Pulse Check: Champ Week

If any of my readers are familiar with Bracket Science, you will recognize this post, but it may not be what you are expecting. During Champ Week, Pete would post an entry called the Pulse Check, which would gauge potential tournament teams based on specific criteria that he found to be useful in identifying Final Four and Champion contenders. I, as usual, am going to take a different approach. Since a lot of the data in the Pulse Check will eventually be compiled into the Spreadsheet, I am going to present below the 2016 data that will not change, and when all games have finished, the rest of the data usually in the Pulse Check will be compiled into the Spreadsheet. This will save myself a lot of repetitive work. Instead, I am going to look at the Pulse Check from years past to see if there are any patterns or tell-tale signs that will aid us in picking the Perfect Bracket. I guess you could say that I am checking the reliability of the Pulse Check.

As I stated above, here is the data that will not change from here until the start of the tournament. If you compile your own Spreadsheet, you are more than welcome to use it. Also, feel free to double check my work. I complete stuff in binges, so all it takes is one mis-click of the mouse and one team has an Elite 8 appearance that belongs to another team. (Please use the comments below to report typos.)



Legend: Prev = Went to Previous Tournament, CY (4+) = Division 1 Coach for at least 4 years,
E8 App = Coach has at least 1 Elite 8 appearance, Conf = Power, Major, Small
FOM+ = Team's style of play benefits from Freedom Of Movement officiating emphasis.

Checking the Pulse of the Pulse Check

The very first thing I am going to do is show the Pulse Check from the 2014 season. This was done by Pete after the 2014 bracket had been unveiled. For those unfamiliar with the Pulse Check, I will explain. Each column represent a stat for that team. If that stat meets a certain threshold, it passes the test. If that stat does not meet the threshold, it fails the test (marked in red) and counts as a Disqualifier for the team. (The threshold was defined by Pete based on his historical study of the brackets). I have added an additional column to the very right (labeled TW), which was done after the tournament had finished (and it stands for Tournament Wins for that team).

The first thing that most bracket pickers will examine is the Pulse Check for the National Champion. As the chart shows, Connecticut had 8 Disqualifiers (DQs). The National Runner-Up -- Kentucky -- had 5 DQs. The team with the fewest DQs -- Villanova with 0 -- only won 1 game in the tournament. If we look at it by seed group, we can see some strange patterns. In the 1-seed group, everything held to expectations as teams with the fewest DQs had the most tournament wins. Yet in the 2-, 3- and 5-seed groups, the team(s) with the most DQs had the most tournament wins in those groups. In the 6-seed group, there is no pattern as three teams had 4 DQs with varying results, and one of the teams with 4DQs (OHST) won the same amount of games (0 wins) as the team in that group with 10 DQs (MASS). At this point, it may seem like the Pulse Check is too unreliable to make any picks, but some outcomes that do not match expectations in this year actually conform to expectations in the 2013 Pulse Check (we'll examine further below). So the real question is this: What does it tell us?

I believe the message sent by the Pulse Check is quite simple: If we are expecting a crazy year for the tournament, we need to apply the Pulse Check in a broad approach rather than a specific approach. If we are taking a specific approach, then we would let expectations based on DQs guide our picks. For example, we would promote FLA, ARI, NOVA and DUKE to the Final Four based on having the fewest DQs in their respective regions. Another thing we might do under a specific approach is to eliminate (according to seed expectations) the team with the most DQs in their seed group in order to have upsets (#1 UVA loses before Final Four; #2 WISC/MICH loses before Elite 8; #3IAST & #4SDST lose before Sweet Sixteen; #5STL, #6MASS, #7UNM/CONN & #8COL lose in their first game). If we followed the specific approach, our successes are green and our failures are red.

