Bot Surge: Weather Models, NegRisk Arb, Cross-Platform Plays

GitHub Bot Landscape — 2026-06-19

Volume and Context

Today's sweep returned 19 repositories with Polymarket exposure, all updated within the last 24 hours. The cohort splits cleanly into three categories: legitimate tooling with real engineering depth, low-signal personal projects, and outright suspicious repos. Operators should calibrate accordingly.

---

Notable Repos Worth Watching

**`BallesJr/polymarket-weather-edge`** is the most operationally interesting find today. The project automates NO-side bets on temperature markets using live METAR observations and Open-Meteo forecast data, applying a Gaussian probability model to identify overpriced YES positions. This is a specific, testable thesis: weather market liquidity is thin and retail participants likely overprice tail outcomes. The same author also maintains `polymarket-edge-model`, a Random Forest calibration pipeline with backtesting. Two repos from the same operator in complementary domains suggests someone building a systematic multi-vertical approach. Watch this account.

**`matthewnyc2/arbitrage`** describes a production-grade async Python scanner for NegRisk multi-outcome arbitrage with 60 passing tests and a paper-trading mode. NegRisk arb — exploiting mispricing across correlated outcome sets in the same market — is underexplored relative to binary arb. The test suite and paper-trading infrastructure suggest genuine engineering investment rather than a demo project.

**`tduic/prediction-market`** targets cross-platform arbitrage between Polymarket and Kalshi with Kelly sizing and an embedded analytics dashboard. Cross-platform plays require careful latency management and account for differing resolution criteria, which is a meaningful operational challenge. No star count suggests this is early-stage or private-use tooling rather than a distributed product.

**`hot3246624/pm_as_ofi`** implements Avellaneda-Stoikov market-making combined with Order Flow Imbalance (OFI) signals. OFI as a predictor of short-term price movement is well-documented in equity microstructure literature; its application to prediction markets is less common and potentially high-value given the lower sophistication of typical counterparties. Nine stars and one fork is modest traction but the strategy specificity warrants a closer read of the codebase.

---

Security Alerts

**`jelllott/polymarket-kalshi-arbitrage-bot`** should be treated as a scam repo. The description is raw keyword stuffing — the phrase "polymarket kalshi" repeated 18 consecutive times. More critically, the fork count of 5,788 against only 217 stars is statistically anomalous and consistent with artificial inflation or a fork-farming scheme designed to signal legitimacy. Do not run any code from this repository.

**`radioman/polymarket-arbitrage-trading-bot`** has a 2016 creation date paired with a description that is also keyword-stuffed and repetitive. The 737MB repo size is outsized relative to any trading bot codebase and may indicate bundled executables or obfuscated payloads. The star/fork ratio (474/131) looks cleaner, but the combination of age, size, and description pattern warrants caution. Avoid execution without full sandbox review.

**`quipmnxailcrrgky/tradingbot`** (98 stars, 60 forks) has a randomized username consistent with bot-generated accounts and a generic description. Elevated engagement on an account with no readable identity is a common distribution vector for credential-harvesting bots. Do not connect API keys or wallet credentials.

---

Broader Market Intelligence

The concentration of BTC/ETH 5-minute prediction market tooling — visible across at least four separate repos today — confirms that short-duration crypto price markets remain the primary automated battleground on Polymarket. Operators holding positions in these markets should assume significant bot-driven order flow, particularly around round-number price levels and scheduled macro events.

The emergence of weather market bots (`polymarket-weather-edge`) is a meaningful signal. If METAR-based automated traders are entering temperature markets, human operators without access to live meteorological data are at an information disadvantage in those markets going forward.

Cross-platform Polymarket/Kalshi arb tooling appearing in multiple independent repos in the same week suggests either a shared information source driving parallel development or a genuine market inefficiency that several operators have independently identified. Either way, the window for easy cross-platform gains is likely compressing.

---

Summary Counts - Total repos reviewed: 19 - Operationally interesting: 4 - Security red flags: 3 - Noise/low signal: 12