As the early season unfolds, many fans of college hockey get very interested in the Rating Percentage Index, or RPI, the system that’s used to determine at-large bids for the NCAA Division I Hockey Tournament. Many of those same fans actually pay more attention to the PWR, or Pair Wise Ranking, which is based off RPI but with a couple more layers.
Of course, playoff winners from each of the six major conferences get automatic bids. The other spots in the Big Skate, as well as the seeding of teams in the tournament, are determined by the RPI value for each team. Since many of the playoff winners also have high ranking in RPI, the number of at-large berths is never the same, and the cut-off point in RPI for an at-large slot is usually a last-weekend-of-the-year nail biter.
Seeding does get adjusted to avoid regional games between conference rivals and also to minimize travel costs. And there are some rules, never used as far as I can tell, that can break ties, such as relative conference strengths.
Since the playoffs are always a bit of a crap shoot for any team—it isn’t a certainty that the league regular season champion is also the playoff champ—serious fans tend to fret a great deal about how teams can earn a slot in the Tournament.
The RPI is a system that uses head-to-head results to calculate an index score. There are four factors that determine RPI.
- Each team’s winning percentage counts as 25% of its Rankings Percentage Index.
- Opponent records count, too: 21 % of your opponents winning percentages are also added in.
- Finally, the remaining 54% is based on the winning percentage of your opponents’ competition. As part of this, the contribution of each individual game is weighted by a factor of 1.2 for a road win or home loss and 0.8 for a home win or road loss.
- There is also a “quality wins bonus” for wins against teams in the top 20. Also, if a team loses in RPI because they beat a low rated opponent, that game is ignored, eliminating a potential punishment for a win.
So let’s look at the Huskies’ RPI after four games. It’s not that hard early in the season, but as the games pile up, you really need a computer model (a sophisticated spreadsheet will do) to compute RPI. Bear in mind that the actual math will be a bit different than what I’m going to show you. That’s because RPI is a dynamic model that changes after each and every result is entered. Getting to that level requires a full model, and The Old Dog, fan that I am, isn’t that consumed with RPI to try and develop that kind of thing.
With that admission, let’s look at an approximate set of calculations. To start, the Huskies’ winning percentage is 0.500, so they get 0.25 X 0.50 or 0.125 for the first criteria. That one’s easy.
Currently, Robert Morris has a 3-3 record. Tech gets 0.21 X 0.50 for 0.105. The same is true for Alaska—also 3-3 through the first three weeks. That gives Tech a total of 0.125 + 0.105 + 0.105 of 0.335. This is also a straightforward calculation.
The third criterion brings additional complication. Besides their two losses to MTU, RMU swept Bentley and split with Army. Tech, of course, stands at 2-2. Bentley is 1-2, so they sit at .333, while Army is 3-1 for a 0.75 win percentage. Alaska’s opponents include Tech, Denver at 6-0 and Penn State at 5-1. While the math gets quite a bit more complex for this criterion due to the home-and-away adjustments, the net result is that Tech picks up an additional 0.345 from the strength of schedule that Alaska (mostly) and RMU (partly) have played.
Tech hasn’t picked up any bonus points for “quality wins.” Nor have they had any adjustments for beating a low ranked foe.
This actually puts Tech in the 8th slot in the nation with an RPI of 0.6800. But don’t get carried away. Northern Michigan is 2-1-1 and is ranked 23rd. And Northern has played Boston University and Michigan State so far, not exactly a pushover schedule. Much as I’d prefer to smack the Wildcats for the RPI discrepancy, that’s just not fair. There just aren’t enough games for any team this early in the season to make the RPI a reasonable assessment tool. Minnesota-Duluth, after getting swept on the road by Wisconsin, is ranked 49th out of the 54 teams that have played any games. (Six ECAC teams haven’t played yet.) I don’t believe for a nanosecond that the maroon Bulldogs (as opposed to the red Bulldogs from Big Rapids) are grossly inferior to the Huskies, no matter how much I’d love to say that.
The truth is that RPI doesn’t really start to make sense until a good portion of the season has been played. After each week’s games, all of the factors change depending on who does what to whom and where it gets done.
Is the RPI really a good way to do this? With just sixteen slots for sixty teams, that means 73% of Division I schools don’t get into a tournament. In NCAA basketball, there are 351 D1 teams, according to several web sites—but only about 120 of those schools would ever realistically have a shot at the tournament, and 68 slots exist. That’s more than half who can never really hope to get in due to the conference they are in and the schedule they play. Another quarter won’t make it due to a weak schedule or mediocre record. So, about 80% of basketball programs don’t make the Big Dance.
Of course, only four teams (out of 129 teams) make the Football Bowl Series playoffs. Sixteen of 125 Football Championship Series (the old Division I-AA) teams make the playoffs. Finally, about 79% of DI baseball teams don’t make the 64 team NCAA Tournament. As these numbers demonstrate, hockey isn’t unusual in how the Tournament is structured, so it would be hard to make a case for expanding the size of the field.
At the same time, RPI does place a great deal of importance on non-conference games. To start, unless a league does well out of conference, the value of winning and even of losing is reduced greatly once conference play starts. Over the past few seasons, the NCHC has stormed through their non-conference opponents, which has increased the value of every NCHC conference win. The opposite has been true for WCHA and Atlantic Hockey teams.
For the Old Dog’s taste, that is one of the weaknesses of using a pure “RPI rules all” approach. It’s not just that conference power has a big effect, it’s the fact that most, or even nearly all, non-conference games are played early in the year. Hockey is really different in that regard, as some teams get much better and some get much worse from the start to the end of the season. This is amplified because hockey teams get very limited formal, fully coached practice time before the season opens, far less than any other NCAA sport.
At the NHL level, we often see teams with low seeds either go deep into the playoffs or win the Stanley Cup. The LA Kings were seeded last in the 2012 Cup playoffs, but still hoisted Lord Stanley when the ice chips settled. Last season, the St. Louis Blues overcame a horrendous start to get a mid-level seed but still won the Cup at the end. In that same season, American International, which had never been in an NCAA tournament, managed to win the Atlantic Hockey playoff and grabbed the 16th seed—and then promptly upset #1 overall seed St. Cloud State in the first round. The Yellow Jackets started slowly but constantly improved throughout the season, and the NCAA tourney win proved how far they had come.
When the NCAA basketball tournament selections are made, RPI is used, but it is not used as a presumably objective tool the way it is in hockey. The selection committee factors in other ranking and rating schemes beyond RPI and they also put a great deal of weight on other factors, including the way a team has played in the weeks leading up to the tournament.
Would hockey be better or worse using more subjective tools to select teams? The Old Dog thinks it might be a two edged sword because there are tons of controversies in NCAA hockey history from the time before the current RPI system was adopted. At the same time, I’m not really comfortable with games in October having such a major impact. To me, one way around this would be for all of the conferences to schedule so that they have one “open weekend” for their teams in January and possibly February. They could then start conference play a bit earlier, and add some spice to the cold winter months leading up to the conference playoffs in March.
Still, when it comes to discussing NCAA tournaments, no matter what the sport, any speculation about how tournament teams are selected always creates controversy. But isn’t that something that sports fans love to argue about?