The Pi-Rate Ratings

January 9, 2019

Advanced Basketball Statistics–Fun Stuff for Stats Buffs

This feature today is not for everybody. You have to be a stats fan for this one to be fun to read. Last year, we were asked to explain some of the advanced basketball metrics used today. Then, a couple weeks ago, we were asked again what certain metrics were. So, this will be an attempt to explain the basic advanced metrics and then to some degree how one might use this data to determine an approximate point spread difference.

If you are not familiar with advanced metrics in sports, it all started with baseball many decades ago. The legendary general manager of the then Brooklyn Dodgers, Branch Rickey, was always many years ahead of his contemporaries. He had basically created the farm system for Minor League baseball in the late 1920’s, and he opened the game to African Americans and then created a pipeline in Latin America for the Dodgers to take advantage there. Around the same time, Rickey was looking for a statistical advantage to evaluating baseball players, using mathematics to find hidden gems of talent that might have been somewhat overlooked by the competition. This was 50 years before Money Ball. Rickey aligned with a mathematics genius by the name of Alan Roth, who had previously tried to show some of his ideas to other baseball owners, but none of these owners had an interest. Rickey was more than interested, and he hired Roth to work for the Dodgers about the same time as Jackie Robinson debuted in Brooklyn.

Roth was one of the first baseball statisticians to realize that RBIs were basically worthless as a stat. For two decades, he worked for the Dodgers charting where players hit balls, what pitches they hit, and who fielded or did not field balls. This technology would not come into the norm for another 40 years.

Many others presented statistical data for baseball through the 1960’s, 1970’s, and 1980’s. Some of these math experts wrote books, such as Earnshaw Cook and his great work called, Percentage Baseball. It is this book that I read many years ago that roped me into the world of advanced baseball statistics.

A lot of you reading this know who Billy Beane is and what “Money Ball” is. Let me clue you in on something. This was not the first big leap into computer-generated advanced baseball statistics. It wasn’t even the first attempt by the Oakland Athletics. Beane’s predecessor, Sandy Alderson, brought the computer big time into baseball, but you could argue that Earl Weaver with the Baltimore Orioles had a basic no frills database of his batters’ and pitchers’ successes and failures against the pitchers and hitters throughout the American League.

About the time that Money Ball had come out in book form, other mathematics experts began looking at different sports. Computer specialists had come up with somewhat successful algorithms to pick winners against the point spread in football, and one or two became quite wealthy until the state of Nevada banned their wagering for life.

In the late 1990’s, the NBA began looking for ways to take advantage of the numbers to maximize talent. Was it worth it to shoot the 3-pointer? Was it better to have a strong rebounding team that maybe didn’t shoot as well than a weaker rebounding team that shot better? What was the best number of minutes to play your star players, and the best number of minutes to play your second team players? Could statistics show enough consistency to partially answer these questions?

Of course, the questions can never fully be answered. Until computers can read the minds of humans, they can never determine if Stephen Curry may have strained his right shoulder lifting his amazing daughter up in the air earlier that day. The computer cannot determine if the star player had a little too much pizza the night before and didn’t get a good night sleep. There is missing data that will be discovered in the future because basketball analytics are far behind baseball in the evolutionary process.
Basketball is starting to catch up now that very expensive software exists in NBA gyms where multiple cameras are placed in the rafters of the arenas, which feed into a computer and can show teams where all 10 players were on the floor for each 1/100 of a second of the game. If the power forward was beaten for an easy jumper when the shooter came off a baseline screen, the computer records this. Within a few years, the game will become every bit as scientific as baseball, and you will see more Cal Tech and MIT grads working in front offices.
By now, you must realize that trying to explain all the advanced basketball metrics would be terribly boring and very difficult to do. I admit that I am not the authority on basketball metrics, but then I get paid for baseball analytics and not basketball analytics.

Here is a brief look at some of the advanced stats for basketball. If you are interested, you should be able to set up these formulas on a spreadsheet and then plug your team’s stats in and have some of the more popular advanced stats for the team you follow. You can even use these stats for lower levels (high school, middle school, youth league), but the formulas must be altered by an amount I cannot give you. There is a difference between NBA and college formulas, and there will be differences as you go down in experience. Some of it has to do with how many fouls it takes to put a team in the bonus and what that bonus is.
Let’s Begin
I must start with the most basic of advanced statistics. This first set of stats will give you a lot more than the basic statistics. They are called, “The Four Factors,” but they are used for both a team’s offense and a team’s defense, so it is really eight factors.

Credit here must be given to the very brilliant Cal Tech statistician Dean Oliver who wrote the number one book on basketball statistics, Basketball on Paper. It is required reading if this is your field of interest. Oliver capitalized on his data and sold it and himself to a handful of NBA teams, but the basketball media wasn’t ready for his ideas.

They scrutinized every move made through his recommendations, forcing the NBA teams to give in to their fans that bought into the media’s opinions. Of course, many of these media hacks cannot balance their own checkbooks, so their scrutiny comes without credibility.  I say this because I was once a media hack in a top 30 market who believed a lot of the preconceived misconceptions of sports.
The “Four Factors” (again eight factors since this is figured for the offense and the defense) are:

1. Effective Field Goal Percentage
2. Turnover Rate
3. Offensive Rebounding Rate
4. Free Throw Rate.

While these factors are still quite valid, they have been surpassed somewhat by more advanced data. For example, True Shooting Percentage is more detailed than EFg%.

Here are the easy calculations for the Four Factors.
1. Effective Field Goal Percentage
This stat adds three point shooting to two point shooting into one stat. A made three-pointer is worth 50% more than a made two-pointer. So, if you make 1/3 of your three-pointers, it is the same as making 50% of your two-pointers.

The formula for eFG% is: (Field Goals Made + (0.5* 3-pointers Made))/Field Goals Attempted.

