RPI Deficiency: What is RPI in Basketball, Football, and Baseball?

As little as 10 years ago, as the NCAA Men’s basketball tournament approached, sports shows started to talk at length about RPI. Invariably, this would lead the half-interested to ask “What is RPI”? While the system is fairly simple and seems to tick a lot of boxes, there is certainly some RPI deficiency too.

In this article we’ll look at the history of the RPI. We’ll touch on why it was conceived as a metric, how it is calculated, and the RPI deficiency that causes it to fall short of its goal.

To receive email updates when new articles are posted, use the subscription form below!

What is RPI?

RPI is short for ratings percentage index. It started in 1981 as a method used by the NCAA to rank and rate teams in college basketball. Since its inception, it has been adopted for use in many other collegiate sports. However, as I alluded to in the introduction, since 2018 RPI has actually seen a decrease in its usage. This is because of a well-known RPI deficiency and because of the existence of other, better, more proven methods for rating college basketball teams.

The system was originally designed as a potential solution to the strength of schedule problem. Especially in college sports, it can be very difficult to determine how good a team is just based on their record. This is the reason stats like strength of record exist – to help contextualize a team’s record by looking at how good their opponents are.

RPI does this by combining three things: your record, your opponents’ records, and your opponents’ opponents’ records. In the next section we’ll see exactly how this is done.

How is RPI Calculated?

Three things are combined to compute this metric:

1. My winning percentage (25%)
2. My opponents’ winning percentage (50%)
3. My opponents’ opponents’ winning percentage (25%)

In this way, it is a weighted combination of many different records. The idea is that if you play better opponents, then their records will be good. Therefore your opponents’ winning percentage will be a larger number and your ratings percentage index will become larger.

But measuring opponent quality isn’t that simple. Remember that the whole point of RPI is that our record isn’t the only indicator of how good we are. Therefore, our opponents’ record isn’t the only indicator of how tough my schedule was! In order to determine how good my opponents were, we should consider their opponents’ records! This is why the formula includes our opponents’ opponents’ records.

How is RPI Calculated?

We want to look now at the precise formula for this stat. Let WP be my winning percentage. Next, let OWP and OOWP be my opponents’ winning percentage and opponents’ opponents’ winning percentage. The RPI formula is: RPI = \frac{WP}{4} + \frac{OWP}{2} + \frac{OOWP}{4} .

This means that the stat is one part how well we’ve done, two parts how well our opponents did, and one part how well our opponents’ opponents did. If this seems weird to you, it should. The emphasis on other team’s performance is actually one of the reasons that this stat doesn’t work very well at all.

Why Ratings Percentage Index is Only Used in College

RPI is primarily used in college athletics, especially so in baseball and in basketball. Why is this the case? Remember that the point of this metric is to assess a team’s record combined with their strength of schedule to try to estimate how good a team is.

It is helpful in settings where two teams with the same record can vary a lot in quality. This happens in leagues where two teams can have dramatically different strengths of schedule.

This is why it shows up predominantly in college sports. College athletics tend to have shorter seasons in leagues with more teams. For example, college basketball is roughly a 30 game season in a 300 team league. Compare this to the NBA where there is a roughly 80 game season in a 30 team league. Picking 30 games out of 300 teams leads to much more variance than picking 80 games from among 30 teams.

How RPI is used

RPI is an old system, dating back to at least the 1980s. The 1980s was truly the wild west of sports statistics – it was much before the moneyball era or the data science revolution of the 2000s. It was a valiant effort that addressed a major problem. Let’s see how it was applied.

RPI in basketball

Perhaps the most visible application of RPI was for the selection process in the NCAA college basketball tournament. As I discussed above, college basketball has a huge problem when it comes to figuring out who is the best. Because the league is so large and team quality varies so much from top to bottom and from conference to conference, teams with the same records can have wildly disparate qualities.

The RPI system was used from 1981 until 2018 as an official metric used by the selection committee in selecting the men’s tournament field. Between the 2018 and 2019 seasons, the NCAA announced that RPI would be retired and would no longer be a contributing factor to the selection of the tournament field.

