Tweet
By Pete Tiernan
ESPN Insider
When you sit down to fill out your bracket after Selection Sunday, you'll be challenged to predict the outcome of three types of games -- pushovers, mismatches and toss-ups.
Pushovers (1 vs. 16 and 2 vs. 15) are the easy games to pick. They effectively give everyone an 8-0 gift to kick-start their bracket. The high seeds are a near-perfect 172-4 in these matchups for a 98 percent success rate. Too bad only 13 percent of the tourney games in the 22-year, 64-team era (176 of 1,386) are this easy to predict.
The second and biggest class of games -- the mismatches -- comprises 54 percent of tourney contests since 1985 (752 of 1,386). These matchups pit teams with a seed difference of four or more positions. As we learned in "Anatomy of an Upset," the higher seed prevails against the long shot exactly 75 percent (564 of 752) of the time. That's a decent success rate, and you wouldn't wind up with an embarrassing bracket if you slotted the high seed to win all mismatches. Odds are, though, you wouldn't win your pool. That's why we gave you some telltale signs of Cinderellas in the upset article. Being able to identify the right upsets is a big part of building a better bracket. Just ask all those lucky souls who picked George Mason last year.
Still, sniffing out a handful of key upsets won't necessarily secure a tourney pool win. Why is that the case?
1. The most reliable traits of Cinderellas usually apply to just a handful of games each year.
2. Those games almost always occur in the first three rounds of the tourney. The fact is, the number of mismatch games drops off considerably from the Elite Eight on -- when accurate tourney prognostication is at its most critical for pool success. Only 40 of the 154 games (26 percent) over the last three rounds have been mismatches.
This is why the last class of games -- the toss-ups -- is the most important type of matchup to predict correctly in your bracket. Toss-up games pit teams within three seed positions against each other. About one-third of all tourney games in the 64-team era (458 of 1,386) involve toss-up pairings. The higher seed tends to prevail in these matchups, compiling a 245-199 record for a 55 percent success rate (14 games involved same-seeded teams). However, that percentage isn't nearly high enough to bank on. More importantly, higher-seeded toss-up teams struggle more with their lower-seeded foes in the later rounds of the tourney -- right when the percentage of toss-up games increases … as does the point value of correct predictions in most bracket pools.
The chart below shows how toss-up pairings become increasingly prevalent in the later rounds of the tourney -- and how the higher-seed advantage evaporates:
Toss-up games by round, 1985 to present
Round Total games Toss-up games % Toss-up games High seed wins High seed losses High seed win %
Round 1 704 176 25.0 % 94 82 53.4 %
Round 2 352 103 29.3 % 57 46 55.3 %
Sweet 16 176 65 36.9 % 41 24 63.1 %
Elite Eight 88 60 68.2 % 29 31 48.3 %
Final Four 44 37 84.1 % 15* 12* 55.6 %
Finals 22 17 77.3 % 9# 4# 69.2 %
Total 1386 458 33.0 % 245 199 55.2 %
* 10 Final Four games involved teams with the same seed position
# 4 championship games involved teams with the same seed position
Over the first three rounds of the tournament, the higher seed prevails in toss-up games about 56 percent of the time. After the Sweet 16, higher seeds win just 53 percent of the toss-up pairings. And that doesn't even take into consideration the fact that 14 games in the Final Four and championship rounds involve like-seeded opponents. You don't need to look any further than the 9-9 pick-'em record of last year's higher seeds to know that seeding isn't a reliable performance indicator in toss-up games.
Round 1: Resolving the 8 vs. 9 and 7 vs. 10 matchups
There are eight games in the first round that qualify as toss-ups -- the 8 vs. 9 and 7 vs. 10 pairings. In the 64-team era, the higher seed has won these games at a 53 percent clip (94-82). No. 7 seeds are reasonably solid bets against No. 10 seeds, notching a 53-35 (.602) record against their lower-seeded foes. No. 8 seeds, on the other hand, are the only higher seed in the first round to post a losing record against the "underdog" seed (41-47, .466).
