Up: 1:2

Algorithmic trading in the Iowa electronic markets

James E. Schmitz

Algorithmic Finance (2011), 1:2, 157-181
DOI: 10.3233/AF-2011-012

Published: Abstract, PDF.
Archived: SSRN.


The Iowa Electronic Markets are small, real-money financial markets designed to aggregate information about future events. The market microstructure of these markets is studied and a market making model is developed to provide liquidity for one set of securities offered by this exchange. A computer program was created to employ the market making model and profit from the market’s inefficiencies. Using invested capital, the system traded 34% of the total market volume and achieved a Sharpe ratio of 9.9. This paper reveals the details of how this algorithmic trader worked to show how it functioned and the value it added to the Iowa Electronic Markets.

Enhanced Content

This chart shows historical data derived from market activity and the algorithmic trader's own trading history. The chart is annotated with significant political events to show how the election process affected market activity and the algorithmic trader's profitability.

Explore the interactive chart below to learn about the algorithmic trader's market making model and to understand how the model's parameters affected its behavior. Read the research paper to learn how this model was derived and to see an analysis of the trading results.

Min Max Value Description
+100 A=
Market Maker's current position
6 t=
Expected holding period for position
3% =
Estimated one-day volatility of the market asset
10 =
Risk aversion parameter
150 Y=
Estimated limit order size posted by other traders
5.0e-5 =
Market impact parameter

The Market Marker used two equations to determine how many shares it was willing to buy or sell at any price. These equations were functions of parameters that included the Market Maker's current position, its risk aversion and the expected volatility of the securities.

The goal of this market making model is to repeatedly buy securities at prices just below fair value and to sell securities at prices just above fair value, making a profit as a result. The difference between fair value and the price the Market Maker can buy or sell is the potential spread profit, and is represented by the parameter s in the below equations. As s gets larger, the profit opportunities available to the Market Maker get larger and the Market Maker becomes more willing to trade a larger number of shares.

The above chart shows the number of shares the Market Maker was willing to buy or sell at any price on the x-axis given an expected fair value of 0.5. The size of the buy and sell orders were calculated using these equations:

Read the information in the other tabs to learn more about the purpose of the other parameters and how they affected the Market Maker's behavior.

The Market Maker would continuously buy and sell securities resulting in a long or short position in the market asset. The parameter A represents the Market Maker's current position and would change every time it made a trade. As its position grows, it becomes reluctant to further increase its position and more willing to reduce its position. As you change A, observe that when it is positive (negative) the Market Maker is more (less) willing to sell and less (more) willing to buy.
After buying or selling a security, the Market Maker would hold on the position for an extended period of time. The parameter t represents the estimated amount of time the Market Maker expects to have to wait before it can trade out of the position. The longer it expects to have to wait, the less willing it is to enter into a position in the first place. As you change t, observe that when it gets larger (smaller) the Market Maker is less (more) willing to either buy or sell.
The security prices were volatile and unpredictable, so it was possible for the Market Maker to lose money from a price change after entering into a position. The parameter provides the model with an estimate of the expected volatility of the securities. As you change , observe that when it gets larger (smaller) the Market Maker is less (more) willing to either buy or sell.
Because of the volatility and unpredictability of the security prices, the Market Maker needed to balance its goal of making profitable trades against its aversion to the risk from the resulting positions. The parameter quantifies the Market Maker's risk aversion. As you change , observe that when it gets larger (smaller) the Market Maker is less (more) willing to either buy or sell.
When the Market Maker would post a buy or sell order in this market, it would set the price equal to the best bid or ask price previously set by another human trader. By trading in this manner, the Market Maker was "joining" other orders at the bid or the ask. The size of the other orders were not visible to any market participant so the Market Maker needed to estimate their size. The parameter Y is that estimate. As you change Y, observe that when it gets larger (smaller) the Market Maker is less (more) willing to either buy or sell.
The Market Maker was just a computer program that knew nothing about the election or the underlying meaning of the securities it was trading. It would buy and sell securities with human counterparties who had opinions and information about the election outcome that the Market Maker did not. Security prices would change as a result of the trading, incorporating the humans' opinions and information into the price. On average, these price changes would result in a profit for the human and a loss for the computer program. The parameter quantifies the expected information content of the humans' trades and the expected adverse price move. I gradually increased this parameter as the time till the election decreased because I expected the information content of the humans' trades to go up. As you change , observe that when it gets larger (smaller) the Market Maker is less (more) willing to either buy or sell.

Managing Editor

Philip Maymin

University of Bridgeport

Deputy Managing Editor

Jayaram Muthuswamy

Kent State University

Advisory Board

Kenneth J. Arrow

Stanford University

Herman Chernoff

Harvard University

David S. Johnson

AT&T Labs Research

Leonid Levin

Boston University

Myron Scholes

Stanford University

Michael Sipser

Massachusetts Institute of Technology

Richard Thaler

University of Chicago

Stephen Wolfram

Wolfram Research

Editorial Board

Associate Editors

Peter Bossaerts

California Institute of Technology

Emanuel Derman

Columbia University

Ming-Yang Kao

Northwestern University

Pete Kyle

University of Maryland

David Leinweber

Lawrence Berkeley National Laboratory

Richard J. Lipton

Georgia Tech

Avi Silberschatz

Yale University

Robert Webb

University of Virginia

Affiliate Editors

Giovanni Barone-Adesi

University of Lugano

Bruce Lehmann

University of California, San Diego

Unique Features of the Journal

Open access
Online articles are freely available to all.
No submission fees
There is no cost to submit articles for review. There will also be no publication or author fee for at least the first two volumes.
Authors retain copyright
Authors may repost their versions of the papers on preprint archives, or anywhere else, at any time.
Enhanced content
Enhanced, interactive, computable content will accompany papers whenever possible. Possibilities include code, datasets, videos, and live calculations.
Algorithmic Finance is the first journal in the Financial Economics Network of SSRN to allow comments.
The journal is published by IOS Press. In addition, the journal maintains an archive on SSRN.com.
While the journal does reserve the right to change these features at any time without notice, the intent will always be to provide the world's most freely and quickly available research on algorithmic finance.
Online ISSN: 2157-6203
Print ISSN: 2158-5571