We are back a week later with another bracket projection and another test of my new app. If you haven’t read last week’s edition, go check it out to understand my inspiration and to get a feel of how these will go. As a recap of last time, I went 41/63 for picks overall, which is a really impressive bracket. However, I ended up with 2 out of the Final Four teams, but without a team in the championship game.
The link for the Google Sheet with my most recent bracket is here once again, this time with two brackets: one of which was filled out with my app and the other I filled out on my own volition. Again, every team that is on the line was my selection. Green means that I was right, red means that I was wrong, and the team above a team in red is the team that actually ended up winning that game. With that, let’s get right into it.
In the first pod of the Midwest Region, we’ve got Purdue (0.9895 win probability) easily over Central Connecticut, and, with the first toss-up game, we go with TCU (0.5452) as the favorite over Villanova. Our first S16 team is Purdue, who has a 0.7893 probability of making it to the Sweet 16 and a 0.7924 chance of beating TCU. Considering Purdue is 2nd in Kenpom as of this bracket projection, the algorithm is pretty high on Purdue in the pod, especially considering the code spit out Purdue with a S16 Rating of 5.3517 (again, out of 5). The code doesn’t like the second pod at all, denoted by the fact that Indiana State was actually given the nod above Creighton as the most likely team from the pod. Creighton is also likely, but they have a S16 Rating about 0.8 less than Indiana State’s. Indiana State is the 2nd most likely upset according to the code and has a probability of 0.4472 against Oklahoma, which is why I decided to choose them for the Round of 32. After I decided to go with Creighton (0.7479) over App State (even though App State is the 10th most likely upset), I went with Creighton (0.6367) over Indiana State for the Sweet Sixteen spot from this pod. However, I was wrong on both accounts, as Oklahoma ended up beating both teams in the simulation to make it to the Sweet Sixteen. For the third pod in the Midwest Region, the algorithm spits out both Alabama and Illinois as possibilities for teams from the pod. I went with Alabama (0.6692) over Saint Mary’s, even though the Gaels were the 12th most likely upset in the bracket, and I went with Illinois (0.9064) over UMass Lowell. I was wrong about Alabama, but I went Illinois (0.4620) over Alabama, which worked out in the end. The final Sweet Sixteen team from this region that I selected was Kansas, and Kansas ended up beating Colgate and Nebraska in succession to make it there in the simulation. I went with Nebraska (0.5369) over Memphis in a toss-up game, and Kansas (0.9012) over Colgate. From there, I went with Kansas over Nebraska, even though Nebraska was technically more likely according to the algorithm. However, Kansas’ 0.5926 chance of making it to the Sweet Sixteen was too good to pass up, and it worked out in the end. From there, I went with Purdue (0.7174) over Creighton and Illinois (0.5753) over Kansas because of the uncertainty of the Jayhawks’ pod. Then, I decided to go with Purdue (0.6823) over Illinois, which were all great picks. Overall, very satisfied with the Midwest Region.
