The 2016 NCAA Tournament field is set: 18 teams will punch each other in the face over the next few days in an attempt to advance to either Columbus, Ohio or Providence, Rhode Island. This marks the first time in history that reasonable humans have willingly charted a course to two interchangeable cities that exist purely to provide some sense of existing human life en route to actual places with purpose and volition.
Regardless of whether you want to set the Selection Committee on fire for consciously attempting to ruin your team's chances for a title run, things are going to happen in the coming week that may or may not make your face melt. The issue, then, is how likely these first round games will degenerate into bears with crocodile snouts fighting crocodiles with bear snouts. To try and illustrate that awesome state of survival of the fittest (I have no idea why National Geographic isn't broadcasting hybrid animal battles every night of the week), let's turn to lacrosse computing machines to determine goal spread projections for the play-in and first round games.
Note: In the past I've illustrated this through win probabilities, but I think that people have a hard time reconciling what a win probability is designed to represent. This was none more evident than when Towson fans flipped out on me last year when the Tigers hung close to Notre Dame and Drexel fans just about sent a mob to my apartment in 2014 when the Dragons used a monster spurt that spanned the late stages of the first half and continued into the third quarter to swamp Penn at Franklin Field. So, instead of providing fuel for lunatic-inspired ire, I'm showing projections through goal spreads this year as I think it'll be easier to grasp the nature of upsets, over- or under-performing, etc.
USEFUL LACROSSE COMPUTING MACHINE TABLE THING
Here is a useful lacrosse computing machine table thing that reflects blended projected goal spreads for all of the play-in and first round games. The table thing utilizes projections from LaxPower, the Massey Ratings, and College Crosse's Simple Ranking System:
SOMEWHAT USEFUL THOUGHTS ON THE LACROSSE COMPUTING MACHINE TABLE THING
Bullet points, ahoy!
- The two games that everyone seems to be pointing at as ripe for a seed upset -- Duke at Loyola and North Carolina at Marquette -- are grounded in lacrosse computing machine support. Each fall within the Pick 'Em zone, which is especially exciting as the two games will kick off ESPNU's nonstop weekend lacrosse coverage.
- There are some big ol' projected spreads on the board here. Notably, Johns Hopkins-Brown stands out as one that seems to exist in a different galaxy than where fans that believe that lacrosse computing machines are useless toasters masquerading as useful crock pots think the relative spread is between the two teams. But, it's important to note that Brown has spent most of this year blasting fools into orbit while the Jays have kind of overdramatically told the world they were dead or doing great every other week.
- It would be pretty surprising to see Maryland fall in the first round, even with focused positional matchup considerations on the table. Avoiding Hartford and Dylan Protesto would be nice for the Terps, but even if the Hawks head to College Park, Maryland is decidedly the stronger side. What's going to be interesting is whether Maryland draws Albany or Syracuse in the quarterfinal round on a neutral field. (The Terrapins would be about a 1.5-goal favorite against the Danes and in a Pick 'Em situation against the Orange. The Selection Committee did Maryland no favors bracketing the Terps with Syracuse at the low end of the field.)
- Hobart would have been in better shape drawing Quinnipiac than Towson, but them's the breaks when you spin the wheel of lacrosse roulette (the Statesmen would only be about two-goal 'dogs to the Bobcats). While Hartford-Quinnipiac is reflected as a 1.5-goal spread, that line would be enticing to a person wanting to make billions of dollars by assuming a home equity loan and wagering all the cash on that game.
- To provide a sense of accuracy in these projections, LaxPower is bagging around 55% of its predictions within three goals and over 78% of its spreads are within five goals; College Crosse's SRS model is clicking on about 58% of its predictions within three goals and hitting over 82% of its projections within 5 goals. Upsets happen and there is an average goal error of around three goals in the LaxPower and College Crosse models, but these are pretty reliable projection systems. It doesn't mean that the models aren't infallible -- upsets happen (sports!) and there is inherently going to be an error in the projection versus the actual result -- or that stronger models exist (actually or theoretically), but these models are solid efforts to help set expectations.