Let’s say that Duke takes 58 total shots in a game. They make 26 of these shots, and 8 of them are three-pointers. The calculation would be:

(26 +(0.5*8))/58 which equals .517 or 51.7%.

It works the same for defense. Let’s say in the same game, Duke’s opponent took 57 shots and made 24 with 7 of them three-pointers.

(24+(0.5*7))/57 = .482 or 48.2%.

1A. True Shooting Percentage combines Effective Field Goal Percentage with foul shooting into one combined scoring stat. As you will see with the 4th factor, there is debate over how to use FT Rate properly.

The NBA formula for True Shooting Percentage is: Pts/(2*(FGA+(.44*FTA))) but this is the NBA formula. As I mentioned above, the formula for college basketball is a little different, and it has to do with different Free Throw rules in the two organizations.

For college, it is: Pts/(2*(FGA+(.465*FTA)))

Let’s look at this for an individual. Here are Steph Curry’s Shooting Stats for his last year at Davidson.

Curry scored 974 points in 2008-09. He took 687 shots from the field and 251 foul shots.

974/(2*(687+(.465*251)))= .606 or 60.6% which for a guard is outstanding.

Compare this to Kareem Abdul-Jabbar’s sophomore season at UCLA in 1966-67, when the NCAA made the mistake of banning the dunk following his dominant first year on the varsity.  Jabbar, known then by his birth name of Lew Alcindor, scored 870 points that year with 519 shots from the field and 274 foul shots.

870/(2*(519+(.465*274)))=.673 or 67.3%.

You can see that a dominant post player like Jabbar was worth more in shooting than a top outside shooter like Curry. This is a relative statement, but it is like saying Babe Ruth was worth more as a hitter than Ty Cobb.

2. Turnover Rate
This measures the rate at which a team commits a turnover or forces the opponent to commit a turnover. We will stick with team stats for now, because the formulas for individuals are a bit more complex.

The calculation for Tunover Rate is:

TO / (FGA + (0.44 * FTA) + TO) for NBA, and
TO/(FGA+(.465*FTA+TO) for College

We will calculate a couple of extremes here. Let’s look at Temple in 1987-88 and Arkansas in 1993-94. Temple’s Coach John Chaney guided the 1987-88 Owls to the regular season number one ranking using an aggressive 2-3 matchup zone defense and a patient offense that valued every offensive possession like gold. Temple did not gamble on offense or defense, as they never attempted to force their offense or try to create turnovers with defensive pressure, preferring to force opponents to shoot poor shots.

Arkansas coach Nolan Richardson guided the Razorbacks to the national title in 1993-94. His teams pressed full court for 40 minutes (40 minutes of Hell) and played up-tempo fast-breaking offense. Arkansas committed more turnovers on offense, but they forced a lot more turnovers than average, and they came up with a lot of steals that led to easy points.

Temple in 1987-88 in 34 games
Offense: 305 Turnovers, 2,050 FGA, 704 FTA
Defense: 423 Turnovers, 1981 FGA, 513 FTA

Offensive TO Rate: 305/(2,050+(.465*704)+305) = .114 or 11.4%
Defensive TO Rate: 423/(1981+(.465*513)+423) = .160 or 16.0%

Arkansas in 1993-94 in 34 games
Offense: 539 Turnovers, 2,363 FGA, 834 FTA
Defense: 725 Turnovers, 2,234 FGA, 817 FTA

Offensive TO Rate: 539/(2,363+(.465*834)+539) = .164 or 16.4%
Defensive TO Rate: 725/(2,234+(.465*817)+725) = .217 or 21.7%

Which team was better at total turnover differential, Temple in 1988 or Arkansas in 1994? It was basically a wash. Temple played conservative basketball about as good as it could be played, going 32-2 and outscoring opponents by 15+ points per game. Arkansas played havoc basketball and went 31-3 outscoring opponents by almost 18 points per game. Both styles worked.

3. Offensive Rebound Rate (and, of course, Defensive Rebound Rate)
This measures the rate a team gets offensive rebounds and the rate in which it limits its opponents from getting offensive rebounds, which is obviously the rate of getting defensive rebounds. These stats allow the statistician to quickly see the opposite without having to perform double calculation. If Michigan State gets 36% of the rebounds on their offensive side of the floor, then Michigan State’s opponents will obviously get 64% of the rebounds on their defensive end of the floor.

The calculation for Offensive Rebound Rate is: Off. Reb/(Off. Reb + opponents Def. Reb),

 and thus the Defensive Rebound Rate is: Def. Reb/(Def. Reb + opponents Off. Reb)

Coach Tom Izzo has his Michigan State Spartans totally dominating the glass this year. Their rebounding margin of 11 boards per game is giving the Spartans an incredible advantage in games (how much we will see later).

Let’s calculate their Offensive Rebound Rate so far this season:
Offensive Rebounds = 190 Defensive Rebounds = 337
Opponents Offensive Rebounds = 176 Defensive Rebounds = 513

Michigan State’s Off. Rebound Rate = 190/(190+337) = .361 or 36.1%
Michigan State’s Def. Rebound Rate = 513/(513+176) = .748 or 74.8%

You can also figure total Rebound Rate, which isn’t a Four Factor, but easy enough by taking Michigan State’s percentage of total rebounds. (190+513)/(190+513+337+176) = 57.8%

4. Free Throw Rate
This is the most controversial of the Four Factors, and there are now multiple theories about how best to calculate this stat. The original formula was simply FTA/FGA. Many metric specialists (including me) believe this is not the best way to calculate free throw rate. For one, this original formula does not calculate made free throws. Shaquille O’Neal would be just as effective and maybe more effective than Steph Curry, and there is no way you can convince me that Shaq’s free throw rate should be as strong or stronger than Curry’s.
There is another school of thought, which is the one the PiRate Ratings have adopted, and that is Free Throws Made per 100 possessions. The calculation is a bit more involved since you need the number of possessions, but total possessions is now kept as a stat in college basketball, and there is a formula that accurately approximates possessions.

