|
Contributed by Peter Tiernan, ESPN.com Bracketologist
|
Since 1985, the NCAA Tournament has been structurally consistent -- with the same number of teams (OK … so in the last few years there's been a 65th bid), the same seed matchups, and even the same days that rounds are played. That consistency is a big reason why historical data has such relevance in predicting future outcomes.
It would be a mistake, however, to think that the dynamics of the 1985 tourney are the same as those of last year's Dance. As the trends below indicate, the tourney has evolved over the last 22 years. It's become lower-scoring, more guard-dominant and ruled by younger teams -- to name just a few volatile characteristics. Check out the trend lines for these attributes:
Number of upsets If you're thinking that the shocking results of 2006 herald a new era of unpredictability, think again. In three of the four previous upset-laden tourneys, March Madness lurched back to relative sanity, decreasing by four to seven upsets the succeeding year. You have to go all the way back to the first couple of years of the 64-team era to find two consecutive upset-heavy tourneys.
Average number of consecutive tourney bids per team
Last year's tourney field was the most inexperienced since 1998, with the average team possessing 3.2 consecutive trips to The Dance. Actually, team experience has been dropping slightly since 2001, when it peaked at 3.7 straight bids. It's tempting to draw a connection between the inexperience of the 2006 field and the upswing in upsets. The fact is, though, that there's little correlation between the two factors. The years between 1991 and 1998 had the fewest average number of upsets per tourney; they also saw the most consistently inexperienced tourney fields.
To gain a full sense of how significant the trend toward inexperienced teams is becoming, you only need to look at the round-by-round averages of the tourney field. In the 64-team era, the average number of consecutive tourney trips for surviving teams has steadily climbed with each succeeding round (see the dark red line below). While the average tourney entrant comes in with three straight bids, the field of first-round winners has four, teams reaching the Sweet 16 have 4.6, Elite Eight contenders rise to 5.4 and the Final Four field has 6.4. There's a slight dip to 6.3 straight trips for the two final combatants, but the ultimate champion averages 6.5 consecutive tickets to The Dance -- the highest of any round.
This trend was practically turned on its ear in 2006 (see the gold line). Last year's field came into the tourney with slightly more experience (3.2 straight bids), but the round-by-round survivors didn't climb as dramatically as historical trends. In fact, the Final Four were only slightly more experienced than the field. From that point on, however, experience won the day. Florida, by far the most tourney-tested semifinalist with eight consecutive trips, dominated upstart George Mason and a UCLA squad with just two straight bids.
Average number of tourney trips per coach
Last year, like their teams, coaches were less experienced than in previous tourneys, averaging 5.6 trips to The Dance. In fact, over the last nine seasons, only 2001 had coaches with fewer average trips to the tourney. While the correlation between team inexperience and upsets doesn't hold, there's a stronger relationship between the number of coaching tourney bids and upsets. Throughout the "upset-light" years of the tourney, coaching experience was at its highest. In many of the upset-heavy tourney years -- like 1985, 1986, 2001, 2002 and 2006 -- coaches had the least experience in the tournament.
Round-by-round trends for 2006 show a similar departure from historical norms as the team experience trend. While coaches in last year's field of 64 had the same average years of tourney experience (5.6) as all other coaches in the modern era, those that advanced were decidedly less experienced than the average advancing coaches. With each round, the gulf between historical averages and last year's numbers grew, peaking in the Final Four, where the 2006 semifinal coaches were more than four years less experienced than the overall average.
Average regular-season winning percentage for tourney field
Last year's tourney field had the best average winning percentage of any season since 1998 -- and the third best of the 22-year 64-team era. As we'll see in the "performance against seed expectations" analysis, though, winning percentage is a misleading factor: Just because a team has a strong regular-season record, that doesn't mean it will be successful in the tourney. By the same token, a strong record isn't necessarily a sign of tourney trouble either.
Average wins in last 10 games for tourney field
In only two years of the modern era did the field enter the tourney with fewer average wins in its last 10 games. Despite what college hoops pundits suggest, pretourney momentum doesn't necessarily translate into March Madness success. As attribute PASE statistics show, teams that have won more than seven of their last 10 games slightly underperform in the tourney (winning 708 games when their seed position suggests they should win 710). That said, in the seven highest upset years, the tourney field won fewer average games in their last 10 (7.20) than they did in the eight lowest upset years (7.32). So pretourney momentum at least has some correlation to upset prevalence.
