“Don't judge each day by the harvest you reap but by the seeds that you plant.” ~ Robert Louis Stevenson
There's little else that resembles March Madness. There are few sporting events that can duplicate the emotion, the jubilation and the finality of the NCAA Men's Basketball Tournament. Every March, 68 of the best college basketball teams in the United States are provided a reset. Your program's rich history or this season's domination of conference play mean very little now. Each school is set upon (somewhat) equal footing with 67 others in search of a national championship.
I say "somewhat" because the schools are, as we all know, seeded. The top team in the Atlantic Coast will probably be a one-seed. And the winner of the annual Big Sky tournament is slotted into a 14, or thereabouts. I'm not saying they'll always win, but the haves do start the tourney playing the have-nots.
These top seeds have rightfully earned their placement based on a long regular season inside and outside their conference. But how often does this translate into success in the national tournament? Will the best teams rise to the top? Everyone's talking about the mid-major trend: The Butlers, the Wichita States and the Florida Gulf Coasts. Is there a trend towards more success at the higher seeds? Is there more parity?
I have analyzed the performance of seeds in the NCAA Men's Basketball Tournament since 1985 and, where appropriate, I have also reviewed the last ten tournaments to identify whether any trends are establishing themselves. The following charts are broken down by round to help you confidently fill in that bracket of yours.
THE FIRST ROUND
I know, I know. The first round is now technically the four games that open the tournament and whittle the field from 68 to 64. I'm sorry Mr. NCAA, but we all still call the opening Thursday/Friday games “The First Round”. The following graph shows the probability of the favorite (i.e., the lower seed) winning in The First Round. Like expected, the percentages gradually decrease as the seeds increase. The 5/12 myth appears to carry some weight as the probability of a 5-seed winning is below that of a 6-seed. And, since 1985, 8-seeds win less often than 9-seeds.
Many claim that parity within the tournament is a relatively new phenomenon. Perhaps more underdogs are surviving further into the tournament; however, the graph above shows that the winning percentages of all The First Round matchups are fairly consistent when comparing tournament games back to 1985 with those over the past ten years. In fact, only the 5 and 6-seeds have had materially less success when narrowing our analysis over the shortened time horizon. In both cases, the winning percentages dropped 7% and 4%, respectively. Likewise, 3 and 8-seeds saw significant increases in their First Round winning percentage over the past 10 tournaments. 3-seeds have won 5% more First Round matchups over the recent ten year period and 8-seeds increased by 7%.
Based on these results, I have categorized The First Round games into three separate categories: The slam dunks, the free throws and the coin flips. The 1 through 3-seeds are the slam dunks. At over 90%, these teams have proven year after year to be virtual locks to advance to the next round. The 4-seeds are like the free throws. You should have no problem, but about one of every four is an unexpected miss and you should treat your bracket as such. The question becomes which of the 13-seeds will upset a 4. Finally, the 5 through 8-seeds are a near coin flip. Based on my analysis, a 5, 6 and 7-seed will be upset twice for every five games and there are pretty much two upsets a year in the 8/9 match. Therefore, it might be prudent to select two 12-seeds, one or two 11 and 10-seeds, and finally two 9-seeds.
THE SECOND ROUND
It's fun to pick the upset specials in The First Round. But we all know you didn't actually see Florida Gulf Coast with your own eyes as they rolled through the Atlantic Sun tourney last year before knocking out Georgetown and San Diego State from March Madness. I know you still showed everyone your bracket last year, but, in the end, you were just lucky. The Second Round is a different animal. You need to treat each pick as if they could go all the way. You must decide whether you really want to risk giving up on a top seed this early. The First Round is about minimal casualties. The Second Round is about pushing forward.
The following graph depicts the winning percentage of seeds in The Second Round since 1985. This chart will help you identify which seeds are likely to move forward and which seeds you should stay away from. Based on these results, I wouldn't be giving up on any top seeds yet. The analysis also suggests that you can start crossing off 7-9 seeds and any team 13 and higher.
2-seeds in The Second Round are almost as intriguing as The First Round's 5/12 matchup. 2-seeds can only meet up with 7 or 10-seeds in The Second Round and 69% of the time the 2-seed will win. However, their winning percentage illogically drops to 60% if they happen to square off against a 10-seed and increases to 75% versus a 7. The numbers seem to lean towards advancing a 10-seed if you are going to take a chance with a Second Round upset of a 2-seed.
The results from the chart above can be simplified by grouping several of the seeds. There are a few distinct cliffs in the results. For example, 1-seeds have far better results than 2-seeds, the results of 7-seeds drop dramatically from those of the 6-seeds, etc. The following chart captures each group of seeds and examines whether or not The Second Round results have changed significantly over the past ten years.
As you can see, there is very little movement in the results of the top seeds within the past ten tournaments. However, the winning percentages of the 10 to 12-seeds have actually dropped in the recent past whereas the 13 to 15-seeds have had a bit more success. These movements should be taken with a grain of salt since the sample size of higher seeds in Second Round games is much smaller. In fact, just under 20% of teams advancing to The Second Round are a 10-seed or higher.