With the West Region, I went against the grain a little bit. The first and last pods are normally always the easiest for the Round of 32 since it’s normally the 1- or 2-seeds and then whichever team the algorithm tells me to choose for the toss-up game, and that’s exactly what I did with the UNC-St. John’s matchup. North Carolina is clearly the favorite over Norfolk State, with a win probability of 0.9665, and St. John’s is a slight favorite over Boise State with a win probability of 0.6029. From there, I went a little crazy. North Carolina is the most likely team to make it out of this pod according to the code, but St. John’s is also likely, so I decided to ride with the Red Storm, ideally for some value points. Unfortunately, as you can see, it did not work out well for me, as UNC made a run all the way to the Final Four. With the next pod, I went with BYU (0.7614) over Grand Canyon and Duke (0.7920) over Akron, before narrowly choosing BYU (0.5387) over Duke. Duke wasn’t even listed as a likely team from this pod, so that’s why I rolled with the Cougars, and no upset alerts were flagged from this pod, so just an unlikely sequence of events. Could be worse, but since I went with St. John’s over UNC, I decided to go with BYU over St. John’s for the safe play, so I lost a Sweet Sixteen team in the first round. That’s just how the cookie crumbles sometimes. I felt no need to play it safe with BYU according to the algorithm since it was the clear favorite of the pod and there was no upset alert at all, so I felt comfortable going with St. John’s and BYU, which may have been playing with too much fire, especially considering BYU’s issue of being an alleged “computer tricker.” I performed a little better on the lower side of the West Region. I went with Clemson (0.5828) over Northwestern. Northwestern was flagged as the 5th most likely for an upset, but I went against it because their probability was below the critical point for 11-seeds. 11-seeds above a win probability of 0.4731 have historically won 61.90 percent of first round games since 2001, while 11-seeds below this probability have won only 38.81 percent of games against 6-seeds. I also went with Clemson since they were flagged as the most likely team of the very uncertain pod (one that also listed Kentucky and Northwestern as likely teams). Drexel was flagged as the 8th most likely upset, but I went with Kentucky at a 0.7742 win probability. Then, I went with Clemson (0.4396) over Kentucky with a little bit of value pick, which ended up working out. Unfortunately, the toss-up game predictor for the last pod failed, as Kansas State beat Colorado State (0.5807) for a spot in the Round of 32. I went with Arizona in back-to-back games in this pod, since they were a clear favorite for the pod (5.7082 S16 Rating), over South Dakota State (0.9638 win prob for Arizona), and over Colorado State (0.7741 win prob for Arizona; and Kansas State for that matter). As I mentioned, I went chalk for the Sweet Sixteen of the West Region, with BYU (0.6609) over St. John’s (neither of which made it to the Sweet Sixteen) and Arizona (0.7611) over Clemson. From there, Arizona (0.6128) was the easy pick of the bunch. In hindsight, I would’ve stuck with an Arizona pick for the Final Four team out of this region. Arizona had a F4 prob of 0.3629, while UNC had a F4 prob of 0.2546. I will say, the pick of St. John’s over North Carolina was a little ambitious, but sometimes you have to roll the dice.
In the South Region, we started with Houston (0.9919) over Jackson State and Texas A&M (0.5697) over Seton Hall in the toss-up game. Unfortunately, the algorithm was not prepared for a run by the Pirates, which is exactly what I ensued. I played the numbers and went with Houston (0.8389) over Texas A&M. Seton Hall overcame even greater odds, as Houston had a 0.8770 probability of beating Seton Hall according to Kenpom numbers. With the upset over Houston, Seton Hall took out one of my Final Four teams. Not great stuff. I think even more infuriating is the fact that I had a separate Big East 8/9 seed to make a run and upset a 1 seed in the previous region in St. John’s. Sometimes you can’t get ‘em all. From there, we went San Diego State (0.7206) over McNeese State (which missed) and Baylor (0.7386) over UC Irvine (which hit). McNeese State and UC Irvine flagged the 7th most and 4th most likely upsets respectively, which essentially means that I didn’t want to have the winner of this pod go farther than the Sweet Sixteen. I went with San Diego State (0.4388) over Baylor, which was probably the best pick, even in hindsight. San Diego State had a S16 Rating of 3.6779, followed by Baylor at 3.2072. San Diego State also had a 0.3627 probability of reaching the Sweet Sixteen, while Baylor had a 0.4567 probability. At the end of the day, not the biggest issue with this region (cough cough Seton Hall), so we move. Utah State had the worst S16 Rating out of the entire pod and Colorado (0.5847) had the most likely upset in the field. Give me the Buffaloes. Auburn (0.8615) had the best S16 Rating, reaching a rating of 5.3261, so I went with them over Louisiana Tech. I genuinely thought about Auburn (0.6871) and Colorado long and hard. Give me the Buffaloes part 2? From the algorithm, I went with Auburn in this instance, but you can see in my raw selections, I actually went with Colorado and that was the move. Colorado ended up upsetting both Utah State and Auburn in back-to-back rounds to make it to the second weekend in this simulation. In the final pod, we went with Florida Atlantic (0.5982) over South Carolina (against my better judgment) and then Tennessee (0.9343) over High Point. Now, here’s where we run into a little issue with the S-curve I put together on Monday night (technically Tuesday morning) at 2am. Since South Carolina actually beat Florida Atlantic, we have an SEC matchup (one that will have happened twice by this point) in the second round. So uh. Anyways, Vols (0.7211) on (Rocky) top over Florida Atlantic (and South Carolina for that matter) for the Sweet Sixteen. I went Houston (0.8042) over San Diego State, which actually turned out to be Seton Hall over Baylor (??), and Tennessee (0.5152) over Auburn, which ended up being Colorado instead. With an easier Elite Eight matchup, the Vols fly into the Final Four over Seton Hall. I had Houston (0.6517) over Tennessee, which would have had a 0.5290 chance of making it to the Final Four had they made it with my selected teams. Better than a flip of the coin for the Cougars. Wasn’t meant to be.
