4 Big Stories from Tourney Day 1 (by the Stats)

Harvard was the big upset, but statistically, Cal overcame some heavy odds as well.

Statistically Optimized Drinks on Harvard!

Most people would look at our 89 percent odds for a New Mexico victory and walk away. Harvard did it and said, "So you say there's a chance?"

Even though Harvard headed into the game both with poor 10.88 percent victory odds and the stain of knowing that "poor" and "Harvard" were in the same sentence somewhere, the Crimson were able to pull it out of the Lobos last night.

By the analytics, it was an upset, although perhaps not as big of one as you'd think. After adjusting for opponents, the Crimson actually had a more efficient offense than New Mexico this season. Harvard averaged 1.078 points per possession scored; New Mexico averaged a slightly less 1.045. The problem comes, however, with Harvard's 0.994 points allowed per possession, far below New Mexico's 0.956 mark.

It was that defense that spurred the way for Haaavahd though, holding los Lobos to 0.984 points per possession last night. Harvard, meanwhile, finished the game pretty darn close to their regular season offensive efficiency, finishing with 1.079 points per possession. Now that's consistency.

About Them 12-Seeds...

Earlier this week, I wrote an an entire article about the most likely 5-12 upsets in the NCAA Tourney. Naturally, it was numberFire's most read article of our March Madness coverage. Naturally, the two least likely upsets by the numbers were the two that actually happened, with Cal beating UNLV and Oregon beating Oklahoma State.

I love statistically unlikely outcomes sometimes, don't you?

It may not seem that way, but these upsets were almost in Harvard's expertly manicured ballpark. Oregon only had a 27 percent chance of winning, by the numbers. Cal only sat at 19 percent.

Oregon may have been the Pac-12 champs, but their analytics were awful. They hold the No. 127 offensive efficiency in the country and a 24.9 percent consistency rating that didn't leave much room to improve marked their resume. However, they didn't exactly need it when they were able to hold Oklahoma State's No. 78 offense to 0.775 points per possession. That sits 0.291 less points per possession than the Cowboys averaged during the season, a number that adds up quickly as the game progresses.

Cal, meanwhile, held the nation's No. 140 offense and No. 83 defense heading into the Tourney, rankings that sent my bracket into a corner, cowering in fear. UNLV, meanwhile, sat No. 99 and No. 17, respectively. So how did Berkeley manage to pull it out this time? Because UNLV simply couldn't make buckets: the Runnin' Rebels' .373 effective field goal percentage (eFG%) sat .110 percentage points lower than Cal, and they simply couldn't make up the difference elsewhere. UNLV's 0.859 points per possession scored was lower than each of their three Mountain West Conference Tournament games.

(Almost) Good Ole' Southern Cooking

If there was one 16 seed that we thought could pull the upset, it would have been Southern: we gave them a 4.29 percent chance of pulling it off. And wouldn't you know it, it almost happened.

Not that it was much of a surprise that it was close for a while, mind you. Southern finished the season with the nation's fourth best defensive efficiency; yes, that's fourth as in "Top Five in the entire country". Even after adjusting for those oh-so-tough SWAC opponents, the Jaguars only allowed 0.878 points per possession. That number finished only behind Florida, Louisville, and, well, Stephen F. Austin among Division I teams.

And it was that defense that rode against the Zags. Even though they were facing the nation's No. 2 most efficient offense behind Indiana at 1.197 points per possession, Southern was still able to hold them to just 1.000 points per possession in this game. And by holding form with their own 0.906 points per possession number, coupled a free-throw factor (FT/FGA) of .081 higher than Gonzaga, they were able to stay in the game.

Southern couldn't pull it out in the end, but they gave it a good run. For Gonzaga, the win ultimately came from their ludicrous advantage on the boards. Olynyk and Co. grabbed 29 percent of available offensive rebounds; Southern only grabbed three percent of theirs.

The Davidson Epic Fail

I'm pretty sure that I almost caused a partial mental breakdown in some offices yesterday when I tweeted about how we had told people that Marquette wasn't very good, not once, but twice.

But they were in your optimized bracket! the voices on the Twittersphere shouted in unison. You can't say that!

Oh, but I did, because the facts were there: our projections gave Marquette the highest chance of being upset of any team seeded one through four. We gave Davidson a 25.72 percent chance of winning, and by George, they almost pulled it off.

The weakness coming into the Tournament for Marquette was their defensive efficiency: they sat 89th in the country with 0.980 points allowed per possession. Going against Davidson's No. 12 most-efficient offense, there was the potential for a major mismatch. However, it was the Marquette offense that actually sputtered; they allowed only 0.921 points per possession, but they scored only 0.937 themselves. Before the Tournament, their No. 48 overall average was at 1.089 points scored.