While it may make you feel better to focus on ‘accurate’ predictions, like having West Virginia win over 50 percent of simulations over Duke or nailing the 1 point difference in the Stanford / Marquette game, it is more useful to look into inaccurate predictions and figure out what caused our forecast to be off.  When a team AccuScore has winning over 50 percent of the time does not win we typically like to pass the buck and point the finger at what we call “deviants”. These are the players and coaches whose performance deviates massively from what their past performance would indicate.  When this deviant behavior causes our 50+ percent prediction to go wrong, we like to point these cases out.

GEORGETOWN HOYAS
The storyline after Davidson came back from 17 points down to beat the Hoyas was Stephen Curry’s amazing second-half.  Obviously, Curry deserves all the credit in the world for leading his team to the Sweet 16, but the Hoyas did a decent job on Curry holding him to 8-21 from the field (5-15 from three point range).  For the game the Hoyas had a MASSIVE edge in FG percentage (63.4 percent to 38.6 percent) and in three point shooting (55.6 percent to 21.4 percent).  The Hoyas also got the slight edge on the boards (26 to 23).

So how does a team that leads the country in defensive field goal percentage lose a game when they have a 25 percentage point advantage in shooting percentage? 

The answer is turnovers.  Georgetown averaged a shade over 13 turnovers per game this year but they committed 20 turnovers against Davidson.  Davidson meanwhile only committed 4 turnovers.  Turnovers cost Georgetown last year as well when they committed 6 more turnovers than Ohio State in their Final Four match-up and proceeded to lose by 7.  Last year, AccuScore was pleased by Georgetown’s turnover problems because it had Ohio State winning the game.  This year, we are not so pleased with Georgetown’s inability to take care of the ball.  It may be time for John Thompson III to consider not running the Princeton offense when he has players who struggle executing it.

UCLA’s CHANCES GO DOWN, BUT THEN GO UP
The UCLA Bruins were able to avoid the upset thanks to Kevin Love’s clutch offense and defense.  However, overall #1 North Carolina was able to dominate Arkansas and after factoring in all the data from the tournament so far, in current simulations North Carolina is actually beating UCLA by a slight 51 to 49 percent margin.  Before the tournament started it was UCLA with the simulation edge. 

However, even though the data is now favoring UNC the Tar Heels’ chances of winning the whole thing is 13.9 percent which is basically the same percentage chance they had in simulations run before the tournament started.  With two wins under their belt and a slight edge over UCLA you would think UNC’s chances would go up significantly, but they haven’t because the East is the only Region where the Top 4 Seeds all advanced. 

Meanwhile, UCLA faces Western Kentucky in the Sweet 16 and no one would be surprised if they faced 7 seed West Virginia in the Regional Final.  That is such an easier path than UNC who has 2 of the hottest teams in the country in their Region (Washington State and Louisville) and they aren’t even the 2 seed, that is former #1 Tennessee.  UCLA is winning the tournament in 20.2 percent of simulations (up 6 percentage points since the start) which is significantly higher than UNC’s 14 percent.