What is a virtual tout?
What is a virtual tout? An automated, computerized event forecasting model. The model is supervised: Data relating to past performance are used to predict future performance and future events.
A tout is a person with special knowledge that others lack. At a racetrack, a tout collects information about racing stables and horses and gives tips to betters. There are touts in all sports, with many selling advice to sports betters.
The thinking behind the book Without a Tout: How to Pick a Winning Team was that sports betters didn’t need to rely on touts. They could analyze individual and team performance data to beat the odds (as defined by money lines and point spreads in Las Vegas). To make money on sports betting, touts and modelers have to be right 52.4 percent of the time.
The book Without a Tout: How to Pick a Winning Team showed how baseball performance data and engineered features could be used to build models for predicting winning teams in baseball. It also showed how to generate probability estimates by playing thousands of simulated, hypothetical baseball games.
Realizing that many sports betters were not going to build their own predictive models, Tom Miller created an information service named The Virtual Tout®. For a number of years, Tom provided betting advice for baseball, football, and basketball, suggesting which games and teams offered the best opportunities for beating the odds. The results of The Virtual Tout® were sold through American Sports Analysts, an established sports information service in Madison, Wisconsin.
The back cover of Without a Tout: How to Pick a Winning Team suggests many uses of event forecasting models: They have relevance to any competitive arena. We can predict which product will be chosen, what consumers are willing to pay, or which firm will win in the marketplace. As long as there are relevant data from the past and ways of keeping score, we can use models to our advantage.
There are similarities between event forecasting in sports, where the goal is to predict which team will win a game, and event forecasting in politics, where the goal is to predict which candidate will win an election. Sports data are organized by teams, players, and performance measures of teams and players. Political data are organized by location (districts, counties, states), voters, and characteristics of voters (demographics, party affiliation, needs, desires, values).
An election is a game that can be won with data and analytics.
Moving from sports to politics, The Virtual Tout® offers automated, computerized election forecasting. We select relevant data sources, including past political events, preference surveys, and geo-demographic information. We use politically relevant data, along with proprietary, engineered features, to predict future political events—election outcomes.
The Virtual Tout® today draws on state-of-the-art machine learning methods to predict margin-of-victory percentages between pairs of candidates in elections. These proprietary methods have proven to be more accurate than political polls. See The Virtual Tout Archive.
The Book that Started It All #
- Miller, Thomas W. 2008. Without a Tout: How to Pick a Winning Team. Madison, WI: Research Publishers LLC. [ISBN-13: 978-0-9727297-8-9] Research Publishers LLC is now located in Manhattan Beach, California.
Back to About The Virtual Tout