What would a broad approach look like? Instead of using the team's specific DQ value to pick advancement rates as we did in the specific approach, we would use the DQ values of the whole group to identify the strength or weakness of the group. For example, suppose we set an arbitrary rule that any seed-group whose average of DQ-values is greater than their seed, it signals weakness of the group. So for a quick calculation: 1s = 2, 2s = 2, 3s = 4.25, 4s = 2.75, 5s = 5.5, 6s = 5.5, 7s = 7.25, 8s = 7.25, 9s = 7, 10s = 8.75, 11s = 7.6, 12s = 8.2, 13s = 9.5 (Since there are only 13 stats, 14- thru 16-seeds would have at maximum 13 DQs and always out-perform our arbitrary rule. In fact, we could possibly ignore 12s and 13s because it is very unlikely that each and every team in these two groups would approach 12-13 stat DQs.) Anyways, greens pass our arbitrary rule and reds fail our arbitrary rule. Our 1s and 2s had the same average-DQ and coincidentally had the same tournament performance: a 1- and 2-seed made the Final Four and another 1- and 2-seed made the Elite 8. Our 3s failed the test and so did their performance: Only one 3-seed met seed expectations by reaching the Sweet Sixteen while the other three failed (and one 3-seed didn't even win a single game). Our 4s passed the test and so did their performance: All 4-seeds met seed expectations by reaching the Sweet Sixteen and one even exceeded by reaching the Elite 8. 5s failed the test and so did their performance: Only one 5-seed met seed expectations by winning one game, the rest did not even do that. 6s passed the test and their performance half-passed: one met expectations by winning one game, one exceeded expectations by winning two games, and two failed to win a game (but all of this may need to be balanced with expectations for 11-seeds). 11s passed our test with flying colors and their performance did as well: Although two failed to win a game, the other two far exceeded expectations by reaching the Sweet 16 and the Elite 8 -- giving a total of 5 tournament wins for a seed-group not expected to win a game. By now, you should see the patterns I am trying to establish. If you would like, you can do the 7- thru 10-seeds on your own (7-seeds will be the oddballs to the pattern), but the pattern and results of the broad approach would be a far safer approach than that of the general. I always love feedback and participation, so if you find any other patterns or approaches, please leave them in the comments section.

Checking the Pulse of the Pulse Check.........In a Saner Year

As I stated above, the 2013 Pulse Check showed more results meeting expectations than the 2014 Pulse Check. I just wanted to take the opportunity and do a quick run-down that shows the reliability of the Pulse Check in a not-so-turbulent year.

For starters, this Pulse Check was done before the bracket had been unveiled and before all teams had finished all of their conference tournament games. Nonetheless, it is approximately close to what the values would have been when the bracket was unveiled.

If we took the specific approach, we would have LOU, KU, IND, and OHST in the Final Four. We would also eliminate (according to seed expectations) the team with the most DQs in their seed group in order to have upsets (#1 GONZ loses before Final 4; #2 GTWN/MIA lose before Elite 8; #3 UNM & #4 STL lose before Sweet 16; #5 UNLV loses first game; not all 6s, 7s and 8s are accounted for on this list so there is no way to know which would be chosen by the specific approach). While the specific approach fails to predict the Final Four based on fewest DQs in a region, it nails the failed seed expectations for the worst team (by number of DQs) in each seed-group.

Conclusions

So how do we tie all of this together and put a pretty bow on top of it? For starters, we could expand the list of DQ-stats up to 16 so that we can apply the broad approach to 14-, 15- and 16-seeds. If we don't want to add new stats or find enough stats that are reliable, another option would be to weight certain stats: If a team did not pass that stat, it would count as 2 or 3 DQs rather than the usual 1 DQ. The most important thing to take away from all of this: Whether it is the Pulse Check or any other system of analysis, know when it works and know when it doesn't work, and when in doesn't work, still try to find something useful from it. I can say this from experience that when something does not work at least half of the time, then it is not worth any of my time. I personally like the Pulse Check because it feels like I have a cheat sheet to the tournament. I'm going to continue working with it for the next few years, trying to find the right combination so that my cheat sheet gives me more As than Fs.