To improve your odds of correctly picking 7 vs. 10 matchups, consider taking high scoring No. 10 seeds that beat their opponents by a solid margin. Tenth-seeded squads that score more than 75 points and win by more than seven points per game are 16-9 against No. 7 seeds. All other No. 10 seeds are 19-44. So if you took No. 10 seeds that meet the offensive output and victory margin conditions, and No. 7 seeds in all other situations, your record would be 60-28 (.682). That's 13 percent better than sticking with the higher seed. Had you followed these rules in 2006, you would've been 3-1. No. 10 seed North Carolina State met the scoring and margin criteria and the Wolfpack won. But No. 10 seed Alabama also won -- and the Tide didn't score 75 points a game.
In the 8 vs. 9 matchup, the No. 8 seeds that tend to succeed against their lower-seeded foes are tourney-tested, battle-scarred teams. Eighth-seeded squads that went to the previous year's tourney, but have won no more than three straight games heading into The Dance are 29-16. All other No. 8 seeds are a paltry 12-31. If you picked for the former and against the latter, you'd be 60-28. That's 46 percent better than a "higher seed" toss-up strategy, and 28 percent better than picking the No. 9 seed in this pairing. If you had used this strategy last year, you would've gone 4-0, since No. 8 seeds George Washington, Arizona and Kentucky all fulfilled the conditions and won, while eighth-seeded Arkansas didn't have the statistical chops and lost to No. 9 Bucknell.
Altogether, these two first-round toss-up strategies lead to a 120-56 record over the course of the 64-team era. That works out to a 68 percent success rate -- 28 percent and 26 games better than the 94-82 record that a higher-seed strategy would yield.
Round 2: The all-important 3 vs. 6 and 4 vs. 5 games
There are four pairings that qualify as toss-up games in the second round -- 3 vs. 6, 4 vs. 5, 11 vs. 14 and 12 vs. 13. Of the 103 second-round toss-ups that have been played in the 64-team era, all but eight have been 3 vs. 6 or 4 vs. 5 matchups. In the few 11 vs. 14 and 12 vs. 13 games, the higher seed holds a commanding 7-1 advantage.
That leaves 95 3 vs. 6 and 4 vs. 5 games in the 22 years of the 64-team era -- 4.3 per tourney. (Last year, there were only two of these matchups: No. 3 Gonzaga beat No. 6 Indiana, and No. 5 Washington got by No. 4 Illinois.) When you consider that 11 Final Four teams have fought through this matchup, six have made it to the final and one -- Syracuse in 2003 -- has won the championship, it's pretty clear that these are important "gateway" picks for building a solid bracket.
Unfortunately, the correlation between higher seeding and success isn't nearly as powerful with the 3 vs. 6 and 4 vs. 5 games as it is with the lower-seeded second-round toss-up games. No. 3 seeds nearly split their games with No. 6 seeds (25-24) and No. 4 seeds edge out No. 5 seeds, 25-21. Following a strict higher-seed strategy in Round 2, the best you would do is a 55 percent success rate (57-46). How can you improve upon that? There isn't much room for improvement with the two lower-seeded matchups. So the challenge is to find conditions that might improve upon the 50-45 record of higher seeds in the 3 vs. 6 and 4 vs. 5 pairings.
In the 3 vs. 6 matchup, the sixth-seeded squads most likely to topple No. 3 seeds are led by up-and-coming coaches who've been to the tourney less than six times and have yet to make an Elite Eight appearance. These No. 6 seeds are 11-3; their counterparts are 13-22. By picking No. 3 seeds in all cases except when the sixth-seeded team has an upstart coach, you would've achieved a 33-16 record. That's 32 percent better than the 25-24 record of a pure higher-seed strategy. It also would've accurately predicted the winner of last year's Gonzaga-Indiana matchup.
In the 4 vs. 5 pairing, the most successful No. 5 seeds have compiled solid records and soundly defeated their opponents. Fifth-seeded squads with at least a .700 winning percentage that have an average victory margin of more than eight points per game and fewer than three straight pre-tourney wins are 14-7. The rest of the No. 5 seeds are 7-18 against No. 4 seeds. By taking the No. 5 seeds that fulfill the winning, momentum and margin criteria and going with No. 4 seeds in all other situations, your record would be 32-14. That's 28 percent better than simply taking the fourth-seeded squads. Last year, these rules would've correctly identified Washington as a No. 5 seed with the right stuff to knock off No. 4 Illinois.
If you had employed all these strategies to second-round toss-up games, you would have achieved a 72-31 (.699) record. That's a 15-game swing over the 57-46 record of higher seeds -- and a 26 percent improvement.