Now, which teams should you consider if you do take a chance in advancing this year's Cinderella to The Sweet Sixteen? Sending a high seed onward can be quite the commitment. You could be dumping the best looking girl at the dance for the new girl in town. The graph below shows each seed's success in The Second Round since 1985 against high (i.e., 9-16) seeds from The First Round. For example, 6-seeds can meet either 3-seeds or 14-seeds in The Second Round and the graph below indicates that 6-seeds are successful against 14-seeds in 86% of their matchups.
I picked up two takeaways from the graph above: 2, 3 and 4-seeds only win about 60% of their games against 10, 11 and 12-seeds, respectively, and 10 and 11-seeds have never lost to a higher seed in The Second Round. That being said, there is a very small sample size of 7, 10 and 11-seeds facing off against higher seeds in The Second Round. Regardless, I would focus on the 10 through 12-seeds if advancing a high seed to The Sweet Sixteen. Essentially, you can lock in the best of both worlds. You have a 40% shot if these high seeds face off against a heavy favorite in The Second Round and they are practically a sure thing if they meet a team seeded even higher than they are. The question then becomes: How many 10 through 12-seeds should you advance and, more importantly, which ones?
Since 1985, between two and three high seeds (i.e., schools that fall between a 9 and 16-seed) make The Sweet Sixteen on average. This rate is only slightly higher when narrowing the analysis to the past ten years, but still falls between two and three teams. Two of the last three tournaments have seen four high seeds advance to the Sweet Sixteen and, since 1985, the tournament has never seen five. Likewise, there have been two instances since 1985 (1995 and 2007) where none of The First Round underdogs have advanced past The Second Round. Therefore, if you are advancing higher seeds through to your Sweet Sixteen bracket, advancing two, maybe three, 10-12 seeds would be a reasonable move.
THE SWEET SIXTEEN
By the time The Sweet Sixteen begins, you'll have a fairly strong indication of whether you still have a shot at winning your pool or whether you can start researching for your Fantasy Baseball draft or your office Masters pool. The strategy begins to shift from identifying Cinderellas to selecting the underrated among the best in the country.
It's time to put an end to the Cinderella discussion. You recall Loyola Marymount's tragic story, Gonzaga's initial 1999 run or George Mason's improbable 2006 tournament. These teams, however, are among only a selection of upstarts that made it to The Elite Eight from a 10, 11 or 12-seed. The following graph shows how few teams above a 3-seed actually move on to The Elite Eight. Based on these results, you should expect about five or six 1 through 3-seeds to advance and the remaining be filled out by the field.
And you're wrong if you think there is a trend towards more upstarts punching tickets to The Elite Eight. The chart below narrows the scope to the past ten tournaments and the percentage of 1 to 3-seeds increases by 3%. In other words, you are much closer to expecting six 1 through 3-seeds to advance. As for the aforementioned Cinderellas, since 1985 there have been 13 10 through 12-seeds advance to The Elite Eight, but only three over the past ten years – George Mason in 2006, Steph Curry and Davidson in 2008 and VCU three years ago.
The problem with the pie charts above is that they reflect the fact that fewer high seeds are expected to make it to The Sweet Sixteen to begin with. I need to make sure all teams who survived the opening week are placed on equal footing. The following graph attempts to do just that. I have looked at the winning percentages of each seed in Sweet Sixteen matchups since 1985.
The results give me the confidence to move forward any 1 or 2-seeds left in my bracket. 8-seeds have had surprising success in The Sweet Sixteen, but their seven victories since 1985 are only from a sample size of ten. The 3-seeds have won half of their Sweet Sixteen battles; however, their success depends on the level of their opponent. The 3-seeds are 41% versus 2-seeds and they win 65% of their Sweet Sixteen games when facing a higher seed.
The most striking result from the graph is that there is very little disparity between the success of 4 and 6-seeds and that of 10 and 11-seeds. You could argue that taking a chance and advancing the 10 through 12-seeds will have as much success as penciling in the 4 through 6-seeds into The Elite Eight. The key determining factor is the likelihood of the higher seeds moving on even further. Since 1985, there have been 22 teams at a 4 through 6-seed advance into The Final Four as opposed to only three squads who were a 10 through 12-seed.
THE ELITE EIGHT
Only one thought should cross your mind once you've narrowed your brackets down to eight teams. Who do I think can win it all? Scratch them out if you don't think one of these eight teams can cut down the nets. Even the casual NCAA fan can make somewhat of an educated choice at this stage since they have either recalled seeing one of these schools play in last year's tournament or they have caught them this season on a Saturday afternoon.
The chart below shows the distribution of schools advancing from The Elite Eight into The Final Four. As you can see, 74% of the teams moving on are a 3-seed or better. The other notable takeaway from the chart below is how, on average, there are half the 2-seeds than there are 1-seeds and there are half the 3-seeds than there are 2-seeds. Based on these results, you will likely want to advance one or two 1-seeds to The Final Four, one 2-seed, one 3-seed depending on how many 1-seeds you expect to move on and, finally, one from the remaining field.
