Okay, so today I decided to mess around with Massey ratings for junior tennis. I’ve heard about this system before, but never really dug into it. Basically, it’s a way to rank players based on their match results, but focusing on the margin of victory, not just wins and losses.
First, I needed some data. I found some recent junior tournament results – luckily, scores were readily available. Just plain old sets won and lost, nothing fancy.
Then came the tricky part: setting up the equations. The Massey method involves creating a matrix, where each row represents a match, and each column represents a player. You put a ‘1’ for the winning player, a ‘-1’ for the losing player, and ‘0’ for everyone else. The right-hand side of the equation is the score difference (like, sets won minus sets lost).
It starts looking something like this (imagine this is a simplified part of a much bigger table):
- Match 1: Player A beats Player B, 6-4, 6-3. Score difference = +2 for Player A.
- Match 2: Player C beats Player A, 7-5, 6-2. Score diffence = +2 for player C.
So I’m there, filling this thing in, row by row. It got pretty tedious, I have to be honest. Lots of flipping back and forth between the results sheet and my spreadsheet. I may have made a mistake in transcribing scores, I will need to double check my table later.
Solving the System
Now, this is where my rusty math skills got tested. You end up with a system of linear equations. I used an online matrix calculator. I copy-pasted my huge table of 1s, -1s, and 0s into the calculator, hit “solve,” and crossed my fingers.
It spit out a bunch of numbers – the Massey ratings! I found that Player C is rated higher than player A, while player A is still rated higher than player B.
The numbers themselves don’t mean much on their own, they’re relative. What matters is the order and the differences between the ratings. A higher rating means a “better” player, according to the system.
What I Learned
Honestly, it was a bit of a slog, especially the data entry part. But it was cool to see how the system works. It’s definitely more informative than just looking at win-loss records. If you can do it on the game level instead of sets, I figure that the numbers produced will be more reliable, but I will test this some other time. I am also looking forward to comparing it to Elo and Glicko rating systems in the future.
Would I recommend doing this by hand for every tournament? Probably not, unless you’re really into spreadsheets. There are probably tools out there that automate this. But for a one-off experiment, it was a fun (and slightly painful) way to spend an afternoon.
