Adjusting College Stats to Statis Pro Basketball

Suppose we have two teams in Statis Pro Basketball - Team A and Team B. We have two different sets of stats, but we know that Team A is stronger than Team B. How should Team A's stats be adjusted?

First, we need to know the answer, "how strong"? The site I prefer to use is Real Time RPI.com to determine the relative strengths of conferences.

Note that I said conferences, not teams. Teams play within a conference, and their stats already reflect relative strength within a conference - a team can play up to 1/2 of its total games as conference matchups. If Team A and Team B are within the same conference, use the unadjusted stats.

If Team A and Team B are within different conferences, we will look at Team A and Team B's relative RPI. Let's assume that Team A is a Big East Team (like Rutgers) and Team B is a Colonial Athletic Team (like James Madison).

Go to the Conference RPI page of Real Time RPI. Let's look at how strong the Big East is to the Colonial Athletic

Big East: 0.6000
Colonial Athletic: 0.5194

Adjustment: 0.6/0.5194 = 1.1552

The 1.1552 becomes the factor we will use. In general, the Big East is 15 percent better than the CAA.

For Team B, we use the regular stats
For Team A, we use adjusted stats.

Multiply field goals made by the factor.
Multiply 3PG made by the factor.
Multiply FT made by the factor.
Multiply all rebounds by the factor.
Divide personal fouls by the factor.
Multiply assists by the factor.
Divide turnovers by the factor.
Multiply blocks by the factor.
Multiply steals by the factor.

If you need to know the new adjusted points per game, determine how many points the player would have scored for the season with the adjusted numbers and divide by games played. In general, the rule is that we want to reward the player by making good things better and making bad things not as bad.

At least, I know how to adjust Central Oklahoma's stats. Problem: Central Oklahoma does not play in a Division I conference. What should the conference multiplier be?

There are all sorts of ways to answer this question. My way was to determine what conference the #202 team was in and consider that as Central Oklahoma's conference.

Central Oklahoma's proxy conference/Conference USA (the one UTEP belongs to)

= .5051/.4952 = approximately 1.02

Note that these numbers have changed since then. But the conclusion has been decided. Central Oklahoma is 2 percent better than their statistics indicate. We adjust the stats of the Central Oklahoma Bronchos by 2 percent, and go from there.

Why do we adjust Central Oklahoma's stats up? Why not adjust UTEP's stats down?

My reasoning has to do with how the first round of the NCAA tournament concludes, with very strong team playing very weak teams. The score is usually something like 90-50, and not 50-10. I would expect something like 50-10 if the weaker team were rounded down and the stronger one unchanged. Changing the stronger team should create a score something like the typical first-round blowouts of the NCAA tourney.

Next: we'll take a peak at Central Oklahoma's card set.

How Strong is a Division II Team?

In 2007-08, the UTEP Miner's womens basketball team started off their season with a game against St. Mary's (Texas), a high-ranked Division II school. I decided to have my fictional UTEP Miners play the Central Oklahoma Broncos, a real life Division II team ranked #16 among Division II teams at the time.

If both of these teams were pro teams, card creation would be easy: load in the statistics, print out the cards, and play the game. However, one would expect that the two teams wouldn't be on an equal basis. So how do you alter the cards? And which set of cards do you alter?

I figured that at least for now, I would let the UTEP cards alone and make alterations to UTEP's prospective opponents. We would look at Central Oklahoma's statistics, but we also know that Central Oklahoma's statistics are derived from playing other teams in Division II. How do we change the cards to reflect the difference?

The first question was: How strong is Division II compared to Division I? We really don't have an answer for this question in basketball. Division I teams don't play Division II teams that often, and when they do, the sample of Division II teams probably isn't a representative sample. This might be an insurmountable problem.

However, there is a division that is similar to Division II - the football division that was formerly called "I-AA" before they changed it to some convoluted name. I-AA teams play BCS teams all the time, usually at the beginning of the season. Was there a way to rank the I-AA teams compared to their BCS counterparts?

As it turns out, this has already been done. There are sites on the internet that rank Division I and Division I-AA teams together in the same table - after all, if you're making bets, you want to know exactly how strong that unknown Division I-AA team is. I managed to locate such a table and found that the strongest Division I-AA team was ranked about #50 overall.

There are about 120 BCS-type football teams. I came up with the calculation: 50/120 = 5/12. The strongest Division II basketball teams should be 5/12 of the way down the table of Division I basketball teams.. This puts that team at about #142 on the total ranking of 340 or so Division I women's basketball teams.

However, the I-AA teams, when they appear, did not appear in close order. There was an average gap of about four places between appearances among I-AA teams in the BCS/I-AA table. I came up with a new rule:

The #1 Division II women's basketball team would be ranked about #142 overall in Division I women's basketball,
The #2 Division II team would be ranked about 142 + 4 = #146 overall in Division I,

and so on.

Central Oklahoma at the time was ranked about #16. This put Central Oklahoma equivalent to about #202 on the scale of Division I women's basketball teams.

Meaning that Central Oklahoma is, believe it or not, better than UTEP. I don't find that too implausible. Central Oklahoma was nationally ranked at the time, and UTEP was sitting around .500.

The next question: Now that we know the team's relative positions to each other, how do we adjust the stats? More on my next post.