Hovell Index

The Hovell Index was created by John Hovell in 2005. It is an approach to predicting the winner of the NCAA College Basketball Tournament. The index includes 15 fields that are converted into a numeric value. The 15 numeric values are tallied into a "Total Index Value". The Total Index Values are then used to pick the winner of each matchup in the bracket.

Fields in the Hovell Index

Years of experience (per player in the starting five)
Height of each of the starting five
Points for - ranking
Points against - ranking
# of ranked teams beat
Free throw %
Rebound ranking
Field goal %
Three point ranking
Number of wins (last 10 games)
Depth (# of players who play more than 10 minutes per game)
Turnover ranking
RPI
# Naismith potentials
Seed

Translation of Fields into Values

This translation is the "magic" of the Hovell Index. Each raw statistic is converted from a pure stat to a "Hovell Index" value. All of these values are calculated through a conversion table. The conversion table values have been pre-determined and slightly modified over the years as the Hovell Index matures and improves.

The Hovell Index prioritizes each of the fields in this order - experience, rpi, # ranked teams beat, # Naismith potentials, # of wins in last 10 games, height factor, field goal %, points against - rank, points for - rank, free throw %, # turnovers, rebounding rank, depth, three pointer rank, seed. Due to that prioritization, each conversion table offers more or less points depending on the priority of the field. For example, an above average team will earn 101 points for experience, 75 points for RPI and 70 points for the number of ranked teams they beat (notice the declining points even though the team performed well in all of those fields).

There is an Excel spreadsheet (attached here) that automatically calculates the Hovell Index after all of the raw data is entered.

Here is exactly how each conversion is calculated (for each team):

Experience:

Start with the class rank of each of the starting five. A starting freshman earns 1 point. A starting sophomore earns 3 points. A starting junior earns 6 points. A starting senior earns 10 points.
Tally the points of the starting five
Use the conversion table to calculate the Hovell Index value
Example: In the 2010 tournament, Duke had 4 starting juniors and 1 starting sophomore. (4 starting juniors * 6 points)+ (1 starting sophomore * 3 points) = 27 tallied points. 27 tallied experience points converts to 69 points in the Hovell Index.
Rationale of the Ranking - experience seems to be the most important attribute of a championship team

RPI:

Start with the team's RPI value right before the tournament
Use the conversion table to calculate the Hovell Index value
Example: In the 2010 tournament, Duke had an RPI of 4, which is worth 73.8 points in the Hovell Index.
Rationale of the Ranking - RPI is a common factor for trying to compare apples to apples. Using this as one input to a larger system seems to help validate the RPI rankings.

Number of Ranked Teams Beat:

Start with the number of teams ranked in the Top 25 (ESPN) that this team beat (including conference tournaments)
Use the conversion table to calculate the Hovell Index value
Example: In the 2010 tournament, Duke had beat 5 ranked teams which is worth 50 points in the Hovell Index.
Rationale of the Ranking - Similar to the RPI ranking, so it is almost a double counting of the strength of schedule. This statistic proves that the team has the ability to beat great teams.

Number of Potential Naismith Award winners:

Lookup the list of potential Naismith Award winners and enter the number of potential winners that this team has
Use the conversion table to calculate the Hovell Index value
Example: In the 2010 tournament, Duke had beat 1 potential Naismith Award winner, which is worth 40 points in the Hovell Index.
Rationale of the Ranking - While basketball is one of the ultimate team sports, there is an advantage to having top talent.

Number of wins in last 10 games:

Count the number of wins in the team's last 10 games (admittedly, I add or remove 'half points' if the team beat ranked teams or lost to un-ranked teams)
Use the conversion table to calculate the Hovell Index value
Example: If a team won 9 of their last 10 games, but the one loss was to a ranked team, I would probably assign 9.5 points (so you can see this is a little subjective), which equates to 40 points in the Hovell Index.
Rationale of the Ranking - There seems to be an advantage to going into the tournament on a 'hot streak'.

Height Factor:

This one is a little tricky.
Enter the heights of the five starters for all teams (ordering them from shortest to tallest in the row for the team).
Some teams have point guards over 6 feet, but some are under 6 feet tall. So, I use 5.12 (for 5 feet, 12 inches) as the equivalent to 6 feet. For 5'11", I use 5.11. For 6'1", I use 5.13.
Calculate the average for each of the height columns
Highlight the cells (based on the column average) where that player is taller than average
Count the number of highlighted cells per team
Use the conversion table to calculate the Hovell Index value
Example: In the 2010 tournament, Duke had starting heights of 5.15, 6.05, 6.06, 6.07 and 6.08. 5.15, 6.05, and 6.06 were 'above average' for 2010, so Duke earned 3 'height factor' points, which is worth 32 points in the Hovell Index.
Rationale of the Ranking - Height can be an advantage, especially when its above average (the other fields cover for the 'talent' of that height).