Sweet 16: The 1 vs. 4 and 2 vs. 3 "gateway" games
Of the 176 Sweet 16 games in the 64-team era, 65 of them have been toss-up games -- and all but eight of these have involved 1 vs. 4 or 2 vs. 3 pairings. There have been six 6 vs. 7 matchups and two 5 vs. 8 matchups. The importance of correctly picking the 1 vs. 4 and 2 vs. 3 games can't be overstated. The 57 matchups have yielded 33 Final Four teams and four champions. Odds are if you don't do well with these games, you won't do well with your bracket.
In the 1 vs. 4 pairing, top seeds hold a commanding 22-10 advantage over fourth-seeded squads. When is it worth going out on a limb and picking a No. 4 seed? Look for underdogs that have won more than seven of their last 10 pre-tourney games and beat their opponents by an average of more than 10 points. These No. 4 seeds are 5-1; all others are 5-21. Picking fourth-seeded squads with these traits and top seeds in all other matchups would yield a 26-6 record. That's 18 percent better than the 22-10 record of top seeds. In 2006, this strategy would have identified No. 4 LSU as a Duke killer and stuck with top seed Villanova in its battle with Boston College.
In the 2 vs. 3 pairing, No. 2 seeds have posted a strong 16-9 record against No. 3 seeds. But considering that No. 3 seeds account for six of the 14 Final Four teams that survive this matchup -- and the only champion (Michigan in 1989) -- it would be a mistake to automatically advance second-seeded squads. The No. 3 seeds that fare best against their second-seeded opponents possess these three qualities:
1. They have coaches who have been to the tourney fewer than eight times.
2. They're coming into the tourney with no better than a two-game win streak.
3. They score 75 points and win by more than eight points a game.
Third-seeded squads with these characteristics are 6-2. The rest are 3-14. That means you could achieve a 20-5 record (25 percent better than 16-9) by taking No. 3 seeds that fulfill the requirements above and avoiding ones that don't. Last year, these rules would have correctly led you to take No. 2 UCLA in its matchup with No. 3 Gonzaga.
As for the lower-seeded 6 vs. 7 and 5 vs. 8 matchups, the telltale sign of an underdog victor is simple. No. 7 and No. 8 seeds that average less than 76 points per game offensively are 7-1 against their higher-seeded foes.
By applying all the rules for the 1 vs. 4, 2 vs. 3, 5 vs. 8 and 6 vs. 7 matchups, your forecasting accuracy for Sweet 16 matchups would be 82 percent (53-12). That's 29 percent and 12 games better than the 41-24 advantage of a higher-seed strategy.
Elite Eight: Revenge of the underdogs
The Elite Eight is the point in the tournament where the correlation between seeding and success evaporates. In these quarterfinal toss-up games, lower seeds actually own a 31-29 winning record over higher seeds. Of the 60 matchups, 42 have pitted top seeds against No. 2 or No. 3 seeds -- and the No. 1 seed has coin-flip reliability, posting a 21-21 record. The remaining 18 games have involved 2 vs. 4, 2 vs. 5, 3 vs. 4, 3 vs. 5, 4 vs. 6, 4 vs. 7, 6 vs. 8 and 7 vs. 8 pairings -- and the higher seed actually has a losing record (8-10).
In the 1 vs. 2 and 1 vs. 3 games, the lower seeds that win these matchups have been to the tourney at least four times in a row and defeated their opponents by at least 10 points per game. No. 2 and No. 3 seeds with both these attributes are 17-3. Those without them are 4-18. Taking the lower seeds in the former situation and the top seeds in the latter yields a 35-7 (83 percent) record.
In the other 18 games, higher seeds that get at least 50 percent of their points from guards are 5-1. The more backcourt-challenged high seeds are just 3-9. Picking against them and for the guard-oriented squads would give you a 14-4 record.
Taken together, the rules for both types of Elite Eight toss-up games would result in a strong 49-11 (82 percent) prediction record. That's 20 games -- and a whopping 69 percent -- better than a higher-seed strategy. If you had observed these rules last year, you would have correctly chosen No. 3 Florida over top-seeded Villanova and No. 4 LSU over No. 2 Texas, while incorrectly picking top-seeded Memphis over No. 2 UCLA.
