gwyneth-chen

Generate optimal bid and ask quotes for crypto pairs using the Avellaneda-Stoikov framework.

13|3|Updated Mar 31, 2026
One-click install
npx skills add https://github.com/cubexch/ai-fund --skill gwyneth-chen
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: gwyneth-chen
Source: https://github.com/cubexch/ai-fund/tree/main/skills/gwyneth-chen
Command: npx skills add https://github.com/cubexch/ai-fund --skill gwyneth-chen

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Streamline crypto market making by enabling automated, risk-aware quoting that captures spread while maintaining healthy inventory and defending against toxic flow.

Core Features & Use Cases

  • Real-time Avellaneda-Stoikov quoting across multiple exchanges
  • Dynamic inventory skew management to control risk
  • Adverse selection detection and safety pauses
  • Maker rebate-aware quoting with P&L decomposition across venue rebates

Quick Start

Provide a pair and current inventory to generate the best bid and ask.

Frequently Asked Questions about gwyneth-chen

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How does Avellaneda-Stoikov market making work for crypto quoting?

Avellaneda-Stoikov market making generates optimal bid and ask quotes for crypto pairs by modeling spread capture against inventory risk. It calculates a reservation price adjusted for current holdings, then skews quotes to manage exposure while maximizing captured spread.

How do I generate real-time bid and ask quotes across multiple crypto exchanges?

To generate real-time quotes across multiple venues, provide a trading pair and current inventory data. The skill estimates fair value per venue, applies maker rebate-aware pricing, and outputs optimal bid and ask levels with inventory skew adjustments.

Can I adjust inventory skew to control risk during crypto market making?

Yes, inventory skew management dynamically adjusts bid and ask quotes based on your current holdings. As inventory builds in one direction, the reservation price shifts and quotes are skewed to attract offsetting flow, controlling overall portfolio risk.

What's the best way to detect adverse selection and toxic flow in market making?

Detecting adverse selection involves monitoring quote toxicity and flow patterns in real-time. The skill includes safety checks that identify toxic flow and can trigger automated safety pauses, preventing losses from informed traders picking off stale quotes.

Does maker rebate-aware quoting support P&L decomposition across venues?

Yes, maker rebate-aware quoting factors venue-specific rebates into the optimal bid and ask calculation. It decomposes P&L across multiple exchanges by isolating spread capture, inventory gains, and rebate contributions for each venue.

What are the limitations of automated crypto market making during high volatility?

During high volatility, automated market making faces increased adverse selection risk and wider optimal spreads. The skill enforces safety checks and can pause quoting when volatility thresholds are breached, but extreme conditions may still result in inventory imbalances.