{"id":6532,"date":"2017-03-15T17:36:53","date_gmt":"2017-03-15T21:36:53","guid":{"rendered":"https:\/\/www.furman.edu\/news\/2017\/03\/21\/ncaa-tournament-upsets-furman-math-professors-calculate-the-odds\/"},"modified":"2022-11-06T20:40:17","modified_gmt":"2022-11-07T01:40:17","slug":"ncaa-tournament-upsets-furman-math-professors-calculate-the-odds","status":"publish","type":"post","link":"https:\/\/www.furman.edu\/news\/ncaa-tournament-upsets-furman-math-professors-calculate-the-odds\/","title":{"rendered":"NCAA tournament upsets no surprise for these Furman math professors"},"content":{"rendered":"

This story was written prior to the start of the NCAA tournament.<\/em><\/p>\n

For 12 years running, ESPN writer Peter Keating has been giving readers of his \u201cGiant Killers\u201d blog a \u201cmetrics-based forecast of the big upsets\u201d in the early rounds of the NCAA men\u2019s basketball tournament. Since 2013, those metrics have come from increasingly complex calculations by Furman math professors Kevin Hutson, Liz Bouzarth and John Harris \u201991, and there\u2019s a reason they keep getting asked back: They\u2019re pretty darn good at it.<\/p>\n

For instance, after crunching the numbers in 2016, the Furman trio zeroed in on four first-round games most likely to result in an upset (defined as a team defeating an opponent seeded at least five spots higher). Three of them\u2014No. 14 Stephen F. Austin over No. 3 West Virginia (36 percent probability), No. 13 Hawaii over No. 4 California (30 percent) and No. 11 Wichita State over No. 6 Arizona (40 percent)\u2014ended up being just that.<\/p>\n

\"\"

Liz Bouzarth<\/strong><\/p><\/div>\n

With that in mind, if you\u2019re a Virginia, Cincinnati, Creighton, Minnesota, or, especially, a Maryland fan, you may want to stop reading now if you don\u2019t want your tummy to be even more fluttery when your team takes the court this week.<\/p>\n

\u201cOur most likely upset is the Maryland-Xavier game,\u201d Harris said. \u201cIt\u2019s got a 51 percent upset probability, so our model would say that Xavier (seeded 11th) has a slightly better chance of winning that game than Maryland (seeded sixth) does even though it\u2019s a 6 vs. 11.\u201d<\/p>\n

The models also had Kansas State knocking off Wake Forest in First Four battle of 11 seeds on Tuesday night, which it did, before projecting a 46 percent chance of the Wildcats of upsetting Cincinnati, also seeded sixth, on Friday. UNC-Wilmington, meanwhile, is given a 42 percent of defeating No. 5 Virginia, similar to the chances of No. 12 Middle Tennessee knocking off No. 5 Minnesota and No. 11 Rhode Island sending No. 6 Creighton home.<\/p>\n

That would continue a trend that\u2019s seen No. 12s win 50 percent of their 20 games against No. 5s since 2012.<\/p>\n

\u201cVirginia is classified for us as a first-cluster giant, and UNC-Wilmington is classified as a third-cluster killer. Historically speaking, at least over the past 11 years that we\u2019ve been looking at data, there\u2019s a 40 percent chance in this grouping that killers will upset the giants,\u201d Hutson said.<\/p>\n

Why? Well, that\u2019s where it complicated. The exact methods Hutson and crew use are protected proprietary information by ESPN, and what information is made available isn\u2019t exactly easily digestible to the average person.<\/p>\n

The short version is \u201cgiants\u201d \u2013 the higher seeds \u2013 are lumped into four broad families based on their style of play, as are the \u201ckillers.\u201d Past results show trends in certain matchups between families, and as the data builds on itself the ability to predict future outcomes based on past performances theoretically becomes more accurate.<\/p>\n

\"\"

John Harris<\/strong><\/p><\/div>\n

Virginia, for instance, falls in the category of being a good team that isn\u2019t great at anything, and history shows those squads are vulnerable to free-wheeling, 3-point gunning opponents like UNCW.<\/p>\n

\u201c(These kinds of giants) are kind of generic. There\u2019s nothing special about these giants. They don\u2019t overly excel in any areas, as opposed to cluster-three killers,\u201d Hutson said of Virginia. \u201cThese are kind of below average in certain things, but they are way above average in terms of creating steals and way above average in their tendency to shoot a lot of 3-pointers in terms of other killers.\u201d<\/p>\n

In general, the least vulnerable giants rebound well, especially offensively, and don\u2019t turn the ball over, while the most successful killers play a high-risk, high-reward game. They shoot lots of 3s, they crash the offensive glass, and get steals.<\/p>\n

What makes the work by the Furman professors unique is its ability to use information from past tournaments to increase the personalization of individual game predications in the present.<\/p>\n

