Understanding the RPI, 201: Double Dipping

Happy 2017, everyone, and welcome back to this RPI case study. Considering the winter break and all the festivities that surely came with it (including FSU picking up a big time RPI win at UVA), let’s recap the lessons covered so far.

In Understanding the RPI, 101 we discovered that:

A) Half of a team’s RPI is determined by its opponents’ winning percentage

2) Shakespeare would not like the RPI

D) Losing a game can be better than winning a game by 42

Garnet) Playing a crappy team that’s in an even crappier conference is better than playing a crappy team in a really good conference

Understanding the RPI, 102 taught us about the “Real Estate” rule:

1) Wins on your home court, even against elite teams, don’t count for much

B) Wins on the road, even against crappy teams, really add up

4) A house of cards (not you, Kevin) can be created by scheduling more road games

Gold) UF’s ill-timed construction plans now look diabolically genius

Alrighty then. Now that everyone is caught up, let’s dive in. With this being a 200-level course, we’re really starting to get into the weeds.

What’s Good for the Goose is Good for the Gander

If you recall back to a portion of URPI 102, we discussed the house of cards that was set up by the enterprising Missouri Valley Conference during the 2005 and 2006 seasons. In short, individual schools increased their RPIs by scheduling more OOC road games. Come conference play, the value of beating a conference opponent increased because now they were accumulating prized top-100 wins as opposed to sub-150 wins. Simultaneously, conference upsets—which occur in every league every season—were now less damaging to one’s resume. This house of cards resulted in the MVC receiving as many bids as leagues like the ACC and Big 12.

What was good for the goose (increasing one’s own RPI by scheduling more road games) was also good for the gander.

Unsurprisingly, this isn’t the only way a conference can collectively work together to increase their potential bids. Alternative methods include:

• Being really, really good and deep. If every conference member went undefeated in OOC play, the collective “opponents’ winning percentage” would skyrocket. This method is, of course, much easier said than done—though the ACC seems to be putting itself in position to give it a go one of these years.
• Double dipping on the best RPI opponents. (Notice I said best “RPI opponents,” which if you’ve learned anything from the first two URPIs, you know that doesn’t always equal the top teams in the nation). This second method is worthy of a deeper look.

Double Dipping

The third, and often less discussed, factor making up a team’s RPI is their opponents’ opponents’ winning percentage, which accounts for 25% of a school’s RPI. This means that how the teams played by your opponents fare on the season collectively accounts for the same percentage of a school’s RPI as that school’s own win-loss record (or at least, the RPI-land’s modified win-loss record). Crazy, I know.

Of course, no one school accounts for 25%. Some opponents’ opponents’ are only played by one school in your six degrees of RPI separation and account for less than half a percent. But...what if you want a school to count for more? Then the conference should make like George Costanza and double dip.

Let’s back up a second.

Before the start of every season, FSU (and nearly any other school) already knows a few things about their schedule. FSU will face 18—soon to be 20—ACC opponents, some of them twice. Playing a team twice means they are counted twice in your “opponents’ winning percentage.” Basically, if you play Team A twice, and Team A finishes 20-10, 40 wins and 20 losses have been added to your opponents’ grand total.

The same concept is true for opponents’ opponents. If multiple of your opponents play the same team, then that team counts multiple times when factoring in your opponents’ opponents. Still with me? Let’s go back to our friends at the handy-dandy RPIforecast for some tangible examples.

The link above takes you to FSU’s profile page, which features lots of cool stuff. For our purposes here, scroll down to the bottom quarter of the page. There you’ll see a breakdown of every single team that currently factors into or will factor into the Seminoles’ RPI. The list is ordered based on the weight at the conclusion of the regular season, from most to least.

Obviously, FSU has the single biggest impact on FSU’s own RPI (the RPI at least got that right). Next up on the list, you’ll find the four teams FSU plays twice in league play this year (Miami, Duke, Clemson, and Notre Dame). But if you look closely, you’ll see that Miami’s future weight is slightly higher than the other three, while Duke’s is less than Miami, but slightly higher than the next two. The reason for this is double dipping.

Miami and FSU shared two OOC opponents this season (George Washington and UF) and both also play Duke twice. Basically, FSU and Miami have more overlap in their schedules than any other FSU opponent. Duke also played UF and plays Miami twice, making Duke the second most connected opponent of FSU.

Scroll down beyond all the ACC teams and you’ll find FSU’s most connected OOC opponents. The results are predictable. Manhattan, Winthrop, and Illinois all took part in the NIT Tip-Off tournament with FSU and the teams essentially played a round-robin. UF is fourth due to their games noted above against Miami and Duke.

Double Edged Sword

The MVC House of Cards was an example of individual teams raising their own RPIs via smart scheduling, which in turn helped the conference collectively by creating more resume-boosting opportunities. Conversely, double dipping is an example of conference teams collectively influencing all of the RPIs in that conference via opponents’ opponents’ winning percentage. Said another way, double dipping isn’t just impacting resume-building opportunities, it’s also directly impacting the collective RPIs.

Of course, this can be double edged sword. If 8 ACC teams had played UF this season, UF would have a larger impact on the RPI of all the teams in the ACC. Which would be good because UF, a good team in a weak high-major conference, is likely to have a fairly high winning percentage (let’s leave aside the probability that UF would have potentially lost 3 or 4 of those ACC games...). However, if the ACC quadruple dipped on Southern Miss this season, a team that might win as few as one D1 game this year would shoot up the RPI impact scale. Not good.

Now, some might argue that the difference between an opponent weighted at 1.75 vs 1.61 is small on the grand scale of things. And this argument would have a point. It is small. But so too is the margin of being the last team selected for the NCAAT and the first team to have its bubble burst. Just ask Seth Greenberg.

In the final installment of this series, we’ll bring it all together and look at ideal scheduling for the RPI.