Our Accepted FT Rate Calculation is: FT Made per 100 possessions.

If you do not have the number of possessions, you calculate it this way:

NBA: FGA+ (.44 * FTA) – Off. Rebounds + Turnovers
College: FGA +(.465 * FTA – Off. Rebounds + Turnovers

An example from a real game–last Sunday’s Michigan vs. Indiana game.
Michigan took 58 shots in the game. They had 16 Free Throw Attemps, 7 offensive rebounds, and an amazing 2 turnovers.

Let’s calculate their possessions: 58 + (.465*16) -7 + 2 = 60.44

In the actual game box score, Michigan had 60 possessions. In other words, this formula is very accurate, and when there is a difference of one possession in the calculation, it usually is because the team that controlled the opening tap also had the last possession of the half.
Michigan made 12 free throws in their 60 possessions, so we now have to normalize this to how many they would have made in 100 possessions, which is quite simple.

12/60*100 = 20.0, so Michigan’s Free Throw Rate in this game was 20.0.

If we use the original formula, Michigan had 16 FTA and 58 FGA for a rate of 16/58 or 27.6%. We feel that this overstates Michigan’s rate here. Because there were just 60 possessions in this game (about as low as a 30-second shot clock game can produce), the rate was inflated.

There is a third school of thought by stating this formula as FT Made / FG Attempted, which is a bit more accurate than FTA/FGA, but we still prefer making our rate per 100 possessions.

Putting it all together
So, now you have the four factors. How can we take this data before a game is played and determine an estimated point spread? It is not an exact science.

Let’s return briefly to baseball. In baseball, you have the infamous WAR stat, where players are rated in wins above a replacement player, a replacement player being somebody you can pick up on waivers or call up from AAA. There is no WAR stat at this time for basketball, although many statisticians have tried to calculate one from game stats. The problem is that it is hard to judge defense in basketball compared to judging pitching and fielding in baseball.

So, the answer is to find a way to determine how much weight to place on each of the Four (Eight) Factors to try to determine which team is better.

In the NBA, this calculation is considerably easier than in college, because strength of schedule only marginally differs in pro basketball, as most teams play an equal schedule strength. It can be argued that Golden State’s schedule is easier than Philadelphia’s schedule, because the Warrior won’t play Golden State, while the 76ers don’t benefit from playing Philadelphia, but that becomes negligible as the season progresses.

In college basketball, the Patriot League and the Big Ten are not close to comparable, so Lehigh’s Four Factors’ stats are not equal with Michigan’s Four Factors’ stats.
Originally, Oliver determined that Effective Field Goal Percentage was by far the most important of the Four Factors, and since there are a lot more shots taken in a basketball game than anything else, it goes without saying that this factor should be the most important. If your team can consistently beat its opponents in eFG%, they will win more games than they lose. If your team has an eFG% that is 10% better than the opponents, then your team is playing at a championship level.

Oliver believed that eFG% was about 40% of the success or failure of a team. He stated that turnover rate was worth 25%, offensive rebound rate was worth 20%, and FT Rate was worth the remaining 15%. In back-testing, these numbers approximated success or failure in the NBA.

It took many hours of algorithm testing for the PiRate Ratings to come up with percentages to apply to these factors. In the end, we had to create two more factors to approach legitimate accuracy.
If you have followed this site during basketball season for some time, you have probably heard about our own creation called “R+T Factor.” This is a refined version of the rebounding rate and turnover rate, which probably is the reason why Oliver gave a bit more weight to turnovers than rebounds. The key is to separate turnovers into steals and everything else. A steal in basketball is worth more than a rebound. When a team steals the ball, the chances of getting an easy basket and/or drawing a foul is much higher than obtaining a rebound. After working with the formula for a few years, we finally came up with one we like.

Our R+T rating is: (R*2) + (S*.5) + (6-Opp S) + T, where
R= Rebound Margin
S= Average Steals Per Game
T= Turnover Margin

In 2017, one NCAA Team had a rebound margin of 12.3 per game.  They had a turnover margin of 1.8 per game (which means that they committed 1.8 fewer turnovers per game than their opponents), averaged 7.1 steals per game, and opponents averaged 6.2 steals per game.

This team’s R+T Rating was: (12.3*2) + (7.1*0.5) + (6-6.2) + 1.8 = 17.5

This team played in one of the top power conferences in the NCAA, and their rating of 17.5 was the best among the power conference teams.  When a power conference team has an R+T rating over 10, they are Sweet 16 caliber.  At 15, they are Final Four caliber.  So, it can be deduced that this team did fairly well in the 2017 tournament.

This team was national champion North Carolina.

This R+T stat tries to estimate the number of extra scoring opportunities a team gets in a game. The stat is much more valuable in the NCAA Tournament where there are 25-30 really strong teams playing. When the pressure is on, many times these extra opportunities decide the outcomes. While effective field goal percentage is still the number one variable, the R+T rating becomes more and more valuable as the tournament progresses. By the Sweet 16, the teams with the best R+T rating usually continue to advance, and in many years, the team with the number one R+T rating weighted by schedule strength wins the National Championship. In every season in the 21st Century, the champion has been among the nation’s leaders in R+T factor weighted against schedule strength.
The obvious second added factor in predicting basketball games is schedule strength. If a team in the Ivy League outscores its opposition by 10 points per game, they are not as good as a team from the ACC outscoring opponents by 10 points per game.

At the start of conference play, one SEC team may have played a non-conference schedule that on average is 10 points weaker per game than another team. Kentucky usually plays a much harder pre-conference schedule than Vanderbilt or Ole Miss. Tennessee has played a more difficult schedule than Missouri.