Finally, we round it off with the East Region. We go with Connecticut (0.9512) over Fairfield with an in-state matchup and Michigan State (0.5650) over New Mexico in the ensuing toss-up game. Michigan State surprisingly has the most likely chance out of the pod, but I go with the Huskies (0.6164) instead, which works out well after the Lobos take down Sparty in Brooklyn. Both Princeton (6th) and Samford (3rd) are flagged for upsets over Iowa State and Dayton. I decide to go with the most likely upset (Samford; 0.2654) and to ride with the most likely team from the pod, which ends up being the Cyclones (0.6897 over Princeton, 0.8233 over Samford). This all works out according to plan, truly showing the power of gut-reactions and stronger information from the algorithm. A faltering Rebels team is flagged as the 9th most likely upset. I decide against it. Utah (0.6319) over Ole Miss and Marquette (0.8548) over Eastern Washington. Marquette and Utah are both likely teams from the pod according to the algorithm, but I ride with the Golden Eagles (0.5896 over Utah), who have the most likely chance of reaching the Sweet Sixteen from the pod at 0.5419. The final pod in the East Region is… volatile. To say the least. The algorithm spits out Texas Tech (3.3617), Wisconsin (3.1160), and Mississippi State (3.0520) as likely teams in the pod, and with a pod as volatile as this one, I quickly realize (and hope) that the Sweet Sixteen team from this pod won’t make it much further. Texas Tech (0.5271) is barely given the nod over Mississippi State and Wisconsin (0.8848) is chosen over Youngstown State, despite the 11th most likely upset being flagged for the Penguins. From there, we go with Texas Tech (0.3893), because why not? And it ends up being right in the simulation. From the Sweet Sixteen on, we go Connecticut (0.5930) over Iowa State and Marquette (0.5780) over Texas Tech. Connecticut and Texas Tech face off in the Elite Eight, but we still maintain the Huskies as our Final Four team, after they defeat the Red Raiders.
At this point, I’m happy with 2 Final Four teams. I think having 3 Final Four teams could be where I would seal the deal and be extremely happy with my bracket, but there is some solace to be taken in the fact that I got the correct seeds, with 3 1-seeds and 1 2-seed. Heartbreaking to have Purdue lose to Tennessee to overcome a Maui Invitational loss, for both my bracket and as I would imagine as a fan.
All in all, we had less correct picks, but with Purdue in the national championship game, we ended up with more points in this simulation. Like I said, I would be more happy with 3 correct Final Four teams, as I would imagine that would give me a significantly better chance. I’ll have to crunch the numbers in the coming weeks to see if that’s actually true.
That’s it for the second edition of filling out my bracket projection. Hoping to improve as time goes on. If you have any questions, hit me up @TSBBracketology on Twitter.
No comments:
Post a Comment