelly. This process-first framing is rare in the GitHub prediction market space, where most repos skip validation entirely. Worth a deeper read for methodology, even if the codebase is small (287KB).
**BallesJr/polymarket-weather-edge** is the most niche strategy seen in this feed: automated paper trading targeting the NO side of overpriced Polymarket temperature markets, using live METAR aviation weather observations and Open-Meteo forecast data fed into a Gaussian probability model. The author also maintains **polymarket-edge-model**, a Random Forest calibration pipeline with backtesting. Two repos from one builder, both methodologically coherent, suggests someone doing real research rather than tutorial-following.
**adibfarah569/PolyTrace** describes itself as a "whale copy-trading terminal" and is explicitly labeled DEMO. It is 198KB and two days old. Treat as early-stage. Whale copy-trading tooling is a category with real potential on Polymarket given the public on-chain address data, but this implementation is too new to evaluate.
Security Alert
**quipmnxailcrrgky/tradingbot** (98 stars, 60 forks, 34MB) and **bit-nexusxtitmtdsuy/Polymarket_Bot** (32 stars, 17 forks, 36.6MB) both carry randomized or obfuscated account names, large binary-heavy repo sizes inconsistent with their described functionality, and high fork counts relative to stars. These are consistent patterns seen in credential-harvesting or wallet-draining packages distributed through GitHub. **Do not clone or run either repo without full static analysis in an isolated environment.** Treat as hostile until proven otherwise.
**King102681/polymarket-bot** has no description and is 311MB — an extraordinary size for a trading bot. Flag for inspection.
Bottom Line
The Polymarket tooling ecosystem is maturing fast on two tracks simultaneously: disciplined quant infrastructure (PMXT, arbiter, BallesJr repos) and low-effort or potentially malicious clones competing for the same search visibility. Operators sourcing open-source components need tighter vetting hygiene. The Kalshi/Polymarket BTC arb cluster warrants a liquidity audit — if the edge is real, it is already crowded.