The Elite Eight data showed a surprising shift when comparing results dating back to 1985 with results over the past ten years. I have already established that the majority (specifically ¾) of teams making it this far consist of the top three seeds in each regional; however, how do the best of the best fair at this point of the tournament?
The graph above shows that 1-seeds have won more than they've lost in Elite Eight matchups since 1985, but that they have regressed back to the mean in the past ten years. It is clear that a certain parity now exists amongst the top four seeds. In addition, the confidence of higher seeds (i.e., those above a 4-seed) is growing as they have claimed victory in 47% of Elite Eight battles over the past ten years and in five or their last eight.
THE FINAL FOUR
After the field is pared down to four teams, your brackets can be categorized into one of three possible scenarios: (i) You and the rest of the world have Louisville, Syracuse, Duke and Kansas and it's going to come down to a tie-breaker, (ii) you have nobody left but the flier you're riding on an 8-seed Baylor, or (iii) you never read this article and you were mathematically eliminated before the East Regional Final ever tipped off.
The composition of The Final is typically made up of 1, 2 or 3-seeds. Referring to the chart below, you will see that 83% of National Semi-Final winners are 3-seeds or better. The highest seed to ever advance to The Final was an 8 seed. This happened twice: Villanova in 1985 and Butler in 2011. In fact, since 1985, no 7-seed has ever even qualified for The Final Four let alone a Final and 6-seeds have only appeared in The Final twice themselves (the Danny Manning-led Kansas Jayhawks in 1988 and Michigan's Fab Five in 1992).
As with the earlier rounds of the tournament, I have again analyzed the winning percentage of select seeds in The Final Four (or National Semi-Final) round and compared the results dating back to 1985 with the results over the past ten years. Two trends jump out when looking at the historical data since 1985 – the winning percentage of 1-seeds are approximately 10% higher than that of 2-seeds and, secondly, the difference in winning percentage between the remaining seeds is only marginally different.
The past ten years tell a more striking story. The winning percentage of 1-seeds in Final Four games is still out ahead but jumps 14% to 71% over the past 10 years which is nearly 30% higher than the winning percentage of 2-seeds. In fact, no 1-seed has lost a National Semi-Final game since Connecticut fell to a 2-seed from Michigan State in 2009. Also, there have been seven Final Four games over the past ten years involving a 4-seed or higher and they have won only two of the contests. In fact, both victories were by the same school on Butler's improbable back-to-back visits to The Final in 2010 and 2011.
Coin flip or crystal ball? It is time for you to pencil in who wins the whole enchilada. This is often the quickest decision you make while filling out your bracket. But, isn't this square worth the most points? A correct answer can vault you from 34th to the money and a loss can pry the prize directly from your hands.
NCAA men's basketball champions are a select few. The list of title holders can get a bit repetitive. Since 1985, only ten schools out of the 29 champions had never hoisted the trophy before. A first time champion has occurred only six times over the past 20 seasons and only once in the past ten (Florida in 2006).
Your final selection may be where you try to distinguish yourself from the pack. Let's say you have a 1-seed facing off against a 4 or a 5-seed. The numbers show that you may want to resist the temptation to go with the higher seed and roll the dice with the tiebreaker instead. Since 1985, a higher seed has beaten a lower seed in only six title bouts and it hasn't happened since Florida knocked off UCLA in 2006. Of those six upsets, half of the games pitted teams with more than a one-seed disparity.
The pie chart below shows that just over three of every five tournament winners are 1-seeds. And it's good to note that 90% of winners are among the first three seeds just in case you are thinking of rolling that Cinderella all the way through to the championship. It has been 17 years since a team higher than a 3-seed won it all and that was Mike Bibby and Jason Terry's 1997 4-seed Arizona Wildcats.
The pie chart above is reflective of which seeds have already advanced to The Final so it is biased towards the favorites among the lowest seeds. But, once you arrive to the big dance, how does your tournament seeding affect your performance in the championship game? One final chart shows the winning percentage of seeds in The Final.
Not only do more 1-seeds advance to The Final, but they more often come away victorious. In 17 Finals involving a 1-seed and a higher seeded school, the 1-seed is 13-4 since 1985. For example, 2-seeds have had terrible misfortune against 1-seeds in the national championship game. The only time since 1985 that a 2-seed beat a 1-seed in The Final was back in 1986 when Louisville upset Duke. On the flip side, 2-seeds have won 60% of their games versus other seeds.
The chart above shows quite the success rate for 3 to 4-seeds over the past 10 seasons; however, the sample size over the most recent ten years is only three games. Based the results above, your safest bet is to crown the 1-seed if you have any left. If you gave up on the top seeds prior to The Final, the results are not significantly different above a 1-seed to trust anything other than your gut.
LOOKING FOR MORE
For the 2014 NCAA Men's Basketball Tournament, I have also analyzed the impact of a school's conference and the experience of their head basketball coach in the following articles:
You can find a list of the entire 2014 NCAA tournament field in my Cheat Sheet which also summarizes my findings from this analysis and in the two other articles listed directly above.
Bob Sullivan writes periodically for SportingCharts.com and can be followed on Twitter at @mrbobsullivan.