389. Unintended Consequences of Artificial Intelligence in Financial Markets
Invited abstract in session MD-1: Álvaro Cartea, stream Keynotes.
Monday, 14:30-16:00Room: Great Hall
Authors (first author is the speaker)
| 1. | Álvaro Cartea
|
| Mathematical Institute, University of Oxford |
Abstract
Abstract: How can artificial intelligence and learning algorithms affect the integrity of market? We discuss three lines of research:
1. Spoofing and quote-based manipulation. Market making learning algorithms will find optimal strategies that manipulate the limit order book. Manipulation occurs to induce mean reversion in inventory to an optimal level and to execute round-trip trades with limit orders at a higher probability than was otherwise likely to occur; spoofing is a special case when the market maker prefers that manipulative limit orders are not filled.
2. Market making. We show that algorithms can tacitly collude to extract rents and we show that that tick size (coarseness of price grid) in the limit order book matters: a large tick size obstructs competition, while a smaller tick size lowers trading costs for liquidity takers, but slows the speed of convergence to an equilibrium.
3. Breaking anonymity in a limit order book. Empirically, we find that liquidity providers use excessively large limit orders to break the anonymity of limit orders by signaling themselves to other liquidity providers. Importantly, we find that liquidity providers respond differently to signaled limit orders. Specifically, they avoid trading with each other and focus on picking-off soft flow from retail limit orders. We use a model of the limit order book to show that signaling occurs in a collusive equilibrium. In equilibrium, signaling occurs so that colluding firms can identity each other in an anonymous book to avoid sniping each other. This allows colluding firms to share profitable flow from retail limit orders.
Keywords
- Financial Modelling
- Artificial Intelligence
Status: accepted
Back to the list of papers