Research on how market makers use cross-market relationships to determine quoted prices.
DeFi & Yields ·
Market makers often begin their pricing strategy not with price prediction, but with a more immediate question: what should an asset be worth right now? This distinction matters because fair-value models aim to translate available market information into an indifference price for quoting, rather than forecast where prices will move in the near term. When quoting an instrument that is not the primary price-discovery venue, market makers can learn the relationship between their market and a deeper, faster-moving reference market—for instance, observing that a product consistently trades at a fixed ratio to a leader instrument, then applying that ratio to incoming leader price changes. This basis model recognizes that two related markets move at different speeds: the leader price changes immediately, while the relationship between them typically shifts more slowly.
Constructing a fair-value estimate requires careful choices about which inputs to trust. Using the midpoint of a thin market or a stale last trade can be misleading, since both may contain little genuine information. Instead, a model should use the simplest inputs that have an economic reason to contain signal about the target price. An exponential moving average with a specified half-life offers a practical framework for deciding how quickly to weight older observations relative to new ones, avoiding the pitfall of accidentally giving more statistical weight to markets that broadcast more frequent messages.
Liquidity conditions add another layer of nuance. When the displayed book thins sharply, a sudden jump in the observed basis may not deserve the same confidence as a similar move from a healthy order book, and reducing the effective weight of observations from thin periods can improve estimates. However, liquidity itself does not determine whether a price is correct.