You may have heard that the Wake Forest basketball team has an important game tonight @ Clemson. As the marquee bubble game of the night, I decided to try to quantify its importance to our NCAA Tournament chances. To analyze this, I first compiled our remaining win percentage odds from Kenpom. Here is the current breakdown:
@ Clemson: 41%
@ Duke: 23%
@ Virginia Tech: 47%
Next, I assigned NCAA Tournament odds to each of our possible remaining records:
Using a random number generator to simulate our season 1000 times, I was able to combine our current expected win distribution with our projected NCAA Tournament odds. The result? (Drumroll please...)
49%! So, as a starting point, our NCAA Tournament chances are roughly a coin flip. Now how much would tonight’s game swing that percentage?
I re-simulated the season 1000 times, first with our Clemson win% at 100%, then again with our Clemson win% at 0%. Here’s that table:
In this table, the percentages on each record row represent the likelihood of that record, given the Clemson result of that column. For example, if we beat Clemson, there is a 21.3% chance we finish 10-8. If we lose to Clemson, there is a 3.4% chance we finish 10-8. These win probabilities are then combined with the aforementioned NCAA Tournament odds, resulting in the percentage at the bottom of the column: a win vs. Clemson brings our odds up to about 64%, while a loss would drop us to about 35%. In other words, tonight’s game swings our odds nearly 30%.
Additionally, the swing is essentially centered around 50%, meaning: after tonight, Wake will either be clear favorite or clear underdog to make the tournament. That’s a huge difference. But how does that compare to the rest of our schedule?
I repeated the above process for each of our five remaining games, as if each was the next game, and I ended up with a pretty interesting result:
In essence, taken individually, all five games are equally important. They all swing our tournament odds by roughly 30%. The percentages change dramatically (a win over Duke moves us to 68%, while a loss drops us to 39%; a win over Pittsburgh moves us to 53%, while a loss drops us to 24%), but the delta is consistent. So what’s going on here?
I believe this result is a byproduct of a couple related factors:
- The way I set up our NCAA Tournament odds, the biggest jump is from 8-10 to 9-9
- Because all five of our remaining games are relatively winnable and losable, no one game is particularly essential to reaching 9-9 (note the consistency of the values under the Win column in the 9-9 row)
To test this idea, I pretended that 8-10 would likely put us in the field. I moved the 9-9 odds to 95%, and moved the 8-10 odds to 75%, so the second win is the critical one, rather than the third:
This result makes sense: if you believe we just need 2 more wins, the Duke game is relatively unimportant, while losing the Pittsburgh game would be devastating. Conversely, if I change the odds to emphasize the need to reach 10 conference wins or more (lets say 10% for 8-10, 20% for 9-9, and 75% for 10-8), the Duke game takes on a much larger meaning, relatively speaking:
Ok, enough charts - what’s the key takeaway?
The simple takeaway is that Wake is a quintessential bubble team right now, and every game is critical. Clemson is a massively important game, I don’t mean to undersell that. In fact, I’ll even repeat myself:
After tonight, Wake will either be clear favorite or clear underdog to make the NCAA Tournament.
But even in that context, it’s important to remember that — with ballpark percentages — we’ll have around a 1⁄3 chance of missing the tournament even with a win tonight, and around a 1⁄3 chance of making the tournament even with a loss tonight. So, long story short: tonight’s game is huge, but not make or break.
I hope that analysis makes sense. If you have any questions, or if you want me to run some different scenarios, let me know in the comments. If you guys like these charts, I can certainly keep doing them as the season progresses. Enjoy the game tonight and Go Deacs!
-Using this kenpom data relies on the (false) assumption that the games are independent, meaning that they do not influence each other. In reality, the Clemson result will have some impact on our odds in the next four games; treating the odds as constant understates the variability, which causes the tail ends of the distribution (i.e 11-7, 6-12) to be understated. Overall though I believe this is a minor issue which does not significantly impact the conclusion.
-This analysis does not adjust for different types of the same record. In other words, I am treating all 8-10 records identically. Would a 2-3 finish with wins over Louisville and Pittsburgh be a better resume than a 2-3 finish with wins over Pittsburgh and Virginia Tech? Maybe. I don’t buy that that’s necessarily true. Either way, I did not attempt to adjust for it.
-The NCAA Tournament odds I used (e.g. 75% if we reach 10-8) are entirely subjective. It’s just my opinion, based on my analysis of the bubble.