Final Four: The "four factors" strategy
In the 37 (of 44) Final Four toss-up games, the higher seed holds a 15-12 advantage over the lower seed. Ten games have involved teams with identical seeds (nine 1 vs. 1 matchups and one 2 vs. 2 game). The best way to predict the outcome of these "same-seed" semifinal tilts is by picking teams in this order:
1. Take any "big six" conference team over a mid-major school.
2. Take any team that has a victory margin of more than 20 points (in the case where both teams have 20-plus margins, take the squad with the higher one).
3. Take the team with the longer winning streak.
4. Take the team with the better bench scoring.
If you had applied these prioritized rules to the 10 semifinal same-seed games, you would be a perfect 10-0 in your Final Four forecasting.
The "mismatch-seed" Final Four games break down into two categories: Top seeds vs. No. 2-4 seeds and games not involving No. 1 seeds (2 vs. 3, 2 vs. 4, 2 vs. 5, 3 vs. 4, 4 vs. 5, 4 vs. 6 and 5 vs. 8 games). In the matchups involving top seeds, stick with No. 1 seeds that have the following:
1. A preseason All-American.
2. A coach who's been to the tourney more than three times.
3. An average victory margin of more than 11 points.
4. An offensive output above 77 points per game.
Top seeds with these attributes are 8-3. Avoid all other No. 1 seeds; they're 0-5. Putting the two together, you'd go 13-3 in these matchups.
In the other 11 mismatch-seed, toss-up games, pick the higher seeds if they're from a big six conference, have a winning percentage lower than .900 and get more than 20 percent of their points from guards. These teams are 7-1. Combine this with the 3-0 record of lower seeds in the other matchups, and you'd achieve a 10-1 record in these games.
When you add it up, the three sets of rules above yield a 33-4 (89 percent) success rate, 13 games and 65 percent better than a higher-seed, flip-a-coin-on-same-seeds strategy. The rules also would have resulted in a correct pick of No. 2 UCLA over No. 4 LSU.
Championship: Coaching, victory margin and offense
Of the 22 championship games played in the 64-team era, 17 have involved toss-up seed pairings, and the higher seed is 9-4 -- with four same-seed matchups. If you followed a higher-seed strategy and picked the more offensive-minded team in the same-seed games, your prediction record would be 12-5 (75 percent).
That's not too bad, but you can do better. Here's how: Take every team whose coach has appeared in the tourney at least six times more than the other team's coach. In games where the coaching experience difference is less than six trips, take big six conference teams over mid-majors first and the team with the higher victory margin second. These rules yield a 15-2 record. The only two exceptions were Kentucky's 1996 victory over Syracuse, when Rick Pitino bested Jim Boeheim despite making 10 fewer trips to the tourney, and UConn's 1999 victory over Duke, which owned a 25.4-point average margin of victory to the Huskies' 16.7-point margin.
The 88 percent success rate is 25 percent better than the accuracy of the "higher-seed, higher-scoring" rule. What's more, it would've correctly identified Florida, the higher victory-margin team, as the winner over UCLA in last year's finals.
Bottom line: Toss-up rules yield a 75 percent success rate.
Employing a strict higher-seed strategy to predict the outcomes of toss-up games in the tourney only results in a 55 percent success rate (252-206, assuming a split in the 14 same-seed pairings). You could have dramatically improved your odds of picking toss-up winners by following the simple rules described above.
Generally speaking, scoring punch, victory margin and the lack of a long pre-tourney winning streak are the recurring traits of lower-seeded victors from round to round. If you remember nothing else in making toss-up picks, remember to look more favorably on lower seeds in toss-up games that score more than 76 points a game, win by more than eight points and have no more than a two-game win streak (or even a loss) entering the tourney.
That said, if you use every tip outlined above to make your toss-up picks -- and past performance dictates future results (always a big "if" during March Madness) -- you would boost your prediction accuracy to 75 percent (342-116), a 36 percent increase over an exclusive seed-based strategy. Your performance in last year's tourney would've been even more impressive. Instead of the 9-9 pick-'em results of going with the higher seed, you would have gone 16-2 (.889) -- a 77 percent improvement. Bear in mind, though: While these toss-up rules show a strong correlation to tourney advancement, they aren't necessarily the causes of success.