Though the reason that RPI was retired was its well known deficiencies, the system was bizarrely still used in the selection of the women’s tournament field for another 2 years. Now, both the men’s and women’s tournaments use the NCAA Evaluation Tool (NET) as an official deciding factor. NET ratings are a much more mathematically mature system that better represents the actual quality of a team.

RPI in baseball

Though not nearly as well known, RPI is also used in NCAA baseball to rank teams. It was introduced in college baseball to replace an archaic system that rewarded points for wins based on which quadrant a team was in. For example, you got the most points for a top 25 win, one less point for a win over a 26-50th ranked team, etc.

The old baseball system was not unlike basketball’s current quadrant system used to evaluate strength of schedule. However, the points system in baseball resulted in a totally arbitrary metric that penalized or rewarded certain teams based on the strength of their schedule and not on their resume.

To fix this issue, ratings percentage index was introduced in college baseball. However, as we’ll see in the next section, ratings percentage index still has many deficiencies that reward and penalize certain teams unfairly.

The Deficiency of RPI

As has been hinted at multiple times, there is a serious RPI deficiency. It isn’t a very good system. In fact, its main deficiency is that which it tried to solve in the first place: it does not correctly measure and incorporate a team’s strength of schedule.

The shortest description of what is wrong with it is as follows. RPI overly credits teams that play hard schedules and overly punishes teams that don’t. Nearly 75% of a teams ranking via this metric is based on other team’s records.

That is, only 25% of a team’s rating is based on your own performance. Even worse, the more you win, the worse your opponents’ record will be! That means that while a win will increase your winning percentage, it will decrease your opponents’ winning percentage.

So what would make a better system? I always like to say that you shouldn’t get credit for playing good teams, you should get credit for beating good teams. You should even get credit for losing to really good teams by small margins. RPI doesn’t do this. It only looks at strength of schedule and record totally independently. In the next section we’ll see an example of the RPI deficiency in action.

An Example of RPI Deficiency

There is a classic example of how this stat can go wrong. In this example, 2 teams play each other 3 times to start the season so that those are the only games they’ve played. This could easily describe a three game baseball series. Suppose team A wins all 3 games. Let’s look at how each team’s RPI stacks up in this example.

Team A’s record is 100% and their opponents winning percentage is 0%. In this example, team A’s opponent’s opponent’s is … team A! Therefore, the OOWP number in the formula above is 100%.

Do the same exercise for Team B and we get WP = 0%, OWP= 50%, OOWP = 0%. If you go through the formulas, then each team has the exact same RPI. That means that team A can beat team B every single time they play but RPI can’t figure out which team is better!

Team A		Team B
100%	WP	0%
0%	OWP	100%
100%	OOWP	0%
50%	RPI	50%

Obviously this example is a bit contrived, but it does show how RPI can be flawed.

The Core Problem with RPI

While the previous example highlighted one deficiency of RPI, it was only a symptom of a larger problem. RPI places far too much emphasis on the strength of schedule and too little emphasis on a team’s performance against that schedule.

Consider an extreme case where team A plays only the top 10 teams in the league while team B plays only the bottom 10 teams in the league. For team A, their opponents’ winning percentage will be nearly 100% while for team B, OWP will be nearly 0%.

This means that team B’s RPI will max out at 50% while team A’s RPI will bottom out at 50%. No matter what team A or team B does with their schedule, team A will be judged the better team because they played a much harder strength of schedule. This can’t possibly be a good measure of team quality.

Though this may feel like another contrived example, consider how RPI will treat the worst team in the best conference versus the best team in the worst conference. Just by being in the best conference, a team will get a massive boost in RPI without having to do anything against their schedule.

In the RPI system, team’s don’t control their own destiny. They could win every game by historical margins but if their strength of schedule is weak, there is nothing that can be done to overcome this fact. A team could lose every game by historical margins, but if their strength of schedule is strong, then RPI will still look upon them rather favorably.

This is one of the reasons that the NCAA now uses NET ratings which explicitly take into account margin of victory. Now, even if a team plays a very easy schedule, they can achieve high NET ratings if they really dominate against their schedule. Teams with easy schedules can still convince the committee they are worthy with good performances. They still control their own destiny. And, at the end of the day, that is an important thing.

To receive email updates when new articles are posted, use the subscription form below!