If you look at the historical "wins in last 10 games" averages of round-by-round survivors, there is a case to be made that pretourney momentum confers some advantage. While the average tourney field of 64 has won 7.3 games, typical Elite Eight teams have won 8.1 -- and the champion has won 8.2. This trend didn't hold in 2006, however. From the Elite Eight on, each succeeding round of survivors was actually less successful in its last 10 games. In fact, Florida was "colder" than the average 2006 tourney entrant, winning just seven of its final 10, compared to 7.2 wins for the field. That might be one reason why so few people picked the Gators in their Final Four bracket.
Average winning streak entering tourney
The other sign of March Madness momentum -- winning streak entering The Dance -- has a more complicated relationship to tourney performance. Teams came into last year's tournament with the smallest average winning streak of the 64-team era -- and what happened? Upsets skyrocketed. In fact, for the last four years, upsets have risen as win streaks declined. Before we make too much of the trend, however, look at 2002. That year, teams entered the tourney with the modern era's longest average winning streak -- and fell victims to the highest number of upsets. Moreover, as the attribute PASE stats show, teams losing a game before coming to The Dance (usually in their conference tournaments) actually perform better than expectations, winning 710 games when their seed position dictated only 681 wins. Then again, last year's finalists, UCLA and Florida, both had more than four-game pretourney winning streaks. The lesson: Don't be swayed either way by the streaks -- winning or losing -- teams carry into The Dance.
Average points scored per game for tourney field
Scoring punch among tourney combatants has been steadily dropping about half a point per year since the high-water mark in 1989. That year, the average tourney team scored a whopping 82.7 points a game; in 2006, the average team scored just 73.0, the lowest total of the modern era.
Despite the downward trend in scoring punch, offensive output remains a key indicator of tourney success. From 1985 to 2006, the average points scored by teams in each succeeding round of the tourney has climbed almost linearly from 76.7 a game for the overall field to 82.2 for a semifinalist and 84.1 for the ultimate champion (see graph below). PASE statistics on attributes confirm the value of having a strong offense; teams that score more than three points above the tourney field average have won 634 games when seeding suggests they should've only notched 587 wins. The plus-0.117 PASE value is good for sixth best among the 56 attributes studied.
While scoring remains a key performance indicator, the results of the 2006 tourney signal a growing importance in the value of defense. Last year's round-by-round tourney survivors weren't nearly as high scoring as historical trends. In fact, last year's Final Four averaged just 73 points a game -- exactly the same as the overall tourney field. (The typical semifinalist is more than six points more prolific than the 64-team field.) Florida's dismantling of UCLA does show that a strong offense still trumps a stout defense in the tourney … but the unsettling fact is that scoring punch may no longer be the reliable predictive factor it was for 21 years.
Average points allowed per game for tourney field
Just as scoring offense has steadily declined since 1989, so too has the average points that a tourney contender allows per game. Last year's field gave up only 64.5 points per game, the lowest of any year in the modern era. In fact, the last three years have been among the four lowest of the last 22 years.
Despite the declining numbers in points allowed per game, scoring defense remains a poor indicator of tourney performance. Based on PASE, allowing fewer points per game than the tourney field average ranks 51st among the 56 attributes studied. Teams possessing this attribute won 694 games when their average seed position dictated about 728 wins.
Don't tell last year's tourney field that scoring defense didn't matter. In every round, the survivors were stingier than the historical average. And while the overall numbers show that points allowed per game slightly increase for round-by-round advancers, from 68.2 to 69.6, last year's surviving fields actually decreased from 64.5 to 62.0 before increasing for the champion to 64.8. Will stingy defensive teams have similar success in this year's tourney? If so, look for another upset-laden March; three of the five heaviest upset tourneys have coincided with years when scoring defense was low.
Average victory margin per game for tourney field
Throughout the 64-team era, the average margin of victory for the tourney field has held nearly constant, never dropping below 7.4 points or exceeding 9.2. The 8.5 margin for the 2006 field is exactly the average of all the teams in the modern era. Don't think for a second, though, that margin of victory isn't an important factor in the tourney. Of the 56 attributes analyzed for PASE, five of the top 10 involve scoring margin. Teams that beat their opponents by more than 15 points a game win 43 games more than the 245 games that seeding suggests they should win, for the highest PASE (plus-0.424) of all attributes.
Average per game margin of victory by percentage
Evaluating victory margin by sheer points can be a misleading method. After all, an 8.5-point margin in a low-scoring year like 2006 is a much larger relative margin than, say, an 8.8-point margin in the high-scoring 1989 tourney. By analyzing margin as a percentage increase between points allowed and points scored, we get a more accurate reflection of victory margin. In that light, the 2006 tourney field boasted one of the highest margin percentages in the 22-year history of the modern era. Only 2000's 14.0 percent and 2005's 13.3 percent gaps were larger.