Once conferences have played more than half of their league schedules, you can even calculate ratings based only on conference games played and then take those ratings and rank the conferences overall to get a more accurate rating for every team.

For example, let’s say that on February 20 with 80% of the Big 12 conference games in the books, Texas Tech is 1 point better than Kansas, 3 points better than Iowa State, and so on down to Oklahoma State being 14 points weaker than Texas Tech. Let’s say that Stephen F. Austin is 3 points better than Abilene Christian in the Southland Conference and 5 points better than Sam Houston. Overall, the Big 12 is calculated to be 17 points better on average than the Southland conference, so Texas Tech would be 17 points better than SF Austin, 20 points better than Abilene Christian, and 22 points better than Sam Houston. Oklahoma State would then be 8 points better than Sam Houston, since they are 14 points weaker than Texas Tech.
We don’t actually figure the ratings this way, but we have an algorithm that does a similar calculation for every team based on their overall strength of schedule for the season. It is a close cousin but goes more in-depth than the Quadrant system in place by the NCAA Selection Committee and used by our Bracketology experts when they pick their weekly selections, which by the way you can now see our PiRate Rating Bracketology at the Bracket Matrix, at Our abbreviation there is “Pi.”

There are many additional advanced analytical basketball ratings. Also, you can break down the individual ratings for all of the Four Factors, as well as ratings that calculate individual offensive and defensive efficiency and the Usage Rate, which tries to estimate how much a player is used in his team’s games by looking at what he does while he is in the game. Some teams most efficient players may not have the top usage rates on their teams, while less efficient players get more game usage. Teams can look at these stats and good coaches can adjust their lineups to get their more efficient players more game time, while limiting players that may be harming the team. Then, there are coaches that continue to play the wrong players for too many minutes, while their actual more efficient players don’t play enough. There is a phrase for these coaches that continually do this: We call it “Soon to be unemployed.”

March 24, 2010

Sweet 16 Preview


From Sweet to Elite

Advanced Level Bracketnomics


Hello PiRate Basketball fans.  Our system worked well, but the idiots (us) in charge of the data didn’t have the guts to play all the upsets.  We still have nine teams alive, and our top-rated teams according to our system are still there, except for Kansas. 

We told you in the first round that Georgetown and Vanderbilt were the most ripe for upset bids based on their R+T scores just barely above zero.  We were there on other double-digit ups as well.

Before we preview the Sweet 16 games, let’s refresh you on the PiRate formula components.

Scoring Margin—We look for teams with a minimum scoring margin of 8 points per game, give precedence to teams with double-digit scoring margins, and develop huge crushes on teams with scoring margins of 15 or more points per game.  We award one point for as little as a 5-point scoring margin, 3 points for 8 or more, and 5 points for 10 or more. 

Teams with a negative margin who have made it to the Sweet 16 are eliminated and are automatically picked to lose the next game (unless of course there is a rare instance of their opponent also qualifying for elimination.)

Field Goal % Margin—We look for teams that have a +7.5 or better difference in field goal percentage versus opponents’ field goal percentage.  We give special consideration to teams with double-digit field goal percentage margins, and if we see a team hitting better than 48.0% and yielding less than 38.0%, we circle that team in red because they are going to be tough to beat if they are a member of one of the Big Six conferences (ACC, Big East, Big Ten, Big 12, Pac-10, or SEC).  We award one point for FG% margins of 5.0 or more, 3 points for margins of 7.5% or more, and 5 points for double-digit margins. 

Like above, teams arriving at the Sweet 16 with a negative field goal margin are eliminated.

Rebound Margin—This is actually part of a multiple statistical entry, as we combine it with turnover margin as well.  However, we do separate rebounding because offensive put backs are vitally important in the Big Dance.  We are looking for teams with a +5.0 or better rebounding margin.  We award one point for a rebounding margin of 3.0 or better and 3 points for a margin of 5.0 or better. 

Teams with a negative rebounding margin receive -2 points, but they are not eliminated yet.

Turnover Margin & Steals Per Game—Teams with negative rebounding margins can make up for it with exceptional turnover margins, especially if they get a lot of steals that lead to great fast break opportunities.  We don’t award points solely on turnover margin and steals; we incorporate those stats into a multi-statistical formula we call “R+T.” 

R+T is a formula that applies weighted advantages to steals and turnover margin, while adding rebounding margin into the equation.  Rebounding margin is already factored into the formula by itself, but it receives fewer awarded points.  This stat balances out the rebounding with the scoring and field goal margin, and it allows us to look at the number of extra scoring opportunities a team normally receives. 

The Formula for R+T is:  R+ (.2S*1.2T), where R is rebounding margin, S is steals per game, and T is turnover margin.  Whenever this stat is negative, this team is immediately eliminated.  If this stat is less than one, don’t figure on this team staying around in the Dance.  All four teams that fell below one in R+T lost in the first round, including heavy favored Georgetown and Vanderbilt.  We award the result of the R+T in points.

Power Conference & Strength of Schedule—We give extra weight to teams that are members of the Big Six conferences.  We give a little weight to the teams from the top of the mid-majors (such as Missouri Valley, West Coast, Colonial, and Mountain West).  We deduct for teams from the lower conferences (such as America East, MAAC, Big West, and Patriot). 

We look at the strength of schedule as produced by, and multiply that number by 100.  50.00 is a mid-point, so if that number is 52.37, we consider that schedule to be 2.37 points stronger than average.  If the number is 46.28, then that schedule is 3.72 points weaker than average.  This is incorporated into our criteria.

Record Away From Home—Every team is playing on a neutral floor, so we throw out the home won-loss records.  A team that is 26-9 overall, but 17-0 at home is actually a .500 team away from home.  Likewise, in some rare instances a team might be 22-10 with a home record of 14-6 and a record away from home of 8-4.  Winning two –thirds of one’s games away from home would make this team more likely to beat the 26-9 team on a neutral floor, all else being equal.

