Sasha Stoikov's Lecture on Market Making Circulates
A lecture on market making and microstructure by Cornell's Sasha Stoikov is circulating widely within the quantitative finance community. The talk covers the Avellaneda-Stoikov model, inventory risk, and adverse selection, with applications to automated bots on prediction markets. The content is being highlighted as essential material for those building high-frequency or market-making trading systems.
- Sasha Stoikov, a senior research associate at Cornell, has a background in the industry, having worked as a Vice President in the high-frequency trading group at Cantor Fitzgerald and as a consultant for Morgan Stanley. - The Avellaneda-Stoikov model, originally detailed in the 2008 paper "High-Frequency Trading in a Limit Order Book," provides a mathematical framework for determining optimal bid and ask quotes by considering both inventory risk and market volatility. - A central concept of the model is the "reservation price," an adjusted mid-price calculated based on the market maker's current inventory (q), their risk aversion (γ), and market volatility (σ), which allows the quoted spread to be dynamically skewed to manage inventory. - The model extends earlier work by Ho and Stoll (1981) and has been implemented in open-source trading software, including the Hummingbot framework for cryptocurrency market making. - One of the key challenges the model addresses is adverse selection, where a market maker risks providing liquidity to a better-informed trader just before a significant price move. - Stoikov's more recent research includes defining a "micro-price" for multiple cointegrated assets, which uses order book imbalances as a predictor for short-term price moves. - In tests, execution algorithms based on the "micro-price" concept for cointegrated ETFs were found to be able to save half of the bid-ask spread compared to other strategies. - Stoikov's research has demonstrated a linear relationship between order flow imbalance at the best bid/ask and short-term price changes, with the slope of this relationship being inversely proportional to market depth.