While the margin of last year's round-by-round survivors nearly mirrored historical numbers through the first three rounds, 2006 Elite Eight combatants were 20.5 percent better than their opponents, as compared to 18.5 percent overall. The Final Four and championship game numbers dipped below the historical average -- but Florida's 22.4 percent margin was higher than that of the average champion. It was another telltale sign that, despite their No. 3 seed, the Gators had the statistical chops to take the tourney.
Average percentage of points from top five scorers
Given all the pundit chatter about the value of a deep team, it seems like strong bench play would be a key tourney performance indicator. Not true. Of the 56 attributes analyzed by the PASE method, getting at least 20 percent of scoring from the bench was the 31st-most significant factor, netting 671 actual wins against 659 expected victories. Part of the issue is that there isn't a huge gulf between the most starter-dominant and bench-reliant squads. The 1987 field got the highest percentage of points from the top five scorers at 81.1 percent, while the lowest percentage in 1999 was only 3.4 percent more bench-reliant at 77.7 percent.
Dependence on starters stays fairly constant for the round-by-round field of survivors, with less than a 1 percent difference between the highest and lowest percentages. That said, the 2006 advancers relied much more heavily on their starting units. Over the first four rounds, the percentage of starter scoring among bracket survivors surged from 79.4 percent to 85.0 percent. It did decline for the final two and the champion, but those drops just brought the numbers more in line with the historical average. The bottom line: Don't be swayed -- negatively or positively -- by a team's bench strength. In a short tournament with ample rest between games, depth just isn't that important.
Average percentage of points among top five scorers from backcourt
Given all the expert talk about the importance of strong guard play in the tourney, it should come as no surprise that the percentage of points a team gets from its backcourt vs. its frontcourt is a significant predictor of tourney success. What is surprising, though, is how much more important frontcourt scoring is than backcourt scoring. It's true that the percentage of points the tourney field gets from guards has risen over the last 20 years. In 1987, the average team got 42.2 percent of its points from the backcourt; last year the typical squad got nearly 10 percent more. In fact, only 2004 saw a higher reliance on guard scoring (by a scant tenth of a percent).
When it comes to advancing in the tourney, however, the more frontcourt-dominant teams tend to fare better. Based on PASE analysis, teams that get less than 40 percent of their points from guards (and therefore more than 60 percent from forwards and centers) won 27 more games than the "seed-expected" 393 wins. The averages of round-by-round tourney survivors tell the same story. Over the last 20 years, teams surviving into the later rounds of the tourney get more contributions from their big men -- with the champion fully 5 percent more reliant on frontcourt scoring. This "big man" trend wasn't quite as dramatic last year. But the Gators, like the Tar Heels and the Huskies before them, were clearly the most frontcourt-dominant team in the Final Four. And look what happened.
Average class of top five scorers (1 = freshman, 4 = senior)
To look at the trends in the average class of each team's top five players in the tourney field, you wouldn't think age had much bearing on March Madness performance. After all, the difference between the youngest (2.91) and oldest (3.09) tourney field, is negligible. Tourney teams have pretty consistently fielded an average starting unit of juniors. The starters for the 2006 tourney field have exactly the average class composition of all starting units that came before them (2.99).
The important trend for class composition is only apparent in the round-by-round numbers for tourney survivors. Over the last two decades, there's been a slight decline in the average age of advancing starters, from 2.99 for the 64-team field to 2.94 for the Final Four and 2.89 for the eventual champion. What's noteworthy about 2006 -- and really the last four seasons -- is that the trend toward succeedingly younger tourney survivors is more dramatic. Look at the yellow line below. The 2006 64-team field was comprised, on average, of junior starting units. The Sweet 16 was more like four juniors and a sophomore, the Elite Eight was three juniors and two sophomores, and the Final Four was two juniors and three sophomores. Youth was even served in the finals, with the eventual champion averaging almost all sophomore starters.
This year isn't an aberration. Since 2003, when a Syracuse squad, averaging four sophomores and a freshman, upset the veteran Kansas Jayhawks, younger teams have found more success in the tourney. Connecticut (2.20) was younger than Georgia Tech (3.00), North Carolina (2.80) was younger than Illinois (3.40), and both Florida and LSU were youngsters (2.20). What's it all mean? In the era of early leaps to the NBA, a senior-laden team isn't so much a sign of experience as an indictment of the team's lack of star talent.
Freelance writer Pete Tiernan has been studying the NCAA Tournament for 17 years. E-mail him here. Or go to www.bracketscience.com and check out the BracketMaster, which lets you do your own bracket analysis. |