Before the first round, our formula picked Duke as the overall favorite based on their 34.4 PiRate score.  The Blue Devils no longer own the top score after the first two rounds.  Their criteria score fell a little, while another team elevated just enough to post a higher score.  The new leader in the clubhouse is none other than Kansas State.  This surprised us all here, but the Wildcats were impressive in wins over North Texas and BYU.  Their defense was stifling, and their offense, while not spectacular, clicked in spurts.  KSU controlled the boards in both games as well.

The Wildcats have had few great moments since in the last 20+ years.  This team is starting to bring back memories of the glory days in the Little Apple when Tex Winter introduced his triple-post (triangle) offense and Jack Gardner had the Cats running and gunning.

Of the 16 teams remaining, five come from conferences outside of the Big Six conferences, but each of the quintet’s PiRate criteria scores reveals that they belong in the Sweet 16.  None of the five (none of the entire 16) have scores in single digits.

Now, it’s time to look at the eight, Sweet 16 games, using these criteria.  The number you see in (Parentheses) after the team is their PiRate Criteria Score.  All of these scores have been update to reflect their two wins in the Big Dance.                                                                            


East Regional


#1 Kentucky (29.22) vs. #12 Cornell (14.56)

The Wildcats are the one team that also qualifies in the 48-38% field goal margin.  John Calipari no longer officially owns any Final Four appearances to his name, after the NCAA upheld the vacating of all Memphis wins during Derrick Rose’s playing career (his U Mass team had to vacate that appearance as well).  So, we can say he is still looking for his first official visit to the Final Four.  We don’t know with 100% certainty if the Wildcats will make it there, but we are safe in saying they will be one of the Elite 8.  Cornell cannot stop DeMarcus Cousins inside unless they totally sell out on the perimeter.  John Wall and Eric Bledsoe will make the Big Red pay for that tactic, and then Patrick Patterson will break their backs if he hits a three.

Cornell might stay close through one or two TV timeouts, but this game should get out of hand before halftime.


Prediction: Kentucky 88  Cornell 64


#2 West Virginia (29.08) vs. #11 Washington (21.93)

West Virginia wins ugly.  The Mountaineers don’t look pretty, but they keep pounding at opponents until they see an opening.  Then, like a crafty boxer, they exploit that opening and grab the lead on points.  They rarely record a knockout, but they are great at keeping the lead once they get it in the final half.

Washington does look pretty when they play.  Lorenzo Romar’s teams vaguely resemble many of the great UCLA teams from the past.  With Quincy Pondexter and Isaiah Thomas providing a great one-two punch, it is hard to stop the Huskies from scoring 70 or more points.

West Virginia doesn’t usually win games if they give up more than 75 points.  Coach Bob Huggins will devise a game plan to force UW’s big threats to work harder for open shots, and Washington will not reach 75 points in this game.

Prediction: West Virginia 73  Washington 66


South Regional


#3 Baylor (26.04) vs. #10 St. Mary’s (15.47)

This looks like a classic mismatch between a power team from a power conference and a team that should be just glad to have made it this far.  It could be, but we like the way St. Mary’s plays, and we think Coach Randy Bennett is possibly the next Lute Olsen if he so chooses to move on to a school from one of the Big Six conferences.

This will be a great battle between big men.  Baylor’s Ekpe Udoh and St, Mary’s Omar Samhan should balance each other out.  Samhan is a little better offensively, but Udoh is a little better defensively.  Samhan is the more likely to get in foul trouble.

Baylor has more potent weapons in LaceDarius Dunn and Tweety Carter, but the Gaels have more depth.  We just don’t see the Bears running away with this game.  We will pick them to advance.

Prediction: Baylor 78  St. Mary’s 71


#1 Duke (30.48) vs. #4 Purdue (15.37)

Credit must be given to the Boilermakers for making it this far without Robbie Hummel.  They played hard and won a couple of tough games.  Unfortunately, Purdue goes up against one of the big boys.  This is their final game of the season.

Duke may have fallen a notch in winning their first two games, but having to play the play-in winner lowered their strength of schedule.  Emptying the bench may have artificially lowered their criteria score, and we still think Coach K is sitting pretty with his club in a great bracket.

Prediction: Duke 81  Purdue 67


Midwest Regional


#2 Ohio State (22.24) vs. #6 Tennessee (21.16)

These may not be the two best teams left in the Big Dance, or even in this regional, but they may be the two best-coached teams.  Buckeye head guy Thad Matta has definitely produced a better record than his talent on hand should have produced, and Volunteer coach Bruce Pearl has squeezed every last drop of juice out of his big orange.

Two years ago, when Ohio State was the top-rated team, Tennessee built up a 20-point lead against OSU, before the Buckeyes chipped away and came back for the win in this same round.  Vol center Wayne Chism can remember that game well.

We look for this to possibly be the most entertaining game of this round, but we have to go with the Big Ten in this one.  Tennessee is having to go with players that would be considered bench-warmers at Ohio State for almost one quarter of the available playing time.  Pearl will either have to play five reserves for their usual 48 combined minutes per game or go with his top seven until they drop.  Either way, it tips the scale in favor of Brutus.

Prediction: Ohio State 69  Tennessee 63


#5 Michigan State (20.92) vs. #9 Northern Iowa (13.76)

This is another game where we have to discount a team for the loss of a player.  Spartan star guard Kalin Lucas is out for the rest of the year with a ruptured Achilles tendon.  He is the Spartans’ leading scorer, leader at getting to the foul line, leading passer, and best perimeter defender.  Losing him is almost like losing Magic Johnson. 

One thing MSU still has in its favor is a brutalizing inside force with a three-headed rebounding monster.  Raymar Morgan, Draymond Green, and Delvon Roe will see to it that Northern Iowa will not get many second-chance points.

