Quantitative Trader for Polymarket System Enhancement

Senior Freelance Quant/Low-Latency Trading Engineer — Performance-Based Contract

I’m hiring an experienced quantitative trader or low-latency trading engineer to turn an existing Polymarket system into a consistently profitable, diversified portfolio.

What is already built:

- Go execution engine deployed on AWS in `eu-west-2` – I got colocation approval
- L2 WebSocket order books and authenticated fill stream
- Warm, reused HTTP/2 order lanes through Cloudflare
- Pre-signed maker-order envelopes
- Highest-passive maker pricing
- Configurable maker leases and asynchronous cancellation
- Python temporal cross-validation and portfolio-bootstrap framework
- ONNX inference with continuously cached predictions
- Strategies for BTC, XRP, ETH, SOL, and DOGE five-minute markets

Measured engineering performance:

- L2 book source-age p50 approximately 4–7 ms during healthy operation
- Cached-book admission processing as low as 0.007–0.074 ms p50
- Admission p95 approximately 0.02–0.74 ms depending on the active instance and asset
- Warm local signal-to-first-write benchmark: approximately 8.6 μs
- Raw London network RTT approximately 2.9–3.1 ms
- Authenticated Cloudflare/venue HTTP response generally approximately 36–50 ms
- Maker pricing, admission, and cancellation operate independently from model inference
- Current maker lease is configurable; recent experiments use 200 ms

These measurements describe our infrastructure, not guaranteed venue arrival or fill latency.

Current out-of-sample research results include:

- BTC: 13,151 signals, 97.28% directional accuracy, approximately $8.23/day under forced-full arithmetic
- BTC neutral 50% market-fill bootstrap p5: approximately $3.28/day with −$34.53 global drawdown
- XRP: 12,479 signals, 97.47% directional accuracy, approximately $14.38/day under forced-full arithmetic
- XRP neutral 50% market-fill bootstrap p5: approximately $5.66/day with −$64.89 global drawdown
- XRP forced-full historical drawdown: approximately −$35.22

These are causal historical simulations—not achieved live returns. The remaining challenge is converting strong signals and fast infrastructure into reliable live profitability despite queue position, adverse selection, cancellation races, asymmetric fill rates, and correlated portfolio risk.

Responsibilities:

- Audit the complete signal-to-order and fill-reconciliation path
- Diagnose missed fills and adverse-selection bias
- Improve maker pricing, quote lifetime, cancellation, and re-entry
- Measure winner-versus-loser fill asymmetry
- Construct a diversified portfolio across supported assets
- Optimize allocation and risk using walk-forward validation and portfolio bootstrap
- Align replay assumptions with authenticated live fills
- Run a bounded-capital forward test
- Deliver reproducible code, tests, documentation, and telemetry

Success criteria:

- Ten consecutive calendar days of live forward testing
- Positive net realized portfolio ROI after fees
- No unresolved positions counted as profit
- Compliance with agreed exposure, drawdown, and per-market limits
- No manual cherry-picking of trades or markets
- All orders, fills, cancellations, and settlements independently verifiable
- Stable operation across the agreed diversified portfolio

Compensation is success-based and payable after the agreed criteria are achieved. Payment, test capital, maximum drawdown, portfolio composition, test start date, and verification procedure will be agreed in writing before work begins—preferably through an escrowed milestone contract.

Ideal background:

- Professional quantitative or systematic trading experience
- Strong Go and Python engineering
- Limit-order-book and market-making expertise
- Experience with Polymarket or comparable electronic exchanges
- AWS and low-latency networking knowledge
- Expertise in adverse selection, fill modeling, temporal cross-validation, and portfolio risk

Please apply with:

- Relevant trading or execution systems you have built
- Evidence of live execution work—not only backtests
- Your diagnosis plan for fill asymmetry and adverse selection
- Your proposed diversified portfolio methodology
- Availability for a monitored ten-day forward test
- Your requested success fee

Credentials and private keys will never be shared directly. All work must respect venue rules, applicable laws, and the agreed risk controls.