Field Goal Percentage:

Lookup the team's ranking in the top 100 for their field goal percentage (max ranking is 100 even if they are 'worse/higher' than that)
Enter that ranking into the spreadsheet
Use the conversion table to calculate the Hovell Index value (field goal percentage, points against, points for, free throw percentage, 3 pointer ranking, and rebounds all use the same conversion table)
Example: In the 2010 tournament, Duke was ranked worse/higher than 100, so they earned 100 points, which equates to 18 points in the Hovell Index.
Rationale of the Ranking - field goal percentage is pretty key - you need to be able to make your shots. Putting this field just ahead of 'points for - ranking' was a tough call (which is more important, making your shots or stopping shots)

Points Against - Ranking:

Lookup the team's ranking in the top 100 for their points against (aka defense) (max ranking is 100 even if they are 'worse/higher' than that)
Enter that ranking into the spreadsheet
Use the conversion table to calculate the Hovell Index value (field goal percentage, points against, points for, free throw percentage, 3 pointer ranking, and rebounds all use the same conversion table)
Example: In the 2010 tournament, Duke was ranked 11th, so they earned 11 points, which equates to 36.2 points in the Hovell Index.
Rationale of the Ranking - points against is your defensive ranking (which can be hard to find online, by the way). Defense is key to winning championships - notice that the fields using the same conversion table receive an equal priority due to exactly the same point conversion.

Points For - Ranking:

Lookup the team's ranking in the top 100 for their points against (aka offense) (max ranking is 100 even if they are 'worse/higher' than that)
Enter that ranking into the spreadsheet
Use the conversion table to calculate the Hovell Index value (field goal percentage, points against, points for, free throw percentage, 3 pointer ranking, and rebounds all use the same conversion table)
Example: In the 2010 tournament, Duke was ranked 89th, so they earned 89 points, which equates to 20.4 points in the Hovell Index.
Rationale of the Ranking - points-for is your offensive ranking, which is only slightly more important than free throw percentange, turnovers, rebounds, team depth, number of threes and your tournament seed.

Free Throw Percentage:

Lookup the team's ranking in the top 100 for their free throw percentage (max ranking is 100 even if they are 'worse/higher' than that)
Enter that ranking into the spreadsheet
Use the conversion table to calculate the Hovell Index value (field goal percentage, points against, points for, free throw percentage, 3 pointer ranking, and rebounds all use the same conversion table)
Example: In the 2010 tournament, Duke was ranked 65th, so they earned 65 points, which equates to 25.2 points in the Hovell Index.
Rationale of the Ranking - free throws are important in the tournament - especially in close matchups

Turnovers:

Lookup the team's number of average turnovers (per game, all players)
Enter that number into the spreadsheet
Use the conversion table to calculate the Hovell Index value
Example: In the 2010 tournament, Duke averaged 11.2 turnovers per game, which is equivalent to 29.04 points in the Hovell Index
Rationale of the Ranking - turnovers tend to be a predictor for ball control, smart decisions and unforced errors

Rebound - Team Ranking:

Lookup the team's ranking in the top 100 for their rebounds (max ranking is 100 even if they are 'worse/higher' than that)
Enter that ranking into the spreadsheet
Use the conversion table to calculate the Hovell Index value (field goal percentage, points against, points for, free throw percentage, 3 pointer ranking, and rebounds all use the same conversion table)
Example: In the 2010 tournament, Duke was ranked 23rd, so they earned 23 points, which equates to 33.6 points in the Hovell Index.
Rationale of the Ranking - rebounds tend to be the performance of height, smart play under the basket, or just pure tenacity

Depth:

Lookup the team roster
Sort by 'minutes played'
Count the number of players that average 10 minutes or more per game
Enter that ranking into the spreadsheet
Use the conversion table to calculate the Hovell Index value
Example: In the 2010 tournament, Duke had 9 players average more than 10 minutes per game, which equates to 17 points in the Hovell Index.
Rationale of the Ranking - the deeper the team, the less likely they are to be tired and the more likely they are to outlast other teams

Three Pointer - Team Ranking:

Lookup the team's ranking in the top 100 for their team ranking of three pointers (max ranking is 100 even if they are 'worse/higher' than that)
Enter that ranking into the spreadsheet
Use the conversion table to calculate the Hovell Index value (field goal percentage, points against, points for, free throw percentage, 3 pointer ranking, and rebounds all use the same conversion table)
Example: In the 2010 tournament, Duke was ranked over 100, which equates to 18 points in the Hovell Index.
Rationale of the Ranking - three pointers are the great equalizer, but it can be tough to count on them under the stress of the tournament

Seed:

Enter the teams tournament seed number into the spreadsheet
Use the conversion table to calculate the Hovell Index value
Example: In the 2010 tournament, Duke was a 1 seed, which equates to [only] 10 points in the Hovell Index.
Rationale of the Ranking - this probably shouldn't be here because it favors the lower (better) seed, which is exactly what this entire Index is trying to prevent/predict. However, the logic is that the selection committee knows a little something, so a few extra points to a lower/better seed is probably accurate (and the points are very low, so its really only a tiebreaker for close matchups)

Results

The Hovell Index has correctly predicted the winner of the NCAA Men's Basketball Tournament since 2005.

Room for Improvement

These fields might aid in the accuracy of the Hovell Index.

home team - distance to stadium
how can you numerically account for chemistry? Assists?
speed?
division? conference?
steals? (turnovers)
injury?
assists?
use 6 or 7 deep on experience and height, not just starting 5?
team tourney experience? (were they there last year - or how many years over past 4)
watch them play a game (dvr)
coach?
style matchup…
player by player matchup…
tempo
pace
intensity
desire to win
number of times out of 7 each team would win this matchup
number of road wins - especially against ranked opponents
time of possession (can they slow down the tempo)
team that got snubbed - slightly lower rank than they should have
number of possessions during the regular season

References

This article is issued from Wikipedia - version of the Wednesday, January 07, 2015. The text is available under the Creative Commons Attribution/Share Alike but additional terms may apply for the media files.