ESPN Insider
When you sit down to fill out your bracket after Selection Sunday, you'll be challenged to predict the outcome of three types of games -- pushovers, mismatches and toss-ups.
Pushovers (1 vs. 16 and 2 vs. 15) are the easy games to pick. They effectively give everyone an 8-0 gift to kick-start their bracket. The high seeds are a near-perfect 172-4 in these matchups for a 98 percent success rate. Too bad only 13 percent of the tourney games in the 22-year, 64-team era (176 of 1,386) are this easy to predict.
The second and biggest class of games -- the mismatches -- comprises 54 percent of tourney contests since 1985 (752 of 1,386). These matchups pit teams with a seed difference of four or more positions. As we learned in "Anatomy of an Upset," the higher seed prevails against the long shot exactly 75 percent (564 of 752) of the time. That's a decent success rate, and you wouldn't wind up with an embarrassing bracket if you slotted the high seed to win all mismatches. Odds are, though, you wouldn't win your pool. That's why we gave you some telltale signs of Cinderellas in the upset article. Being able to identify the right upsets is a big part of building a better bracket. Just ask all those lucky souls who picked George Mason last year.
Still, sniffing out a handful of key upsets won't necessarily secure a tourney pool win. Why is that the case?
1. The most reliable traits of Cinderellas usually apply to just a handful of games each year.
2. Those games almost always occur in the first three rounds of the tourney. The fact is, the number of mismatch games drops off considerably from the Elite Eight on -- when accurate tourney prognostication is at its most critical for pool success. Only 40 of the 154 games (26 percent) over the last three rounds have been mismatches.
This is why the last class of games -- the toss-ups -- is the most important type of matchup to predict correctly in your bracket. Toss-up games pit teams within three seed positions against each other. About one-third of all tourney games in the 64-team era (458 of 1,386) involve toss-up pairings. The higher seed tends to prevail in these matchups, compiling a 245-199 record for a 55 percent success rate (14 games involved same-seeded teams). However, that percentage isn't nearly high enough to bank on. More importantly, higher-seeded toss-up teams struggle more with their lower-seeded foes in the later rounds of the tourney -- right when the percentage of toss-up games increases … as does the point value of correct predictions in most bracket pools.
The chart below shows how toss-up pairings become increasingly prevalent in the later rounds of the tourney -- and how the higher-seed advantage evaporates:
Toss-up games by round, 1985 to present
Round Total games Toss-up games % Toss-up games High seed wins High seed losses High seed win %
Round 1 704 176 25.0 % 94 82 53.4 %
Round 2 352 103 29.3 % 57 46 55.3 %
Sweet 16 176 65 36.9 % 41 24 63.1 %
Elite Eight 88 60 68.2 % 29 31 48.3 %
Final Four 44 37 84.1 % 15* 12* 55.6 %
Finals 22 17 77.3 % 9# 4# 69.2 %
Total 1386 458 33.0 % 245 199 55.2 %
* 10 Final Four games involved teams with the same seed position
# 4 championship games involved teams with the same seed position
Over the first three rounds of the tournament, the higher seed prevails in toss-up games about 56 percent of the time. After the Sweet 16, higher seeds win just 53 percent of the toss-up pairings. And that doesn't even take into consideration the fact that 14 games in the Final Four and championship rounds involve like-seeded opponents. You don't need to look any further than the 9-9 pick-'em record of last year's higher seeds to know that seeding isn't a reliable performance indicator in toss-up games.
Round 1: Resolving the 8 vs. 9 and 7 vs. 10 matchups
There are eight games in the first round that qualify as toss-ups -- the 8 vs. 9 and 7 vs. 10 pairings. In the 64-team era, the higher seed has won these games at a 53 percent clip (94-82). No. 7 seeds are reasonably solid bets against No. 10 seeds, notching a 53-35 (.602) record against their lower-seeded foes. No. 8 seeds, on the other hand, are the only higher seed in the first round to post a losing record against the "underdog" seed (41-47, .466).
