Understanding Market Functionality and Trading Success

Building on my interest in prediction markets and my previous paper Algorithmic Trading in the Iowa Electronic Markets, I co-wrote a research paper with David Rothschild titled Understanding Market Functionality and Trading Success. This paper is an analysis of trading activity in two markets provided by the PredictIt website.

The general idea of the paper is to analyze the profits and losses of each trader. We decomposed the profits and losses to understand how and why each trader made or lost money. Did they make correct decisions about which securities to hold? Did they execute their trades optimally? Did they earn or pay the spread?

Interestingly, we found that trader execution costs played a bigger role in overall profitability than making correct decisions about which securities to hold. This suggests that prediction markets should be simplified so that executing trades correctly or optimally is easier for less sophisticated traders.

The paper is published through PLoS ONE. The relevant Python code is available on github. The raw data is not downloadable but is available through PredictIt by request.

Before starting this paper I was an active user of the PredictIt website, but while working on it, I didn't execute any trades or hold any positions. The entire project took about three and a half years.


We examine individual-level trading data from several markets in the PredictIt exchange to determine what strategies correlate with financial success. PredictIt provides many markets with futures contracts linked to political issues, ranging from ongoing policy outcomes to political elections. High fees along with restrictions blocking automatic trading and constraining a one-to-one match between people and accounts, combine to severely limit the upside to investment returns over the fixed costs: this ensures that traders are all retail investors. We have the individual-level data from two markets: Democratic and Republican Iowa Caucuses in 2016. This data includes all orders and trades from every trader across the markets. We are able to fully reconstruct market activity and study trader behavior both within and between markets. We show that understanding how markets and trades works is more important to financial success than proxies for (1) confidence or funding (2) information or objectivity in trading. The work should be a call-to-action in favor of simplifying markets and trading for any exchange with retail investors, and for more research into effects of differential trading efficiency in all financial markets.