Some of you may remember the Predictor I had last year, which looked at all of the counting stats (Pass/Rush Yards, Pass/Rush Attempts, Turnovers, First Downs, etc), and spit out a probability of winning the game. It worked decently, for looking at data that can be horribly skewed.
This offseason, I set out to find out which stats led to the most wins, and that's where the Winning Stats series came from. We looked at 16 different stats, and found out how often they lead to wins when a team goes above the league average, both offensively and defensively. It's only natural now I take the data we learned over the summer, pump in the data through the first 5 weeks of the season, weight them accordingly, and have our own set of Power Rankings, which are not subjective in nature at all.
Here's how it works:
- Like I said, there are 16 stats, going from the most important (Drive Success Rate) to the least important (Net Punting Yards / Game).
- WARNING: STAT GEEK TALK - In order to get all the stats on the same scale (since it's hard to compare a percentage to, say, Yards / Drive), I found where each stat lied on a Normal Distribution curve, which only goes from 0 to 1. I used the Mean and Standard Deviation from the past 8 years for this, so we're looking at how good these games fall in relation to over 4200 games, rather than the roughly 150 that have happened so far this year.
- Non Stat Geek Explanation of Previous Bullet - In order to compare each of the stats equally, I made it so each one was in a certain percentile from games over the past 8+ years. For example, the Colts ANY/A average stat this season is in the 91st percentile over the past 8 years, so the value is 0.91. I did this for every stat, for every team.
- Once all the stats are on a level playing field, I need to weight them, as we now know that certain stats lead to many more wins than others (like Passing Stats lead to lots more wins than Rushing Stats). Since I need to keep part of this secret in my computer, all I'll say is that the weighting factors are based off of the Winning Percentages found throughout the summer. Plus I want to keep refining them, so I don't want to give them out.
- I've calculated both an Offensive and Defensive Power Rankings, as well as the Total Rankings. As in all other rankings, a lower number is better.
- I also calculated a Strength of Schedule based on these Power Rankings, to show how easy/hard a team's schedule has been so far this year. I used 2/3 Opponent Rankings, and 1/3 Opponent's Opponent Rankings. Also, the Offense column is the Strength of Schedule the Offense had to face, not how strong the opposing Offenses were.
Ok, enough explaning. Let's get to the numbers, after the jump...
Click any of the headers to sort the columns:
|Team||Power Rankings||Strength of Schedule|
My thoughts on the rankings:
- The Colts are ranked 6th overall, which includes being ranked #1 offensively. If the Defense continues to play at a high level, that ranking will continue to grow.
- I mentioned this in the Open Thread of Stampede Blue Radio on Wednesday: If you take out the Dolphins game (which looks like an outlier at this point), the Colts would be ranked 5th defensively. That one game dropped the Colts that much.
- The Giants are ranked 3rd best offensively, and best defensively, giving them the top spot. However, if you look at the Strength of Schedule numbers, you'll see how easy of a schedule they've played. Sunday's game against #2 New Orleans should be very interesting.
- The Chargers are ranked dead last in defense, although they've played the toughest defense schedule so far. They've also played a pretty easy offensive schedule, which will definitely change on Monday night against the Broncos.
- The 49ers are the most lopsided team, having a very good defense, and a bad offense. The Patriots are also pretty lopsided, with the offense looking better than the "experts' say, and the defense looking worse than the "experts" say.
- The Rams have played some tough defenses this season, but below average offenses. That hasn't stopped them from looking really bad on defense.