Who's #1?: The Science of Rating and Ranking [Kindle Edition]

The book explains different approaches to rating different items in a set. These approaches can be used to rate teams in a sport, how good a web-page is by the traffic it receives, or any other set of objects with characteristics you can quantify. While the book also addresses rankings (and quickly distinguishes between a rating and a ranking), the focus of the book is rating. There are many examples showing the various ratings methods. Each of the common approaches to ratings uses a 5-object set of football teams with scores of matches. The various approaches are compared and tested for past and predictive accuracy in different way, such as with an NFL football season. There is also a lot of interesting trivia about BCS ratings, sports betting and other fan topics.

If you are a huge fan of BCS football and team rankings, this book will appeal to you. There are clear explanations of how ratings are produced using approaches of Massey, Colley, ELO, offense-defense and many others (I mention those 4 because those are the most recognizable). If you are very interested in sports modeling, or use math-heavy approaches to sports betting, this is a must-read book. Successful betting groups use models that rely on approaches very similar to the ones listed here. This book spells out half of the work involved in developing such a model. It also briefly touches the other half of model-development: what inputs to use in your model.

The book is packed full of matrices and proofs. I had to go back and review my linear algebra and eigenvectors to fully appreciate the book. There are also practical tips on solving these problems efficiently, and ways to simplify the problems so you do not try to brute-force your way through a monster matrix. You do not have to be able to solve or understand what a 3x3 matrix to enjoy this book, because each method is discussed in a very readable manner. However, the math behind what is going on is half the book. While it is still interesting without following all the math, I would only recommend the book to someone that is familiar with linear algebra, or will put in the time to become competent at it in order to understand the techniques presented.

While overall the content was excellent and had the right depth, I felt like the sports and spread-betting content could have used more detail. For example, when considering inputs for a football rating, the authors conclude that "Experiments tend to suggest that scoring data provides better overall rankings for NCAA football than does yardage data." One study was mentioned considering 2005 ACC Football, but were there other studies supporting that generalization? The authors also discuss the difficulty in applying ratings to predict spreads, including blow-outs and irregular scoring. While these provide extra challenges to applying ratings to betting on the spread, there are simply ways to attack each of these problems. They ultimately conclude that "simply averaging past scoring differences will probably give you an estimate that is competitive with more complicated techniques". However, even using slight adjustments to inputs can increase the accuracy of fore-site projections.

I gave the book 5 stars because it was very clean and well written, it nicely summarized and discussed many mathematical approaches to ratings, and no other book covered this topic.

"Click to Look Inside" is active for this book, and that's a worthwhile effort for this book. Pretty much every page is interesting.

When I was 8, I loved to look through books like this one. I was fantastic at ignoring the math I didn't understand, while reading very interesting text, talking about math tools, and then looking at the pictures generated by the math tools. Later in life, when I knew some math, I was able to use those tools. If you're not a math person, but have the ability to ignore mathematics while going through a book, then this book is still good for you.

Here, the tools are being used on sports teams, products, movies, and websites. As it happens, part of my job is to study rankings (mostly of websites), and here this book comes along to provide a comprehensive overview.

What's it worth? As an example, the Netflix Prize sought to substantially improve the accuracy of predictions about how much someone is going to enjoy a movie based on their movie preferences. Chris Volinsky won a million dollars for his ranking method. He's one of the endorsers for this book.

There is never a "what is this good for?" moment in this book -- the methods pick out winners. Chapter headings: Introduction to Ranking, Massey's Method, Colley's Method, Keener's Method, Elo's System, The Markov Method, The Offense-Defense Rating Method, Ranking by Reordering Methods, Point Spreads, User Preference Ratings, Handling Ties, Incorporating Weights, What If Scenarios, Rank Aggregation, Methods of Comparison, and Data.

This book is entertaining enough to be considered a high-level popular math book. Anyone tasked with being #1 in a field will find this book useful and lucrative. Highly recommended.