To improve your odds of correctly picking 7 vs. 10 matchups, consider taking high scoring No. 10 seeds that beat their opponents by a solid margin. Tenth-seeded squads that score more than 75 points and win by more than seven points per game are 16-9 against No. 7 seeds. All other No. 10 seeds are 19-44. So if you took No. 10 seeds that meet the offensive output and victory margin conditions, and No. 7 seeds in all other situations, your record would be 60-28 (.682). That's 13 percent better than sticking with the higher seed. Had you followed these rules in 2006, you would've been 3-1. No. 10 seed North Carolina State met the scoring and margin criteria and the Wolfpack won. But No. 10 seed Alabama also won -- and the Tide didn't score 75 points a game.
In the 8 vs. 9 matchup, the No. 8 seeds that tend to succeed against their lower-seeded foes are tourney-tested, battle-scarred teams. Eighth-seeded squads that went to the previous year's tourney, but have won no more than three straight games heading into The Dance are 29-16. All other No. 8 seeds are a paltry 12-31. If you picked for the former and against the latter, you'd be 60-28. That's 46 percent better than a "higher seed" toss-up strategy, and 28 percent better than picking the No. 9 seed in this pairing. If you had used this strategy last year, you would've gone 4-0, since No. 8 seeds George Washington, Arizona and Kentucky all fulfilled the conditions and won, while eighth-seeded Arkansas didn't have the statistical chops and lost to No. 9 Bucknell.
Altogether, these two first-round toss-up strategies lead to a 120-56 record over the course of the 64-team era. That works out to a 68 percent success rate -- 28 percent and 26 games better than the 94-82 record that a higher-seed strategy would yield.
Round 2: The all-important 3 vs. 6 and 4 vs. 5 games
There are four pairings that qualify as toss-up games in the second round -- 3 vs. 6, 4 vs. 5, 11 vs. 14 and 12 vs. 13. Of the 103 second-round toss-ups that have been played in the 64-team era, all but eight have been 3 vs. 6 or 4 vs. 5 matchups. In the few 11 vs. 14 and 12 vs. 13 games, the higher seed holds a commanding 7-1 advantage.
That leaves 95 3 vs. 6 and 4 vs. 5 games in the 22 years of the 64-team era -- 4.3 per tourney. (Last year, there were only two of these matchups: No. 3 Gonzaga beat No. 6 Indiana, and No. 5 Washington got by No. 4 Illinois.) When you consider that 11 Final Four teams have fought through this matchup, six have made it to the final and one -- Syracuse in 2003 -- has won the championship, it's pretty clear that these are important "gateway" picks for building a solid bracket.
Unfortunately, the correlation between higher seeding and success isn't nearly as powerful with the 3 vs. 6 and 4 vs. 5 games as it is with the lower-seeded second-round toss-up games. No. 3 seeds nearly split their games with No. 6 seeds (25-24) and No. 4 seeds edge out No. 5 seeds, 25-21. Following a strict higher-seed strategy in Round 2, the best you would do is a 55 percent success rate (57-46). How can you improve upon that? There isn't much room for improvement with the two lower-seeded matchups. So the challenge is to find conditions that might improve upon the 50-45 record of higher seeds in the 3 vs. 6 and 4 vs. 5 pairings.
In the 3 vs. 6 matchup, the sixth-seeded squads most likely to topple No. 3 seeds are led by up-and-coming coaches who've been to the tourney less than six times and have yet to make an Elite Eight appearance. These No. 6 seeds are 11-3; their counterparts are 13-22. By picking No. 3 seeds in all cases except when the sixth-seeded team has an upstart coach, you would've achieved a 33-16 record. That's 32 percent better than the 25-24 record of a pure higher-seed strategy. It also would've accurately predicted the winner of last year's Gonzaga-Indiana matchup.
In the 4 vs. 5 pairing, the most successful No. 5 seeds have compiled solid records and soundly defeated their opponents. Fifth-seeded squads with at least a .700 winning percentage that have an average victory margin of more than eight points per game and fewer than three straight pre-tourney wins are 14-7. The rest of the No. 5 seeds are 7-18 against No. 4 seeds. By taking the No. 5 seeds that fulfill the winning, momentum and margin criteria and going with No. 4 seeds in all other situations, your record would be 32-14. That's 28 percent better than simply taking the fourth-seeded squads. Last year, these rules would've correctly identified Washington as a No. 5 seed with the right stuff to knock off No. 4 Illinois.
If you had employed all these strategies to second-round toss-up games, you would have achieved a 72-31 (.699) record. That's a 15-game swing over the 57-46 record of higher seeds -- and a 26 percent improvement.
