pmintel.report
← all digests
Jul 20, 2026daily

NegRisk Arb Bot Live, Star-Farming Spam Hits Repos

A clean multi-outcome NegRisk arbitrage scanner dropped today with confirmed risk-free spread detection across Polymarket, Kalshi, and PredictIt — worth cloning immediately. Separately, the arbitrage repo itself is being targeted by a coordinated star-farming campaign, so verify your fork source before you build on it.


NegRisk Arbitrage Scanner Headlines a Strong Batch

Seven bots crossed the desk today. Five cleared the bar for cloning. The standout is **arbitrage** — a multi-outcome NegRisk arbitrage detector that scans the full Polymarket CLOB order book looking for outcome sets whose combined implied probability deviates meaningfully from 1.0. It handles three distinct edge cases: same-venue binary complements, multi-outcome NegRisk field arb, and cross-venue Kalshi mirror matching. If you're not already running something like this, you're leaving money on the table every time a correlated market cluster gets mispriced. The repo computes net profit percentages after fees, which is the number that matters — gross spread is noise.

New Bots: Full Rundown

**polymarket-weather-edge** is a quant weather play: it fits a Gaussian PDF over METAR observations and Open-Meteo ensemble forecasts, then compares modeled probability to the implied probability sitting in the order book. When the market is overpriced relative to the model, it bets NO. Temperature bucket markets are chronically mispriced by retail — this is a real edge.

**DRADIS** is the most ambitious repo in today's batch. It's a multi-strategy engine covering momentum, market-making, arbitrage, time decay, basis trading, and an ML layer (Random Forest + LSTM). It ships with an Ollama/Llama AI advisor for parameter optimization and a Next.js control tower dashboard. The breadth is either an asset or a liability depending on your discipline — a system that does everything often masters nothing. Worth cloning for the infrastructure patterns even if you strip out half the strategies.

**auspex** goes on the monitor list. It tracks distance from current on-chain state to market resolution thresholds and generates resolution-confidence scores. No fees on trades via an integrated swap. Useful for crypto event markets specifically, but the resolution-confidence scoring methodology isn't documented well enough to trust blindly yet.

**pred-arbitrage** also goes to monitor. Cross-venue arb detection across Polymarket, Kalshi, and PredictIt with historical spread tracking. The PredictIt integration is notable given that platform's liquidity constraints, but the spread data over time could be genuinely useful for sizing decisions.

Security: One Malware Template, Two Spam Campaigns

Three alerts worth your attention today.

The **460KB Clone Fingerprint Malware Template** is the critical one. A template repo is circulating that appears to be a prediction market bot scaffold but contains fingerprinting payloads. If you cloned anything in the last 48 hours from an unfamiliar account, audit it. Look for obfuscated fetches in the dependency tree and unexpected environment variable reads at startup. Do not run unknown bot code with wallet keys in scope — ever, but especially right now.

The **Star Farming campaign on the arbitrage repo** is lower severity but operationally important: inflated star counts are being used to push the repo up search rankings. This does not mean the arbitrage bot itself is malicious — our analysis flagged it as worth cloning — but it does mean you should verify you're pulling from the canonical source and not a poisoned fork riding the visibility.

A **garbled username spam campaign** is also active, likely probing for engagement or seeding future phishing infrastructure. Flag and ignore.

Alpha Signals

Three research papers worth reading this week. The most immediately applicable proposes an agentic pipeline that clusters prediction markets by topic using NLP and detects implicit equivalences and hidden contradictions across markets — exactly the kind of systematic fragmentation analysis that surfaces arb opportunities before they're obvious. The M&A deal outcome forecasting paper (LLM over long merger documents) is relevant if you're trading any corporate event markets. The path signature execution framework is further from direct application but worth tracking if you're building execution logic that needs to account for signal history rather than just current state.

What to Watch

Spain blocking Polymarket and Kalshi over licensing is the macro story this week — watch for regulatory copycat risk in other EU jurisdictions. The HN thread on death threats over Iran missile markets is a reminder that resolution disputes on geopolitical markets carry real-world blowback. And the "Nothing Ever Happens" NO-bias bot hitting 471 points on HN means retail is about to flood into naive contrarian strategies — factor that into your model when you see NO prices drift on low-activity markets.