College Baseball RPI - Winning isn't everything, it's only 1/4 of the thing
- The RPI formula is based on a team's winning percentage, its opponents winning percentage, and its opponent's opponents winning percentage, which are weighted to determine their RPI score. Baseball weights its RPI as .25 for winning percentage, .50 for opponents' winning percentage, and .25 for the opponents' opponents' winning percentage. The weighted values of these three categories are then added together to obtain a raw RPI score. The teams are then rank ordered based on this score, a ranking that is imperfect in its current form.
So what's the problem?
There are many, but in a nutshell, 75% of the RPI is a measure of strength of schedule (your opponents' win percentage, and your opponents' opponents' winning percentage), which may or may not tell you the quality of a team. The RPI provides a strong measure of very good teams from better conferences. Its weakness is making meaningful distinctions between teams who are "on the bubble" come NCAA Tournament selection time.
Because the RPI emphasizes who you played over how you performed, weaker teams in the major conferences benefit while strong teams in weaker conferences appear less worthy of inclusion in the tournament.
Which brings us back to the original proposal - providing more value for a road win and less for a home win. How much difference will this make?
To some extent I agree with NCAA statistics guru Jim Wright who believes the change will make very little difference because, "if you're a good team, you're still winning the majority of those games anyway." This is indeed true.
What Wright doesn't add is that regardless of who wins, the difference will be small because the RPI understates the value of winning and over-emphasizes the record of your opponents. The weighting of the RPI is such that the majority of the benefits from this change cannot begin to overcome the three times greater emphasis on who you and your opponents played as compared to how you performed against them.
Under the current system, if you had an undefeated season, the maximum RPI value would be .2500. Under the proposed formula, an undefeated season played entirely on the road (a winning percentage of 1.250 multiplied by the same .25 weighting) results in a maximum value for win percentage of .3125 - a difference of .0625. But this "improvement" diminishes rapidly when applied in real life for a number of reasons -
- For most teams a highly respectable winning percentage is .650, not 1.000, A .650 winning percentage under the old system is equivalent to a .8125 if every game were to be played on the road. Under this scenario, the maximum possible improvement drops from .0625 to .0406. In reality, 1/3rd of the maximum adjustment from the proposal isn't obtainable.
- The benefit isn't significant enough to get teams to change their scheduling patterns. The same teams will continue to play at home, due to weather and economics (since unlike basketball, non-conference scheduling in baseball typically consists of three-game series played in February and March). For teams that play nearly all non-conference games at home against inferior competition (consider 2008 season examples Virginia - 21-4 non-conference, ZERO non-conference road games; and Duke - 24-0 non conference, 4 non-conference road games) the reduction in the value of a home win is irrelevant because they will still aim to enter conference play with a gaudy win percentage built against significantly weak or unprepared teams. When all conference members agree to do the same thing (which is clearly happening), the RPI benefits multiply. The RPI is built through institutionally controlled non-conference scheduling, and when conference play begins the most heavily weighted factor of the RPI - opponents' win percentage begins to compound because of the third criteria, opponents' opponents' win percentage.
- The home team winning 60% of the time is misleading. While this has been confirmed by the NCAA when looking at the 2008 season in total, it masks the early season domination that occurs year after year when the home field advantages are magnified due to weather and the travel demands that accompany teams playing 10-14 games in 10 days as part of a spring break trip when they haven't even practiced outside.
Greg VanZandt deserves credit for his proposal and its rational, well documented approach that would change an old saying to "Its not whether you win or lose, its where you play the game."
Further, this is a proposal that should be adopted by the NCAA Baseball Committee, because it uses well reasoned math and it would represent an acknowledgement of the inequity faced by teams due to geography while answering critics of the current RPI. But there is little chance this will significantly alter the RPI rankings. For it to have meaningful impact, greater emphasis on a team's actual winning percentage in the RPI is needed - in this case valuing a team's winning percentage as .33 of the calculation instead of .25, a shift of just 8 percentage points. The 8 percentage points can be taken from the weighting for the opponents' opponents' winning percentage, placing the majority of the emphasis in the RPI where it should be, on who you played and how you performed against them.
Skeptics will contend that this will simply encourage teams to play a weaker schedule in order to pad their win percentage and make the RPI less valuable. While that might be possible, could the "padding" that skeptics cite really be worse than what is currently occurring in the Duke and Virginia examples cited above or among many other teams with similar scheduling patterns?
Vince Lombardi said "winning isn't everything, it's the only thing".
Under the current RPI formula, "Winning isn't everything, it's only one-quarter of the thing." The proposal from Coach VanZandt presents a perfect time to analyze additional data and make another meaningful change that will value winning more - for the benefit of college baseball.