Northern Iowa is primed to exploit MSU’s misfortune, but we expect the Panthers to come out flat following the huge upset over Kansas.  Jordan Eglseder is going to need help inside as the Spartans attempt to force their offense to score inside the paint.  Adam Koch cannot afford to risk foul trouble, so we see some difficulty here for NIU.  We also do not believe that Ali Farokhmanesh will drain threes all night in this game.  We can see him going 2 for 9.

It’s rather obvious that this is going to be a very low-scoring game, at least until the final minutes when one team may be getting a dozen trips to the foul line.

Prediction: Michigan State 56  Northern Iowa 51



West Regional


#1 Syracuse (27.88) vs. #5 Butler (19.35)

Quickness over brute force strength should be the difference in this game.  Syracuse has been flying a little bit under the radar so far, and the Orangemen are about to reveal to the rest of the nation that they are an Elite 8 team. 

Butler cannot be overlooked, as the Bulldogs are now the best team in the Hoosier state.  However, Butler doesn’t have the horses to exploit the cracks in the SU 2-3 matchup zone.  We see the Bulldogs going through stretches where they cannot score, and you can’t beat Syracuse that way.

A ‘Cuse win should set up the best Regional Final of the four, regardless of their opponent on Saturday.

Prediction: Syracuse 74  Butler 60


#2 Kansas State (31.21) vs. #6 Xavier (18.37)

Xavier has become a household name in the Big Dance, so it’s no longer much of a surprise to see the Musketeers advancing in this tournament.  They just happened to get the wrong team in the Sweet 16, because we just cannot see them matching up inside against the purple and white.  Kansas State can bring two wide-bodies off the bench, and the Wildcats’ guards can hit the glass as well.

The storyline of this game is that KSU will hold Xavier under 40% from the field and rarely give the Musketeers an offensive rebound.  Teams just don’t win in the Sweet 16 unless they can either control the boards of shoot a high percentage.

We look for the Wildcats to set up the game of the tournament in the West Regional Finals on Saturday.

Prediction: Kansas State 77  Xavier 61


Check back with us Saturday before game time for a preview of the Elite 8 Regional Final games.


March 14, 2010

Bracketnomics 505–How to pick your NCAA Tournament Brackets

The Advanced Level Class In Bracket Filling

This is a graduate level class that will earn you a PhD in Bracketnomics.  So you want a scientific method to guide you as you fill out your brackets?  You say you want a system that will take out most of the human-bias, and allow you to pick your teams in a mechanical fashion.  Well, we’ve got one for you that has been back-tested and holds up fantastically through the years. 

What the inventor of the PiRate system did was to discover the vital information that has worked in the past.  He’s been using this formula since the Internet made statistics-gathering easy, and it has been back-tested as far back as the days when the NCAA Tournament field consisted of just 23, 24, or 25 teams.

This method will not pick every game correctly and make you an instant millionaire.  It is geared toward finding the tendencies that historically have mattered most in picking the teams with the best chances of advancing.  Not all teams will be a perfect fit in this formula; what this formula does is pick the teams that have the best chance of advancing and making a deep run into the tournament.

How has the formula performed in recent years?  Last year, it picked North Carolina to win the tournament.  Two years ago, it picked Kansas to win the NCAA Championship.  In 2006, it tabbed George Mason as a team to watch to sneak into the Elite 8 (they went to the Final 4).   It correctly selected Florida and UCLA for the Final Four in both 2006 and 2007. 

There have been a couple of seasons where the criteria didn’t apply successfully, but over the course of time, it has performed accurately more than 85% of the time. 

Beginning this year, we add a couple more factors that we hope make this system even more accurate.

Without further adieu, here is the PiRate Bracket-Picking System.

1. Scoring Margin

For general bracket picking, look for teams that outscore their opponents by an average of 8 or more points per game.  Over 85% of the Final Four teams since the 1950’s outscored their opponents by an average of 8 or more points per game. 

This system really loves a team that outscore their opponents by an average of 10 or more points per game while it worships teams outscoring opponents by an average of 15 or more points per game.  More than 80% of the final four teams in the last 50 years outscored their opponents by double digit points per game.  When you find a team with an average scoring margin in excess of 15 points per game, and that team is in one of the six power conferences, then you have a team that will advance deep into the tournament.

This is an obvious statistic here.  If team A outscores opponents by an average of 85-70 and their team B opponent outscores their opposition by an average of 75-70, team A figures to be better than team B before you look at any other statistics. 


In the days of the 64/65-team field, this statistic has become even more valuable.  It’s very difficult and close to impossible for a team accustomed to winning games by one to seven points to win four times in a row.  This average gives the same significance and weighting to a team that outscores its opposition 100-90 as it does to a team that outscores its opposition 60-50.

2. Field Goal Percentage Differential

Take each team’s field goal percentage minus their defensive field goal percentage.  Look for teams that have a +7.5% or better showing.  50% to 42% is no better or no worse than 45% to 37%.  A difference of 7.5% or better is all that matters.  Teams that have a large field goal percentage margin are consistently good teams.  Sure, a team can win a game with a negative field goal percentage difference, but in the Big Dance, they aren’t going to win four games much less two.  This statistic holds strong in back-tests of 50 years.  Even when teams won the tournament with less than 7.5% field goal percentage margins, for the most part, these teams just barely missed (usually in the 5.5 to 7.5% range).  In the years of the 64/65-team tournament, this stat has become a more accurate predictor.  Nowadays, the teams with field goal percentage margins in the double digits have dominated the field.  If you see a team shoot better than 48% and allow 38% or less, that team is going to be very hard to beat in large arenas with weird sight lines.