Sweet 16: The 1 vs. 4 and 2 vs. 3 "gateway" games
Of the 176 Sweet 16 games in the 64-team era, 65 of them have been toss-up games -- and all but eight of these have involved 1 vs. 4 or 2 vs. 3 pairings. There have been six 6 vs. 7 matchups and two 5 vs. 8 matchups. The importance of correctly picking the 1 vs. 4 and 2 vs. 3 games can't be overstated. The 57 matchups have yielded 33 Final Four teams and four champions. Odds are if you don't do well with these games, you won't do well with your bracket.
In the 1 vs. 4 pairing, top seeds hold a commanding 22-10 advantage over fourth-seeded squads. When is it worth going out on a limb and picking a No. 4 seed? Look for underdogs that have won more than seven of their last 10 pre-tourney games and beat their opponents by an average of more than 10 points. These No. 4 seeds are 5-1; all others are 5-21. Picking fourth-seeded squads with these traits and top seeds in all other matchups would yield a 26-6 record. That's 18 percent better than the 22-10 record of top seeds. In 2006, this strategy would have identified No. 4 LSU as a Duke killer and stuck with top seed Villanova in its battle with Boston College.
In the 2 vs. 3 pairing, No. 2 seeds have posted a strong 16-9 record against No. 3 seeds. But considering that No. 3 seeds account for six of the 14 Final Four teams that survive this matchup -- and the only champion (Michigan in 1989) -- it would be a mistake to automatically advance second-seeded squads. The No. 3 seeds that fare best against their second-seeded opponents possess these three qualities:
1. They have coaches who have been to the tourney fewer than eight times.
2. They're coming into the tourney with no better than a two-game win streak.
3. They score 75 points and win by more than eight points a game.
Third-seeded squads with these characteristics are 6-2. The rest are 3-14. That means you could achieve a 20-5 record (25 percent better than 16-9) by taking No. 3 seeds that fulfill the requirements above and avoiding ones that don't. Last year, these rules would have correctly led you to take No. 2 UCLA in its matchup with No. 3 Gonzaga.
As for the lower-seeded 6 vs. 7 and 5 vs. 8 matchups, the telltale sign of an underdog victor is simple. No. 7 and No. 8 seeds that average less than 76 points per game offensively are 7-1 against their higher-seeded foes.
By applying all the rules for the 1 vs. 4, 2 vs. 3, 5 vs. 8 and 6 vs. 7 matchups, your forecasting accuracy for Sweet 16 matchups would be 82 percent (53-12). That's 29 percent and 12 games better than the 41-24 advantage of a higher-seed strategy.
Elite Eight: Revenge of the underdogs
The Elite Eight is the point in the tournament where the correlation between seeding and success evaporates. In these quarterfinal toss-up games, lower seeds actually own a 31-29 winning record over higher seeds. Of the 60 matchups, 42 have pitted top seeds against No. 2 or No. 3 seeds -- and the No. 1 seed has coin-flip reliability, posting a 21-21 record. The remaining 18 games have involved 2 vs. 4, 2 vs. 5, 3 vs. 4, 3 vs. 5, 4 vs. 6, 4 vs. 7, 6 vs. 8 and 7 vs. 8 pairings -- and the higher seed actually has a losing record (8-10).
In the 1 vs. 2 and 1 vs. 3 games, the lower seeds that win these matchups have been to the tourney at least four times in a row and defeated their opponents by at least 10 points per game. No. 2 and No. 3 seeds with both these attributes are 17-3. Those without them are 4-18. Taking the lower seeds in the former situation and the top seeds in the latter yields a 35-7 (83 percent) record.
In the other 18 games, higher seeds that get at least 50 percent of their points from guards are 5-1. The more backcourt-challenged high seeds are just 3-9. Picking against them and for the guard-oriented squads would give you a 14-4 record.
Taken together, the rules for both types of Elite Eight toss-up games would result in a strong 49-11 (82 percent) prediction record. That's 20 games -- and a whopping 69 percent -- better than a higher-seed strategy. If you had observed these rules last year, you would have correctly chosen No. 3 Florida over top-seeded Villanova and No. 4 LSU over No. 2 Texas, while incorrectly picking top-seeded Memphis over No. 2 UCLA.
