Previewing Day 2 of the Big Ten Conference TournamentPosted: May 25, 2011
Well, day one is in the books on the Big Ten Tournament and Ohio State and Purdue picked up victories. Ohio State had an 8th inning rally which accounted for all five of their runs to beat Minnesota 5-3. Purdue was down 2-0 before firing off seven unanswered runs which, behind the right arm of Matt Morgan was plenty to cruise passed Penn State. You can find plenty of recaps around the web, so I won’t spend too much time on that.
Let’s look ahead to tomorrow where Ohio State will take on the top seed Illinois Fighting illini. Meanwhile, The Boilermakers win got them a match up with the second-seeded Michigan State Spartans. As a result of their losses, fifth-seeded Minnesota will play sixth-seeded Penn State, as well.
What I will do is use a method called the Log5 to determine the likelihood each team wins their games tomorrow.Before we can use the Log5, we must establish a talent level for each team. In my methods, there’s three ways we can go about it: using raw conference winning percentage, using a winning percentage based on actual runs scored and allowed, using a winnings percentage based on estimated runs scored and allowed.
First, let’s look at raw winning percentage. The winning percentage in conference play for each team is below:
From here, it’s just plug-and-chug with the winning percentages into the Log5 formula. The probabilities, based on W% alone, that each team will win the game tomorrow are:
Minnesota v. Penn State: Minnesota should win 54.2% of the time.
Michigan State v. Purdue: MSU should win 54.4% of the time.
Illinois v. Ohio State: Illinois should win 53.5% of the time.
Note: I’ve given Ohio State a 5% boost for the tournament being held in Columbus in all of the probabilities that I’ll list. They do have a home field advantage in playing, well, at home. In the Major Leagues home teams get between a 4-5% boost over a road team of equal strength. I’m just guesstimating here; I can see the argument for the home field advantage to be even greater in the college game.
Now, in a 24-game season like the conference schedule is in the Big Ten, almost anything can happen (as cliche as that sounds). The data suggests that Illinois is not as good as No. 1 seed and that Purdue is better than a No. 3 seed. What I’ll use now are runs scored/allowed in conference play to get a Pythagorean Win %. Using that, we get the following PW%’s for each team:
Now if we input those into the Log5 we get the following probabilities:
Penn State: 52.9%
Ohio State: 56.4%
Illinois outscored their opponents by just ten runs, the same amount Penn State outscored their conference foes by. That put both teams fourth and fifth in the conference behind Purdue (+29), MSU (+29), and Ohio State (+14).
Now, we can get even more granular. Using my estimated runs scored and allowed — which is based on a runs created method and looks at all of the underlying statistics like hits, walks, strikeouts, home runs et al. — to get a winning percentage. Since I don’t have all of that data for only conference play, I’m going to use full-season data in this winning percentage. These are not adjusted for schedule, however, which is something that should be done with half of the teams games (or more) being against non-conference teams. The EW% for each team are:
Here, Illinois takes even more of a hit. The Illini were dead even in run differential (actual runs, not estimated) over the entire season (a.k.a. they were outscored by 10 runs in non-conference games) and their underlying statistics reflect this. Meanwhile, Penn State had a very solid year all around and Purdue is quite possibly the best team in the Conference. I wouldn’t be surprised to see them win the tournament. Speaking of, the EW%’s put into the Log5 yields these results:
Penn State: 68.5%
Ohio State: 54.9%
Here, Penn State looks like a fine bet to advance over Minnesota with the Purdue-Michigan State game being a proverbial coin flip. Ohio State comes out ahead against Illinois, but remember they have a 5% boost due to location. Lower, raise that as you see fit and you’re essentially staring at another coin flip game.
Now, there are obviously caveats. The first being that I didn’t account for any injuries at all. The second, my estimated runs scored aren’t completely ideal — I’m in the process of recalculating them to work better in the lowered run environment of 2011 versus 2010. Truth be told, I’ve never been fully confident in them an always viewed them as estimations. Thirdly, none of the estimated runs scored/allowed figures are adjusted for park or competition. This is something that should be done.
All of that said, tomorrow features at the very least two games that should be entertaining to watch and a third that isn’t so grossly lopsided that it isn’t worth watching. For the record, I’ll take Michigan State, Ohio State and Penn State to win.