3. Rebound Margin

This statistic holds up all the way back to the early days of basketball, in fact as far back to the days when rebounds were first recorded.  The teams that consistently control the boards are the ones that advance deep into the tournament.  What we’re looking for here are teams that out-rebound their opposition by five or more per game.  In the opening two rounds, a difference of three or more can be used.

The reason this statistic becomes even more important in mid-March is that teams don’t always shoot as well in the NCAA Tournament for a variety of reasons (better defense, abnormal sight lines and unfamiliar gymnasiums, nerves, new rims and nets, more physical play with the refs allowing it, etc.).  The teams that can consistently get offensive putbacks are the teams that go on scoring runs in these games.  The teams that prevent the opposition from getting offensive rebounds, holding them to one shot per possession, have a huge advantage.  Again, there will be some teams that advance that were beaten on the boards, but over the course of four rounds, it is rare for one of these teams to advance.  West Virginia in 2005 made it to the Elite Eight without being able to rebound, but not many other teams have been able to do so.  There have been years where all four Final Four participants were in the top 20 in rebounding margin, and there have been many years where the champion was in the top 5 in rebounding margin.

4. Turnover Margin & Steals Per Game

Turnover margin can give a weaker rebounding team a chance.  Any positive turnover margin is good here.  If a team cannot meet the rebounding margin listed above, they can get by if they have an excellent turnover margin.  Not all turnover margin is the same though.  A team that forces a high number of turnovers by way of steals is better than a team that forces the same amount of turnovers without steals.  A steal is better than a defensive rebound, because most of the time, a steal leads to a fast-break basket or foul.  When a team steals the ball, they are already facing their basket, and the defense must turn around and chase.  Many steals occur on the perimeter where the ball-hawking team has a numbers advantage.  So, this system counts a steal as being worth 1.33 rebounds. 

The criteria to look for here is a positive turnover margin if the team out-rebounds its opposition by three or more; a turnover margin of three or more if the team out-rebounds its opposition by less than three; and a turnover margin of five or more if the team does not out-rebound its opponents.  Give more weight to teams that average 7.5 or more steals per game, and give much more weight to teams that average double figure steals per game.  A team that averages more than 10 steals per game will get a lot of fast-break baskets and foul shots.  In NCAA Tournament play, one quick spurt can be like a three-run homer in the World Series, and teams that either steal the ball or control the boards are the ones who will get that spurt.

4a. The All-Important R+T Margin: Consider this the basketball equivalent of baseball’s OPS (On Base % + Slugging %).  Here is the PiRate R+T stat: R + (.2S * {1.2T}), where R is rebounding margin, S is average steals per game, and T is turnover margin.  When this stat is 5 or more, you have a team that can overcome a few other liabilities to win.  When the result is 10 or more, you have a team that has a great chance of getting enough additional scoring opportunities to make it to the later rounds.  When this stat is negative, you have a team that will be eliminated before the Sweet 16.

5. Power Conference Plus Schedule Strength

I’m sure up to this point you have been thinking that it is much easier for Sam Houston or Siena to own these gaudy statistics than it is for Kansas State or West Virginia.  Of course, that’s correct.  We have to adjust this procedure so that the top conferences get extra weight, while the bottom conferences get penalized.  Here is how we do it.  Look at the Strength of schedule for every team in the Field.  You can find SOS on many websites, such as the RPI at  Take the decimal difference for each team in the Field and multiply that by 100.  For example if Team A’s SOS is .6044 and Team B’s is .5777, the difference times 100 is 2.67.  So, Team A’s schedule was 2.67 points (or round it to 3) per game tougher than Team B’s.  Use this in head-to-head contests for every game in your bracket.

6. NEW ADDITION FOR 2010: Won-Loss Record away from home.

Let’s say a big six conference team with a 22-8 record might have gone 16-0 and home and 6-8 away from home, while another team from a different big six conference with a 19-13 record might have gone 9-7 at home and 10-6 on the road.  The overall stats might show the 22-8 team to be better, but they built up those impressive numbers by padding the stat book at home.  Look for teams that can win by double digits away from home against teams in big six conferences.

These are the six basic PiRate criteria used for picking bracket winners.  You might be shocked to see that there are some key statistics that are not included.  Let’s look at some of these stats not to rely upon.

Assists and Assists to Turnover Ratio

While assists can reveal an excellent passing team, they also can hide a problem.  Let’s say a team gets 28 field goals and has 21 assists.  That may very well indicate this team can pass better than most others.  However, it can also mean two other things.  First, this team may not have players who can create their own offense and must get by on exceptional passing.  That may not work against the top defensive teams in the nation, or the type that get into the Dance.  Second, and even more importantly, it may indicate that this team cannot get offensive putbacks.  As explained earlier, the offensive putback is about as important as any stat can be.  So, consider this stat only if you must decide on a toss-up after looking at the big six stats.

Free Throw Shooting 

Of course, free throw shooting in the clutch decides many ball games.  However, history shows a long line of teams making it deep into the tournament with poor free throw shooting percentages, and teams that overly rely on free throws may find it tough getting to the line with the liberalized officiating in the tournament.

Let’s say a team shoots a paltry 60% at the foul line while their opponent hits a great 75% of their foul shots.  Let’s say each team gets to the foul line 15 times in the game, with five of those chances being 1&1, three being one shot after made baskets, and seven being two shot fouls.  For the 60% shooting team, they can be expected to hit 3 of 5 on the front end of the 1&1 and then 1.8 of the 3 bonus shots; they can be expected to hit 1.8 of 3 on the one foul shot after made baskets; and they can be expected to hit 8.4 of 14 on the two shot fouls for a total of 15 out of 25.  The 75% shooting team can be expected to connect on 3.75 of 5 on the front end of the 1&1 and then 2.8 of 3.75 on the bonus shot; they can be expected to hit 2.3 of 3 on the one foul shot after made baskets; and they can be expected to connect on 10.5 of 14 on the two shot fouls for a total of 19.35 out of 25.75.  So, a team with one of the top FT% only scores nine more points at the foul line than a team with one of the worst.  That looks like a lot of points to make up, but consider that this is about the maximum possible difference.  Also consider that teams that shoot 60% of their foul shots and make the NCAA Tournament are almost always the teams that also have the top R+T ratings.  Teams that make the NCAA Tournament with gaudy free throw percentages frequently got there by winning close games at the line.  In the NCAA Tournament, fouls just don’t get called as frequently as in the regular season.  The referees let the teams play.  So, looking at superior free throw percentage can almost lead you down the wrong path. 

