Hi, guys. Nevermind the long screen name. I'm Mike, and I'm a part-time contributor to Lighthouse Hockey. Usually during hockey seasons when teams are actually, you know, playing hockey, I enter all the scores into a spread sheet I designed, with the aim of comparing teams with similar records. We typically also compare the numbers I crunch with the "power rankings" that other sites such as ESPN put out.
Since there isn't any hockey and my nerdy stats finger is getting itchy, I thought I'd modify my spread sheet for football.If you don't care about such stuff, move along. Nothing to see here! If you do like discussing such stuff read on...
Since this is the first time I'm doing this here, I will explain how I crunch these numbers at the end of this post. If there's enough feedback, I'll do this in subsequent weeks.
Much like the opinion polls you see around (where people rate teams based on how good they feel that team is), this approach (exactly the opposite) will never be used to settle ties in the standings, but it's interesting to see how teams compare based on who they beat (and who those teams beat as well). With only six weeks of data so far, some of these are sure to raise eyebrows. But I hope this will add a little bit of context to the still-young season. I will explain how the numbers are calculated after the table. So without further ado...
NFL Power Rankings (through Week 6)
But that's ridiculous, right? No way the Giants are only 16th best in the league right now!
OK, take a deep breath. All this means is that, based on who the Giants have beaten so far, they haven't been too impressive yet, statistically speaking. Most of the teams they've beaten have losing records, for example. And the numbers also take into account who the Giants' vanquished foes have beaten, as well. So, for example, the Niners have all of their wins have been against teams with .500 records or worse. Their two losses have been the only games against teams with winning records. And all of the other Giants victories have been against teams with one or two victories each.
So it makes some sense, from a numbers point of view, that the Giants aren't very high. You'll also notice that the only unbeaten team, the Falcons, are only two spots above Big Blue, in 14th. They haven't beaten a team who currently has a winning record yet. Now, the Falcons (and everyone else) cannot control who they play against. If the Falcons play a few winning teams and are still unbeaten, they will rise straight to the top. But for now, they just haven't been statistically impressive.
So why are the Niners rated so high, you ask? Well, for all those .500 teams they defeated, each of those teams have wins against teams with good records. That helps immensely. You'll understand why when I explain the numbers below. At this point, I should also mention that every loss is treated the same, and margin of victory, home vs. away, etc., all don't factor into the numbers. If you really want to understand how it works, here's how:
- For any given Team A, I add the winning percentages of every team they've beaten, and then divide by the total number of games played. Losses add nothing. So if Team A is 2-1, and they beat two 1-2 teams, then the number would be (0.333 + 0.333 + 0) divided by 3, or 0.222. For now, I will call this number "X".
- After doing that for all teams, I run through the whole exercise again, but instead of using winning percentages, I use each teams "X" number, and then multiply by 10,000 just so the number ends up being a little easier to read. And that ends up being the number in the far right column, the power ranking points.
Since there isn't a lot of data points in even a full season (compared to hockey's 82-game season), that second iteration is important. Say you have two 4-2 teams, and each of those teams have their four wins all against .500 teams, then which of the 4-2 teams is better? Well, if Team A defeated a bunch of 3-3 teams who have in turn beaten a few good teams, while Team B has beaten a bunch of 3-3 teams who have only beaten bottom feeders, then Team A should be considered better. And that's how these calculations work.
OK that was a lot longer than I expected. Feel free to ask questions,and I'll answer whenever I have time. To answer an obvious question for now: Typically, by Week 8 or 9, teams will start to settle more along the lines where you'd expect them to be. But there will always be a few surprises.