A timely book here just ahead of opening day, http://tradingbases.squarespace.com/. Peta relates a lifelong love of baseball and statistics, his experiences as an equity desk trader for Lehman Bros. (15 years) and his subsequent battle back from a horrifying injury sustained by being run over in the streets of NY by an ambulance –as if his Lehman experience wasn’t enough to endure. He suffered a “Theisman grotesque” leg break that left him depressed and basically rehabbing alone in his NY apartment with wife and family living on the west coast.
His passion for trading snuffed by not being able to work, hopped up on pain meds, and trapped in the apartment leads to him to watching more sports than ever before. A baseball lover at heart and a statistical junkie, Peta finds a reason to wake up in the mornings. He decides to try his hand at making a statistical model that would identify edges for baseball team wins and losses that would provide him with a betting edge over the Vegas Line.
Peta eventually creates a hedge fund that bets baseball games that returns 41% in 2011 with similar daily volatility as the S&P 500. The book outlines Joe’s views on gambling. Baseball is his preferred niche since the juice/spread is the smallest in comparison to other sports, the ability to use statistics to get an edge is available, and the natural alignment between the better and the team– rooting for your team to win versus the convolution of winning and beating a point spread.
Joe explains his model with care, logic, and facts–backing up his assertions with anecdotes, experiences and back testing in terms of the body of baseball evidence from the historical stat stockpile. He builds on the pioneer work of Sabermetrics, Bill James, and Nate Silver. In general his system uses time tested relationships of team win/loss records, runs allowed/runs scored, starting pitching assumptions, WAR/PECOTA analysis, and more. He relates his journey on a monthly basis showing his results, the breakdowns of what went right and what went wrong, his acceptance of a “lumpy” higher return than a smooth more accepted rate of return by clinging to the belief that reversion to the mean will occur–eventually. He uses a concept called “cluster luck” to identify “lucky or unlucky” pockets in team’s prior records that should be ignored and removed from overall estimates. This is a key to his being able to form an opinion against the betting line of under or over valuation. His model then picks matchups that should be bet on and he uses a very systematic approach to determining the amount of the fund to bet on any one game–essentially using fund manager skill sets.
One notable opinion of his describes his fondness for “skill sets displayed” versus the recording of errors (mistakes committed and sometimes unfairly attributed). He uses SIERA (skill-interactive ERA) for pitcher evaluation and special modeling for playoffs and interleague games. He also walks the reader through his decision making process for when to tweak the model or when to stand pat—. Over tweaking will result in removing the natural capture of mean reversion. Joe has a friendly writing style and comes across as genuine, interesting and likeable.
I think any spec would like this quick reading book–you will learn something here about baseball stats and baseball betting theory; you may well enjoy the woven storyline of his trading career experiences as these snippets and stories move betwixt his model outlining. It is written for an above average mind, but its not too heavy to put someone off who doesn’t deal with wall street or modeling on a daily basis. I read ever page, every micro-print footnote, and every end note.