Ponder this:  The 1973 UCLA Bruins are considered to be the best college basketball team ever.  That team connected on just 63% of its free throws.  They had a rebounding margin of 15.2, and they forced many turnovers via steals thanks to their vaunted 2-2-1 zone press.  In the great UCLA dynasty from 1964 through 1973 when the Bruins won nine titles in 10 years, they never once connected on 70% of their free throws and averaged just 66% during that stretch.

3-point shooting

You have to look at this statistic two different ways and consider that it is already part of field goal percentage and defensive field goal percentage.  Contrary to popular belief you do not count the difference in made three-pointers and multiply by three to see the point-difference.  If Team A hits eight treys, while their Team B opponents hit three, that is not a difference of 15 points; it’s a difference of five points.  Consider made three-pointers as one extra point because they are already figured as made field goals.  A team with 26 made field goals and eight treys has only one more point than a team with 26 made field goals and seven treys.

The only time to give three-point shots any weight in this criteria is when you are looking at a toss-up game, and when you do look at this stat, look for the team that does not rely on them to win, but instead uses a credible percentage that prevents defenses from sagging into the 10-12-foot area around the basket.  If a team cannot throw it in the ocean from behind the arc, defenses can sag inside and take away the inside game.  It doesn’t play much of a role in the NCAA Tournament.  A team that must hit 10 threes per game in order to win isn’t going to be around after the first weekend.  A team that goes 5 of 14 from behind the arc is better than a team that goes 10 of 28.  Both teams got to the Big Dance, and the team that took only 14 treys per game probably got there with an imposing inside game and great defense—just what is needed to cut the nets six games later.

One Big Star or Two Really Good Players

Teams that got to the Dance by riding one big star or a majority of scoring from two players are not solid enough to advance very far.  Now, this does not apply to a team with one big star and four really good players.  I’m referring to a team with one big star and four lemons or two big scorers with three guys who are allergic to the ball.  Many times a team may have one big scorer or two guys who score 85% of the points, but the other three starters are capable of scoring 20 points if they are called on to do so.  If you have a team with five double figure scorers, that will be a harder one to defend and one that will be consistent.  It’s hard for all five players to slump at once.

We hope this primer will help you when you fill out your brackets this week. 

Now, here is a way to put numbers to the criteria.  It isn’t exactly the way our founder did it every year, but it is a close approximation.

1. Scoring Margin

Award 5 points for every team with a scoring margin difference of 10 or more

Award 3 points for every team with a scoring margin difference of 8.0-9.9

Award 1 point for every team with a scoring margin difference of 5.0-7.9

Award 0 points for every team with a scoring margin difference of 0-4.9

Award -3 points for every team with a negative scoring margin

2. Field Goal % Margin

Award 5 points for every team with a FG% margin difference of 10% or more

Award 3 points for every team with a FG% margin difference of 7.5 to 9.9

Award 1 point for every team with a FG% margin difference of 5.0-7.4

Award 0 points for every team with a FG% margin difference of 0.0-4.9

Award -3 points for every team with a FG% margin difference below 0

3. Rebound Margin

Award 3 points for every team with a Rebound margin difference of 5 or more

Award 1 point for every team with a Rebound margin difference of 3.0-4.9

Award 0 points for every team with a Rebound margin difference of 0-2.9

Award -2 points for every team with a Rebound margin difference below 0

4. Turnover Margin

Award 3 points for every team with a Turnover margin difference of 3 or more

Award 1 point for every team with a Turnover margin difference of 1.5-2.9

Award 0 points for every team with a Turnover margin difference of 0-1.4

Award -2 points for every team with a Turnover margin below 0

5. PiRate R+T Formula

Once again, the formula for R+T is [R + ({.2*S}*{1.2*T})], Where R is rebounding margin, S is avg. steals per game, and T is turnover margin

Award 5 points for every team with an R+T of 10 or more

Award 3 points for every team with an R+T of 7.5-9.9

Award 1 point for every team with an R+T of 5-7.4

Award 0 points for every team with an R+T of 0-4.9

Completely eliminate from consideration all teams with a negative R+T

6. Performance Away From Home

Award 3 points for every team that won 75% or more of its games away from home

Award 2 points for every team that won 60-74% of its games away from home

Award 1 point for every team that won 51-59% of its games away from home

Award -2 points for every team that had a losing record away from home

7. Schedule Strength

Use this to compare when looking at team vs. team.  Take the difference in the Strength of Schedule as given by and multiple it by 100.  For example, Team A with an SOS of .5252 has a schedule 7 points weaker than Team B with an SOS of .5921.  If these two teams face each other, give the Team B an extra 7 criteria points over Team A ([(.5921-.5252)*100]=6.69 rounds to 7).

If you want to compile all this information yourself, the best way is to go to all 65 official athletic websites of the teams in the Big Dance.  You will find up-to-date statistical information.  Some of these stats are available in other places, but many have been found to be riddled with mistakes, or they are not up-to-date.  All 65 school sites are accurate and timely.


Coming Wednesday, we will break down the brackets and show you which teams we expect to advance past the first and second weekends.  We will break down every first round game and show you how the formula works in these games. 

Create a free website or blog at