Bracket Prediction: Highlighting Historical Indicators of March Madness Success

Understanding a little bit of NCAA tournament history is a valuable tool when trying to predict which schools will prevail. In modern tournament play, the teams that do well share a very particular set of characteristics.

With an eye on the Final Four and the eventual champion, this article highlights which historical indicators matter, which are overrated, and how to use them for bracket prediction success.

Dangerous Combination: Great Offense, Bad Defense

Be very wary of teams that play great on the offensive side of the ball, but then can't play a lick of defense. Last year, Missouri was a No. 2 seed largely because of its dynamic offense, which was the very best in the nation. The Tigers defense was ranked 115th and they became the fifth No. 2 seed to lose to a No. 15.

Since 2003, there have been nine tournament teams to be in the top 10 in offense but outside the top 100 in defense. Almost all have lost in the first round. Only one has exceeded expectations.

Moving backward in time with teams that fit this dangerous profile, the No. 6 seed Notre Dame in 2010 (lost first round), Arizona in 2009 (lost first round), Oregon in 2008 (lost first round), Gonzaga in 2006 (went to Sweet Sixteen—lost according to seed), Gonzaga in 2005 (lost as a No. 3 seed in the second round) and Arizona in 2004 (lost in first round).

In 2003, the Dayton team was seeded fourth. It lost in the first round as well. The only team to overcome bad defense and actually produce an upset? The 2003 Notre Dame team, which advanced as a No. 5 seed to the third round.

This year's teams to be careful about? North Carolina State (106th best defense), Iowa State (122), and maybe Colorado State (98). Regardless of seed, these teams will be out of my bracket very early.

Qualify a "Hot" Team

It is common to hear experts or friends say that they make picks based on how hot the team is to finish the regular season. This philosophy may work for some predictions, but it can't be applied to the entire field.

If the streaking teams comes from a mid-major or smaller conference, the hot streak is not predictive. Those wins against lesser competition doesn't translate to the Big Dance, says Bracket Science.

Yet if a team from a bigger conference is on fire, then it does have significance. Yet this too must be qualified. A team who only wins its tournament but was not playing well before that is not one which should be given more credit for its streak.

If you are going to choose a team because of its great play late in the season, it should be one from a big conference and hot play that started before their conference tournament.

RPI Predicts Nothing

By its very nature the RPI is descriptive, not predictive. The tournament selection committee does not give bids or seed based on what it thinks a team might do, but on what the team has done throughout the season. As a result, it is not useful for winning the office pool.

It does not take into account some of the most important factors for prediction, including margin of victory and efficiency ratings. However, it over-emphasizes strength of schedule and only uses three components to get its results. Simply put, it isn't a very smart index.

The NCAA selection committee uses RPI to choose teams and seed, but if you want to impress, it shouldn't be used to predict potential winners.

No. 8 vs. No. 9: A Little Better Than a Tossup

Don't pick the No.8 just because that team is a higher seed. Since the tournament expanded in 1985, No. 9 seeds actually have a winning record against No. 8s. Because it is about 50-50, the 8 vs. 9 matchup is as close to a toss-up as the tournament gets.

Pete Tiernan of Bracket Science, the tournament stats guru, suggests using the kenpom.com Pyth Rating will boost predictions to over 60 percent.

Overall, these games have little value as the No. 8 or No. 9 almost always loses to the No. 1 in Round 2. But, in the end, a couple extra points may make a difference.

Easy as 1, 2, 3

It has been 16 years since anything but a first, second or third seed has won the NCAA title. Even more telling, 14 times in the last 20 years the champion has been a No. 1 seed.

The Final Four is a little more giving. Nine out of the 40 teams in the Final Four in the last 10 years were outside a top-three seed. None of those teams won, of course, but teams outside that top-three bubble will slip into the final weekend every so often.

But rarely does a team outside the top-5 seeds make the Final Four. 37 of the last 40 teams to make the Final Four were a No. 5 seed or better.

This narrows most Final Four selections down to 20 teams and National Champs down to 12 teams. The historical indicators listed ahead will help decide which of those to choose.

Power Six

Since the expansion, only once (UNLV 1990) has a champion come from outside the six biggest conference: the ACC, Pac-10/12, Big Ten, Big 12, Big East, and SEC. Of the 112 teams that have played in the Final Four since the expansion, only 10 have not been part of these conferences.

Interestingly, almost all of those 10 appearances came from just three programs: UNLV has gone three times, Memphis and Butler twice. Very rarely does Cinderella story actually end with the ultimate victory.

This year, however, it might be different. No non-conference school has ever had the profile of Gonzaga. They have been ranked first for several weeks, will have the benefit of a No. 1 seed's favorable road to the championship, and look great on all the best predictive factors to be explained coming up.

The big-conference lock on the championship may very well be broken wide open this March by the Zags. History says the other smaller conference teams don't have much a chance.

Margin of Victory

One of three best tools for choosing which teams to go on a deep run is margin of victory (MOV). The thought of many is, "teams in a better conference or with a tougher schedule will have lower MOV." This is not the case for a national champ—historically, they have dominated all season long.

Of the last 14 champions, 12 have been in the top four of margin of victory, all winning by an average of 14 points. Six of those teams have been the very best in the nation and twice the second-best. Only once (2011) has the champion's MOV fallen below 10.0 and in that year it was still a respectable 7.4.

Not only is MOV a historical indicator of champions, it can be helpful in choosing the three Final Four games. In the 30 Final Four games since 2003, the team with the higher MOV has won 24 out of 30 (80 percent).