\u201cThere\u2019s really no one-size-fits-all idea of what\u2019s going to create an upset,\u201d Hutson said.<\/p>\n

The equation is always changing, too, thanks to new information. In 2017, the professors implemented a new wrinkle called \u201cdecision tree analysis\u201d in an attempt to reflect \u201cmore intuitively how humans make tough choices.\u201d<\/p>\n

\u201cIn creating the decision tree, we rely on the power of statistics to help identify the things we wish we could identify by eye-natural groupings of complicated data,” Bouzarth told ESPN.<\/p>\n

It might be natural to conclude that these are people with whom you should never compete in an NCAA tournament bracket pool, but the reality is Giant Killers isn\u2019t meant to predict the entire tournament, and even if it were advanced analysis of data ultimately ends up in the same place you do: With a guess.<\/p>\n

\u201cOnly friendly ones, and I wish that at we could say that we do better than we do,\u201d Harris said with a laugh when asked if he and his colleagues take part in tournament bracket games.<\/p>\n

\u201cOne of the hard things about this is the analysis we do gives us a lot of insights into these giant-killer matchups, but there\u2019s a whole lot of the bracket that isn\u2019t covered by this sort of analysis,\u201d Bouzarth added. \u201cWe\u2019re not looking at what happens when a 1 plays a 2 in the Final Four. Our data doesn\u2019t really give us that insight, so this is certainly not a tool that is going to give you a perfect bracket the whole way.\u201d<\/p>\n

It\u2019s also important to realize that much more often than not the Furman picks for most likely upsets are relative to the likelihood of other upsets\u2014not to the likelihood of it actually happening. The underdogs are still the underdogs, just not by as much as you might think.<\/p>\n

\"\"

Kevin Hutson<\/strong><\/p><\/div>\n

\u201cIt\u2019s not all that common to have too many games that have an upset probability of greater than 50 percent, because if that were the case then an upset becomes more likely than not,\u201d Harris said. \u201cSome of the games that we get excited about are ones where the upset probability is in the 30-to-40 percent range. That\u2019s a strange way to look at it, because three out of 10 times we\u2019ll be right, and seven out of 10 times we\u2019ll be wrong. But as far as an upset goes, a 30 percent chance of a killer beating a giant is pretty significant compared to a five percent chance.\u201d<\/p>\n

The predictions are built on their initial rankings of the 68 tournament teams, though, and Furman agrees with the NCAA selection committee about who the top teams are \u2013 almost.<\/p>\n

\u201cWe had Villanova, Gonzaga and North Carolina as our 1-2-3, and those three all made it as top seeds,\u201d Hutson said. Instead of Kansas, though, West Virginia is the professors\u2019 No. 4 team despite being seeded third.<\/p>\n

The professors\u2019 rankings also aren\u2019t as bullish on Duke as Las Vegas, which has made the Blue Devils the favorite to win it all at 5\/1, instead ranking Duke 10th<\/sup>.<\/p>\n

In fact, the Furman three give South Carolina, seeded seventh, a better-than-decent 28 percent chance to slay second-seeded Duke in the second round of the regional Furman and the Southern Conference are hosting at Bon Secours Wellness Arena in Greenville<\/strong><\/a>, and as volunteers Hutson and Bouzarth will get a chance to see in person how their predictions match what actually happens.<\/p>\n

Because two of the games are 7-10 and 8-9 matchups and Texas Southern and Troy each have less than a five percent chance of defeating North Carolina and Duke, respectively, in the first round, the second round offers the only chance for Giant Killing. Arkansas has a 10 percent shot at beating UNC in the second round and Seton Hall half that, while Marquette has a 19 percent chance of knocking off Duke if it gets past South Carolina.<\/p>\n

No matter what happens, though, the Furman math department can\u2019t lose. \u201cThat\u2019s the beautiful thing about talking in terms of probability,\u201d Hutson said. \u201cYou\u2019re pretty much right either way!\u201d<\/p>\n","protected":false},"excerpt":{"rendered":"

This story was written prior to the start of the NCAA tournament. For 12 years running, ESPN writer Peter Keating has been giving readers of his \u201cGiant Killers\u201d blog a […]<\/p>\n","protected":false},"author":265,"featured_media":16821,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[15,50,30],"tags":[],"class_list":["post-6532","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-academic-department-page","category-mathematics","category-top-stories"],"acf":[],"_links":{"self":[{"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/posts\/6532","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/users\/265"}],"replies":[{"embeddable":true,"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/comments?post=6532"}],"version-history":[{"count":0,"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/posts\/6532\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/media\/16821"}],"wp:attachment":[{"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/media?parent=6532"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/categories?post=6532"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.furman.edu\/news\/wp-json\/wp\/v2\/tags?post=6532"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}