Hoopville Unveils the 2nd Generation of the TIQ Player Rating System

Hoopville has a new and improved Total Impact Quotient player ratings ready for the 2011-12 season. This might be the most logical rating system we’ve seen for NCAA players, and it has massive advantages over last season’s version.

During the off-season, I grabbed a few books about player rating systems, focusing mostly on John Hollinger‘s “Pro Basketball Forecast” and Dean Oliver‘s “Basketball on Paper.” Armed with those books and a notepad, a couple of cross-country flights felt like short bus rides rather than five-hour journeys cramped into a seat with inadequate leg room for a 6’3” numbers nerd. I feel personally indebted to Oliver, who saved me from an unwanted conversation with an anti-Obama religious conservative from Pensacola, Fla., who seemed determined to lecture me on stock options and/or the likelihood that a Jew marrying a Buddhist would likely end poorly. Give me Advanced Basketball Stats 201 any day of the week.

But I digress. Here are a few major points about this year’s TIQ rating system.

First, let me apologize to all guards out there. The TIQ circa 2010-11 undervalued guards’ assist totals — partly because the formula didn’t adequately account for the points scored when a player assists another player’s basket. Meanwhile, defensive rebounds led to a large contribution in preventing the other team from scoring. Together, the TIQ formula turned guards into oranges being compared to big men’s apples. This year, the formula rewards a more realistic total for assists, and I don’t need to torture the numbers to compare guards to forwards. It’s just apples to apples.

The next improvement is the strong correlation to players’ TIQ ratings and a team’s overall success. In short, for various reasons, last season’s TIQ rating system had almost no correlation to a team’s actual wins and losses. Not so this year.

After changing the recipe, the TIQ now is a strong predictor of overall team success. If you add up the individual TIQ ratings for a team’s players, you can compare the results among teams and make reasonable conclusions. Kansas had the highest TIQ of any team in the Big 12, and the Jayhawks cruised to a conference title. However, Oklahoma had the lowest TIQ total, and the Sooners finished next to last.

Here’s what’s new in this year’s TIQ formula:

• Tweaked to more accurately account for points scored by teammates on a player’s assists, primarily by estimating two-pointers and three-pointers made on those assists and then properly counting those points.
• Elimination of the Opportunity Cost factor, which attempted to gauge how many points teammates would have scored if a player ceded all his shots. This was just a dumb factor in retrospect.
• Addition of a Points Lost from Missed Field Goal Attempts factor. This is similar to what I wanted to do with the Opportunity Cost factor but simply isolates points lost to a player’s missed shots from the field. If a player is jacking up all kinds of shots and only shooting 37 percent, the team will grab a certain number of offensive rebounds that lead to a score. But for the rest of the shots that end in a defensive rebound, this factor counts how many points the team is losing.
• Addition of Points Lost from Missed Free Throw Attempts factor. This factor is like the field goals one, except limited to a player’s lack of production at the free throw line.

The final product is also significantly different. Although the final metric is still the total number of points that a player contributes per 40 minutes, it’s now based on Division I averages for several important statistics. I’m using D-I averages for team tempo, offensive efficiency and defensive efficiency. By normalizing those stats, you can actually compare players outside the context of their team. So we’ll order the rankings based on a player’s contribution per 40 minutes.

The weekly TIQ updates also will include stats for percent minutes played and real TIQ per game. While the TIQ per 40 minutes compares players’ contributions regardless of minutes played, the real TIQ per game metric shows how many points a player is really contributing based on his playing time. The best way to think of this difference is by asking this question: Can my backup point guard who averages six points and three assists in only 12 minutes per game and has a TIQ around 30 (that’s very good) actually maintain that much of an impact if he starts playing 28 minutes per game.

Here’s a more in-depth tutorial of the TIQ.

What qualifies as a good TIQ number?

For 2010-11, the players from the six power conferences averaged a TIQ of 22.3 points. For all you stat-heads out there, the standard deviation was 5.5 points. That means that roughly 50 percent of all players put up TIQ ratings between 16.8 points and 27.8. Players with TIQs better than 33.3 points are in the top 5 percent, while players with TIQs less than 11.3 points are in the bottom 5 percent.

Why is it called a quotient?

It’s mostly a matter of semantics, but the final calculation for players’ contribution per game includes dividing total points contributed by minutes played. And the term for the final result of a division calculation is “quotient.”

What factors go into calculating the TIQ?

I have nine factors in this wonderfully improved version:

• Points scored when a player shoots a field goal or free throw attempt.
• Points scored when a player dishes an assist.
• Points scored when a player grabs an offensive rebounds.
• Points saved when a player steals the ball.
• Points saved when a player blocks a shot.
• Points saved when a player grabs a defensive rebound.
• Points lost when a player commits a turnover.
• Points lost when a player misses a field goal attempt.
• Points lost when a player misses a free throw attempt.

Can I get more information about the statistics used in this formula?

Absolutely. Please contact me about your specific inquiry. I’ll get back to your as quickly as possible.

Why do you only have player ratings for major conferences? Where’s the love for the smaller conferences?

I love smaller conferences, too. However, it’s an entirely manual process to generate these statistics. I only have time to produce statistics for the six major conferences right now, though I’d love to expand this metric to all D-I teams at some point in the future.

Why isn’t every player on a team included?

I’m only calculating stats for players who play at least 10 percent of a game. Everyone else’s real contributions are miniscule, and it would be easier for reserves to post unrealistically high TIQs because they play few minutes and would only need a few points and turnovers per game to have unbelievably good per-40 minute TIQs.

What do you mean by the correlation between TIQ and team results?

In regards to statistics and mathematics, a correlation evaluates the strength in the relationship between two sets of data. A correlation closer to 1.0 shows a direct positive relationship, while a value closer to -1.0 shows a direct negative relationship. For example, teams usually win more games when they shoot well from the field, and they lose when they struggle. This makes sense because you need to score to win, and you’re not scoring much if your field goal percentage is terrible. Therefore, there’s a strong positive correlation between a team’s field goal percentage and winning; the value would be close to 1.0.

In this version of the TIQ, there’s a strong correlation between the sum TIQ of a team’s players and the team’s position in conference standings. To be precise, there’s a 0.7874 correlation between the real TIQ rankings of teams and the final conference standings from 2010-11. That makes the TIQ a pretty darn good measure of team performance and lends credence to the system for accurately measuring player performance.

What are your sources for statistics?

I used easily available information for players statistics, usually from ESPN.com. For team efficiency and tempo information, I use information from the wonderful kenpom.com, run by Ken Pomeroy, another stat aficionado.