Round 2 Predictions | March Madness 2018

The article below provides our predictions for all 2nd Round games in the 2018 tournament. A brief description of how to interpret the predictions is provided in each section. If you’re interested in further detail on our models, check out these articles:

Win Probability & Spread Predictions

For each game, “Team1 Win Prob” represents our model’s probability that Team1 (the better-seeded team) wins the game. Probabilities below 50 are predictions that Team2 will win the game. The “284 Spread” is our model’s prediction of how many points Team1 will win by (for both Vegas and 284 Spreads, negative values indicate Team1 is favored, while positives indicate Team2 is favored). The “Team1 Cover Prob” is our model’s probability that Team1 covers the Vegas Spread. A Cover Prob below 50 favors Team2 to cover (and above 50 favors Team1 to cover). Note that our 284 Spread and Cover Prob models are generated separately, so there will be games where they disagree (e.g., one is predicting Team1 covers and the other predicts Team2 covers):

Team 1Team 2Team 1
Win Prob
Team 1
Cover Prob
9 Kansas St16 UMBC78-10-7.162
5 Kentucky13 Buffalo67-6012
3 Tenn11 Loyola76-5.5-4.727
2 Cincy7 Nevada57-7-3.639
1 Xavier9 Florida St88-5.5-1191
4 Gonzaga5 Ohio St62-3.5-3.967
3 Michigan6 Houston80-3.5-653
2 UNC7 TX A&M94-7-8.943
1 Villanova9 Alabama78-11-11.275
5 West Va13 Marshall81-12-8.354
3 TX Tech6 Florida45-1.5+3.572
2 Purdue10 Butler92-3.5-10.136
1 Kansas8 Seton Hall81-5-3.441
4 Auburn5 Clemson54-1.5+1.230
3 Michigan St11 Syracuse68-8-663
2 Duke7 Rhode Island56-9.5-0.736

Over/Under Predictions

For each game, the Vegas Total can be compared to our “284 Total” (our model’s prediction of how many points will be scored in the game) as well as our “Over Prob” (our model’s probability that the game goes over the Vegas Total). An Over Prob prediction below 50 sides with an Under, while a prediction above 50 sides with an Over. Note that our 284 Total and Over Prob models are generated separately, so there will be games where they disagree (e.g., one is predicting over and the other predicts under):

Team 1Team 2Vegas
9 Kansas St16 UMBC135.5133.242
5 Kentucky13 Buffalo154.5152.241
3 Tenn11 Loyola130129.749
2 Cincy7 Nevada137.5147.080
1 Xavier9 Florida St160159.949
4 Gonzaga5 Ohio St142.5149.576
3 Michigan6 Houston134137.665
2 UNC7 TX A&M151.5145.446
1 Villanova9 Alabama148.5154.177
5 West Va13 Marshall159.5157.745
3 TX Tech6 Florida134144.160
2 Purdue10 Butler144.5155.979
1 Kansas8 Seton Hall154152.841
4 Auburn5 Clemson146.5139.918
3 Michigan St11 Syracuse129.5141.860
2 Duke7 Rhode Island149149.654

Round 1 Recap

Here is how each model’s predictions fared in first round games, with spread and total accuracy graded against closing numbers for each game:

  • Win Probabilities: 26-6 (81%)
  • 284 Spreads: 17-13 (57%)
  • Cover Probabilities: 16-15 (52%)
  • 284 Totals: 18-14 (56%)
  • Over Probabilities: 16-16 (50%)