I was flipping channels the other night when I happened to see the show Arli$$. The main character, Arliss Michaels, is a sports agent. In this particular episode, Arliss was so intent on winning his National League-only fantasy baseball league that he maneuvered a trade to send one of his real-life clients to the American League.
I could sell the grass-court naming rights to Miracle-Gro or maybe High Times magazine. Just think of it, the ATP High Times Grass Court Top 100.
I can relate to that. I’ve been waking up in the morning thinking of new statistical methods I can use to propel myself to the top of the ATP Fantasy Tennis Season next year. One morning I woke up wondering if I could predict grand slam winners by looking at players’ records in Masters Series events. Then I went further. What about looking at their Masters Series records on different surfaces? And how about a surface adjusted ranking (SAR)? Take the top 100 players, upload all of their ATP points on the five different surfaces, then create a separate top 100 for clay court, outdoor hard court, grass court and indoor hard court.
I could sell the naming rights to four top 100 rankings instead of just one. I could sell the grass-court naming rights to Miracle-Gro or maybe High Times magazine. Just think of it, the ATP High Times Grass Court Top 100.
Yes, I’m turning into a fantasy sport fanatic. I’m beginning to understand why fantasy gamers hire biomathematicians at NASA.
Another morning I woke up thinking that I should be able to develop an algorithm to pick my tennis fantasy team. To test it out, I could run it on every tournament this past year and see how much prize money it wins. (ATP fantasy tennis gamers are ranked by the amount of prize money their players win.) Isn’t there a computer program of some sort that can look at the variables in my algorithm, assign weight to their importance, then adjust the weights until my algorithm generates the maximum prize money?
Yes there is.
In the October 16th issue of the New Yorker, Malcolm Gladwell wrote an article about a system that analyzes the scripts of big budget movies and predicts their gross income. The process involves using a computer program known as a neural network. The article also describes a neural network that predicts winners at the dog track using the same variables that greyhound experts use.
Now that I have CFSD (creeping fantasy sports disease), I want to do the same thing for my fantasy tennis team. My algorithm would mirror the process I already use to pick my fantasy team.
I start by picking winners in the draw of each tournament and that already presents a problem. There are thirty-one matches in a thirty-two player tournament and I have to pull up three web pages of statistics for each match. If there are three tournaments that week, that means I have to look at two hundred and seventy-three web pages. That’s totally ridiculous, it takes forever. What to do? I don’t need a biomathematician but I have contracted someone to write a screen scraper program. All I have to do is enter the names of the players and the program will go to the website and download all my data into one file.
Not only is my head now filled with arcane statistical possibilities, but everything I read goes through the CFSD filter. The ATP recently announced that thirteen tournaments will use the round robin format next season. That’s a problem for my algorithm. Each tournament has at least four qualifiers in the first round but when I pick the draw on Sunday morning, I don’t know who those qualifiers are because qualifying doesn’t finish till Sunday afternoon. Most of the time qualifiers lose in the first round and it’s not a big deal, but in a round-robin tournament every one plays at least twice so the qualifier will have a greater effect on the outcome. Also, if the tournament is a twenty-four player tournament, as some of the round robins are, qualifiers will be higher ranked players and that’ll make my predictions even less accurate.
I should be thinking about whether round-robin tournaments will increase the popularity of tennis. Tennis is losing traction in the U.S. ESPN dropped French Open coverage as of next year and though it’s been picked up by The Tennis Channel, it’ll reach far fewer households and viewers will have to pay extra to subscribe to the channel. Instead, I’m worrying about how round-robins will affect my algorithm and thinking of flying to Wisconsin at Christmas and freezing my patootie off just so I can hang out with my friend Billy and get help with this whole computer modeling thing.
I’m supposed to be a sports journalist and that means looking for a good story and showing the human side of sports. The subjects people really care about are the personalities, the drama and, let’s face it, the gossip.
Last week I wrote about a dustup between Mark Philippoussis and Nathan Healey at the Calabasas challenger. I got two email responses from the column. One woman wanted photos for her Philippoussis website and another was interested in pursuing the subject of sportsmanship in tennis. Nobody left me a message suggesting they were wowed by my statistical brilliance.
Two e-mails to zero. Maybe I should pay more attention to that statistic.