Alabama just beat Mississippi 103-88 in Oxford—which basically proves power ratings work. Yes, just this one game did that. (kidding)
Without them (and indeed, any time pre-Jeff Sagarin) the teams would have been considered roughly equal. Both teams were 19-8 coming into the game, so it's fair to say both would be ranked (or not ranked) about the same in the AP top 25. But sportswriters have gotten wise, and consult power ratings to some extent, which is why Mississippi doesn't have a single vote while Alabama is at #14 (the Tide are still under-ranked, so not every sportswriter has caught up with the times)
Granted, it's very likely that Ole Miss would have slid a lot after losing 5 of their last 6 games no matter what decade we're talking about. But they only appeared in the poll once (#22) when they were 15-1, and were left out even at 18-4. Their #63 ranking by KenPom probably had a lot to do with that. Even in the Sagarin era the Rebels would have been close to the top ten at 15-1.
But Bracketology is different, and a good record can be more dazzling. And a bad record a big drawback. On December 28, when Ole Miss was 12-0 and Alabama was 7-5, most Bracketologists had Mississippi around a 6-seed, while Alabama wasn't on the board. Some Bracketologists don't "forecast" the rest of the season as opposed to taking a "Selection Sunday is today" approach, so those results make some sense. But since we use power ratings to forecast the bracket's end result, let's see where Alabama and Mississippi were back in late December and how close our prediction came, in the December 24 Dance Chance:
| Record | Dance Chance | Seed Rank Team Conf Rec Proj Chalk %Odds %Chg Proj Prev 1. Purdue B10 11-1 28-3 30-1 99.7 +0.7 1 1 2. Arizona P12 9-2 28-3 29-2 99.5 +0.7 1 1 3. Houston B12 12-0 27-4 29-2 98.1 +1.4 1 1 4. BYU B12 11-1 26-5 29-2 94.7 +2.8 1 2 5. Iowa St. B12 10-2 24-7 27-4 91.7 +1.3 2 2 6. Florida Atlantic Amer 10-2 27-4 29-2 90.4 +10.4 2 3 7. Alabama SEC 7-5 23-8 26-5 90.3 +2.6 2 2 8. Connecticut BE 11-2 24-7 27-4 88.7 -3.3 2 1
This was our forecast for games through December 23. Not too bad; we had three of the currently consensus 1-seeds as the top three. UConn was a 2-seed, having been a 1-seed the previous week before losing to Seton Hall. Florida Atlantic had just upset Arizona so they were ranked very high in Strength and projected to walk through the American, which didn't quite happen. BYU's Strength was very high, but they've been up and down (up lately after the Kansas upset).
And then there's Alabama, at 7-5, but projected for greater things:
Alabama is only 7-5 but they're favored by the Strength power rating in every game going forward, and projected to finish around 23-8. The Tide recovered [from a loss to Arizona] with a 111-67 blowout of Eastern Kentucky.
Alabama is 20-8 now, so they'd still have to win their last three games to make our projection come true. But they're now a 3-seed in the Bracket Matrix.
Meanwhile Mississippi:
| Record | Dance Chance | Seed Rank Team Conf Rec Proj Chalk %Odds %Chg Proj Prev 73. Mississippi SEC 12-0 19-12 17-14 25.2 +4.1
Mississippi is the final of the three undefeated teams left, and even at 12-0 they're not very close to being a projected tournament team. The problem is their #93 Strength rating, which means they're only favored to win three games going forward, and with variance probably get 5 wins to finish 19-12. One of the games they should win is Bryant, but after that they might only win two SEC games. We'll see if that happens, as the Rebels have already beaten some decent teams—9-3 NC State, 10-2 Memphis, and 8-3 UCF. This is why they're #3 in Success, and #15 Overall in our power ratings. But Strength is best for looking to the future, and the future isn't bright. They're roughly in the same spot LSU was last year when the Tigers started 12-1 before going 2-16 in the SEC. The Tigers did upset Arkansas as part of their early win streak so I think the Rebels might upset Tennessee before their inevitable slide.
Mississippi didn't beat Tennessee (it wasn't close), but they were 5-3 in conference games before really hitting the skids. They're 19-9 now, and (converse of Alabama) need to lose their last three to match our projection. But the basic storyline came true. And Mississippi is no longer in the seedings in the Bracket Matrix. Though it did fluctuate, their Dance Chance—25.2% in late December—is now 24.3%, barely changed after two months.
Tonight's dominant road win by Alabama shows that power ratings really do work. Ok, you can't declare that based on one game—otherwise one wrong game could prove the reverse. But the trajectory of both teams over the last two months shows that even fairly early in the season you have a good basic idea of how the 2nd half of the year is going to play out for the majority of teams. And you don't get there by looking at a team's record alone.
That should be obvious at this point but there are still people stuck on Win-Loss record as a predictor going forward. It's kind of funny, but the pollsters really caught on to Ole Miss being a glass cannon but they didn't catch on to Alabama's quality for a long time. The Tide weren't ranked until they won 7 more games (vs. one loss). But they were a projected 2-seed or 3-seed in the Dance Chance a month prior to that.
And further, we looked at what our final score projection for tonight's game would have been back on December 24th, and the result was: Alabama 90, Ole Miss 75.
A lot less scoring, but the exact margin, from a prediction more than 2 months ago.
The lesson is: check your local power ratings before fretting about your team's NCAA tournament chances—and also before assuming they'll be in based on an early win streak.
Comments