Projection Time!!!

The MVC conference season is two-thirds complete.  It's a great time of year, because with 12 conference games down, we know quite a bit about each team.  However, there still exists a lot of uncertainty of how the next three weeks will play out.

To help understand what could and very well might happen, I built my standard simulation model that I do around this time every year.  In the past, I've relied on Ken Pomeroy to help with the probabilities of winning.  However, he's hidden that information behind a yearly subscription that I'm not willing to pay.  Boo-urns.  End result is that it made me have to actually do more work this year.

The following paragraph and bullet points describe my method.  Feel free to skip this if you want to save yourself the details.  Just know that, from what I have been able to piece together, it's not all that different than what Pomeroy does, but likely has some different assumptions and I probably don't have as many back end adjustments as he does.  Ultimately, I got data from my favorite college basketball stat website, Statsheet.com, I pulled game by game results (efficiencies and possessions) for each MVC game that has occurred this year (all 60 of them).  I then built my model doing the following:

• Determined each MVC team's overall offensive and defensive efficiencies on both home and the road.  This is critical, because there's a difference in how the teams play, and it's not consistent across the teams.
• Took those raw numbers and adjusted them for opponent strength.  This washes out the bias against teams who have played tougher schedules and now have easier remaining games.  This gives us a feel for how a team would perform (home or away) against an average MVC opponent.
• Used the Pythagorean Expectation Method to turn those efficiencies into an unbiased expectation of winning a game against a standard MVC team.
• Using these unbiased probabilities of winning, I pitted teams against one another using the log5 formula to come up with a simple chance of winning for the home team.
• I then used Monte Carlo simulation techniques to simulate out the remainder of the season.
• In the event of ties, I applied the description of the tiebreaker formula from the MVC website.  I used OOC SOS from RPIForecast.com and didn't change those in each scenario (which implicitly assumes the games that happen in the MVC don't impact games from other conferences, which I'm fine with).

The following is a snapshot of what I was seeing in 5,000 runs:

TEAM	Rank	%Thu
CREI	1.4	0%
WSU	1.6	0%
IllSt	4.4	8%
MSU	4.4	7%
EVAN	5.4	32%
DRAKE	5.5	32%
UNI	6.3	50%
ISU	7.1	71%
SIU	8.8	99%
BU	10.0	100%

This lists the teams in decreasing order of their average finishing rank in the conference, and their probability of playing on Thursday.  Interesting conclusions include:

• My model really likes the remaining schedule for Illinois State (gets BU twice, and is a solid home team) and Missouri State (four road games, but SIU is one of them and EVAN is another, who is a poor home team.  This combines with very winnable home games against ISU and BU).
• It sees Evansville and Drake as really volatile teams over the next several weeks, ending at anywhere from 7-11 wins for each.  Schedules are generally tougher than the two mentioned above.  DRKE can own the tiebreaker against both IllSt and EVAN if they win those next games, which tremendously helps their chances.
• UNI can sweep DRKE with the H2H tiebreaker.  They're one game back (5-7) from DRAKE, EVAN, IllSt, and MSU.  It actually likes their chances tonight against CREI.  If they can pull that game out, it would be huge for them, and really puts pressure on the 6-6 teams -- most notably DRAKE.  Believe it or not, tonight's game against CREI has major implications for DRAKE.
• ISU (That's Indiana State) has similar issues that UNI does.  They've been an awful MVC road team so far and they still have dates @WSU and @MSU along with hosting CREI.  The thing I'll say is that this looks a lot at what they've done -- to the extent that Odum continues to get healthier, it's quite possible what is up there is too punitive.

Things are still so much up for grabs.  Tonight is going to be a HUGE night in the MVC.

Other things I noticed when I built this model:

• EVAN is actually a more efficient team on the road than at home.  While their record is the same (3-3), their offense has somehow been way more productive on the road.  It just is imporant to note that when they come to DRAKE in a week or so, it's no easy task for the Bulldogs.
• CREI scores anywhere, practically no difference between home and away.  Sign of a top tier team.
• WSU's defensive performances at home have been absolutely nasty.  I've got them as almost 20 points per 100 possessions better as a defensive team at home.  Going in there is going to be nearly impossible for the Bulldogs.
• Our ability to score on the road is considerably worse than at the Knapp, which is really the reason why we have such differences in results (4-2 vs. 2-4).  I'd be willing to wager that a lot of that has to revolve around three-point shooting and free throws.
• IllSt's defense is about 20 points per possession worse on the road.  They go from the fourth best defense at home to the worst defense in the MVC at home.  It's puzzling, and something to keep note on.  This is the same case with UNI, and why they've struggled this year.  For some reason they can't get that classic UNI defense to work outside of Cedar Rapids.  Both of these teams are the toughest to beat at home outside of CREI and WSU.