Final Four: The "four factors" strategy
In the 37 (of 44) Final Four toss-up games, the higher seed holds a 15-12 advantage over the lower seed. Ten games have involved teams with identical seeds (nine 1 vs. 1 matchups and one 2 vs. 2 game). The best way to predict the outcome of these "same-seed" semifinal tilts is by picking teams in this order:
1. Take any "big six" conference team over a mid-major school.
2. Take any team that has a victory margin of more than 20 points (in the case where both teams have 20-plus margins, take the squad with the higher one).
3. Take the team with the longer winning streak.
4. Take the team with the better bench scoring.
If you had applied these prioritized rules to the 10 semifinal same-seed games, you would be a perfect 10-0 in your Final Four forecasting.
The "mismatch-seed" Final Four games break down into two categories: Top seeds vs. No. 2-4 seeds and games not involving No. 1 seeds (2 vs. 3, 2 vs. 4, 2 vs. 5, 3 vs. 4, 4 vs. 5, 4 vs. 6 and 5 vs. 8 games). In the matchups involving top seeds, stick with No. 1 seeds that have the following:
1. A preseason All-American.
2. A coach who's been to the tourney more than three times.
3. An average victory margin of more than 11 points.
4. An offensive output above 77 points per game.
Top seeds with these attributes are 8-3. Avoid all other No. 1 seeds; they're 0-5. Putting the two together, you'd go 13-3 in these matchups.
In the other 11 mismatch-seed, toss-up games, pick the higher seeds if they're from a big six conference, have a winning percentage lower than .900 and get more than 20 percent of their points from guards. These teams are 7-1. Combine this with the 3-0 record of lower seeds in the other matchups, and you'd achieve a 10-1 record in these games.
When you add it up, the three sets of rules above yield a 33-4 (89 percent) success rate, 13 games and 65 percent better than a higher-seed, flip-a-coin-on-same-seeds strategy. The rules also would have resulted in a correct pick of No. 2 UCLA over No. 4 LSU.
Championship: Coaching, victory margin and offense
Of the 22 championship games played in the 64-team era, 17 have involved toss-up seed pairings, and the higher seed is 9-4 -- with four same-seed matchups. If you followed a higher-seed strategy and picked the more offensive-minded team in the same-seed games, your prediction record would be 12-5 (75 percent).
That's not too bad, but you can do better. Here's how: Take every team whose coach has appeared in the tourney at least six times more than the other team's coach. In games where the coaching experience difference is less than six trips, take big six conference teams over mid-majors first and the team with the higher victory margin second. These rules yield a 15-2 record. The only two exceptions were Kentucky's 1996 victory over Syracuse, when Rick Pitino bested Jim Boeheim despite making 10 fewer trips to the tourney, and UConn's 1999 victory over Duke, which owned a 25.4-point average margin of victory to the Huskies' 16.7-point margin.
The 88 percent success rate is 25 percent better than the accuracy of the "higher-seed, higher-scoring" rule. What's more, it would've correctly identified Florida, the higher victory-margin team, as the winner over UCLA in last year's finals.
Bottom line: Toss-up rules yield a 75 percent success rate.
Employing a strict higher-seed strategy to predict the outcomes of toss-up games in the tourney only results in a 55 percent success rate (252-206, assuming a split in the 14 same-seed pairings). You could have dramatically improved your odds of picking toss-up winners by following the simple rules described above.
Generally speaking, scoring punch, victory margin and the lack of a long pre-tourney winning streak are the recurring traits of lower-seeded victors from round to round. If you remember nothing else in making toss-up picks, remember to look more favorably on lower seeds in toss-up games that score more than 76 points a game, win by more than eight points and have no more than a two-game win streak (or even a loss) entering the tourney.
That said, if you use every tip outlined above to make your toss-up picks -- and past performance dictates future results (always a big "if" during March Madness) -- you would boost your prediction accuracy to 75 percent (342-116), a 36 percent increase over an exclusive seed-based strategy. Your performance in last year's tourney would've been even more impressive. Instead of the 9-9 pick-'em results of going with the higher seed, you would have gone 16-2 (.889) -- a 77 percent improvement. Bear in mind, though: While these toss-up rules show a strong correlation to tourney advancement, they aren't necessarily the causes of success.
Comment