It also has power in picking the Final Four. In the last 15 years, 50 of the 60 (83 percent) Final Four contestants have been in the top 25 in MOV.

Comparing this year's MOV to this historical indicator, only five teams fit the profile of a national champion's profile: Florida, Indiana, Gonzaga, Louisville and Pittsburgh.

Several teams may be popular choices for Final Four consideration, but should not be considered. This list includes Arizona, Miami and Michigan State.

Defense Matters

A stout defense is a requirement to make deep runs in the tournament. There are very few Final Four teams with poor defensive records. Last year, for example, all four Final Four teams were in the top 10 of defensive efficiency ratings, including three of the top four: Louisville (1), Ohio State (2), Kansas (4) and Kentucky (9).

Over the last 10 years, the numbers are actually a little startling. Of the last 40 Final Four contestants, 35 have been in the top 25 in defensive efficiency; four of those five outliers came in just two years, 2011 and 2003.

Without the team of 2003 and 2011, 31 of 32 teams have been in the top 25 in defensive efficiency. Broken down a little further, in that same time frame, 32 of the 40 teams were in the top 20 in defensive efficiency, 20 in the top 10, and 16 have been in the top five.

The bracket implications are clear. Compare your final four to the defensive efficiency ratings. Based on the last 10 years, all should be in the top 25, two should be in the top 10, and at least one should be top five (nine of the past 10 had a top-five defense). If you have teams without a stellar defense, you should have a compelling reason why.

Teams outside the top 20 in defensive efficiency are going to have a difficult time getting to the Final Four. If outside the top 25, it is nearly impossible.

Teams falling outside these guidelines: Michigan (58), Syracuse (23) and Arizona (36).

The numbers don't lie—defense is vital for March Madness success.

Offense Matters More

Offensive efficiency is almost as important for getting to the Final Four, but even more important for winning it all. Of the last 40 Final Four teams, 33 have been in the top 25 in offensive efficiency, 26 have been in the top 20, and like defense, 20 teams have been in the top 10 and 17 in the top five.

In closer detail, the defensive efficiency rank for a Final Four team is actually a little lower than the offensive efficiency rank. So while a Final Four team's defense rank averages at 13.4, the offensive rank average is 15.1. Championship game contestants average much better in both, but defensive rank average is lower again (11.9 to 11.6).

The champions, however, show the necessity of a powerful offense. Over the last 10 years the national champion's offensive efficiency has been first four times, second three times, fourth one time, 11th and 16th. On average, the winning team's offensive efficiency ranking average is 5.7. The defensive efficiency average is 10.8. Great defense gets you to the Final Four, but it is great offense that cuts down the nets.

Teams that fall outside the top 15 in offense have little to no shot to win the title. Even if a team falls outside the top 10, history says this is a high mountain to climb. The exceptions here are 2003 and 2011. Yet, in those years the team was led by an all-time college performance: Carmelo Anthony with Syracuse and Kemba Walker with Connecticut. Not many players in this year's field are capable of doing what those guys did. To win it all in this year and year's past, a team must be one of the best in the nation.

Looking at this year's offensive efficiency rankings implications, Louisville as the champion is on the fence with its 15th ranking. This will be interesting to see how it plays out.

Kansas (unless Ben McLemore pulls a Carmelo), Wisconsin, Michigan State, Miami are not winning the title. Syracuse and Ohio State, while better Final Foul contenders, aren't winning either. Georgetown has no chance.

Of the real contenders, only Florida, Indiana, Louisville, Gonzaga, Duke, Pittsburgh and Michigan have offenses which can take them all the way.

Putting It All Together

With the big indicators in mind now, there are only a few teams with legitimate shots for the national title and only seven contenders for the Final Four.

Starting with seeding: a team needs to be in the No. 5 or better to make the Final Four and No. 3 or better to win it all. Any team outside those numbers should not be chosen.

Then apply the defensive rules. To make the Final Four a team needs to be in the top-25 of defensive efficiency. Michigan is out, VCU too, Kansas St. and Marquette as well.

Then trim by offensive efficiency. Final Four teams need to be in the top-25. This cuts out Kansas, Wisconsin, Georgetown, St. Louis, New Mexico, Oklahoma St., UNLV

And then reduced by margin of victory. To make the Final Four a top-25 rating is necessary. This eliminates Michigan State and Miami.

So the only teams that should be chosen for the Final Four are Florida, Louisville, Indiana, Gonzaga, Ohio State, Duke, and Syracuse.

For the national champion, a margin of victory needs to be over 14 points and offense needs to be in the top six. This includes only Florida, Indiana, and Gonzaga. The Zags don't play in a power conference but this year is so wide open it does not matter for a team as good as Gonzaga. 

Louisville while the No. 1 overall seed, won't have enough offense to win it all. The numbers have them on the outside looking in.

A final tidbit, Pittsburgh fits all the numbers except seed. This is a very dangerous team for the Zags to play in round two and I would not be surprised if an upset occurs. If its wins, Pitt has the path of a No. 1 seed the rest of the way. Interesting to see how that plays out.

While it may be too formulaic, the numbers tell a story for predicting March Madness which is worth taking into account. The historical indicators say only three teams have what it takes to win it all. Now it is time to fill out a bracket, sit back and enjoy the best time of the year.

Link to Printable PDF

Link to Live Bracket

Make your picks for the 2013 NCAA Tournament here with the Bracket Challenge Game