clock menu more-arrow no yes

Filed under:

Penn State Basketball: How Pat Chambers Hedged Penn State's RPI

New, comments

Don't confuse discrediting Penn State's non-conference schedule as cause for the team's 12-1 record with discrediting Penn State's non-conference schedule because it will presumably undermine their NCAA chances.

Rich Barnes-USA TODAY Sports

Not many people are fooled by Penn State's 12-1 non-conference record. All they see is the pedestrian KenPom ranking (#87), the lack of marquee wins (only George Washington at home) and the close margin of victories versus subpar foes (only a 0.10 efficiency margin). Those people need to see more before taking Penn State seriously, and it's impossible to argue that's not a fair judgment. Even the most optimistic of Penn State observers know this team has to prove its mettle against the Big Ten before any postseason scenarios can be taken seriously.

But this article is not about forecasting how many Big Ten wins Penn State will have on March 9th, nor does it try to quantify the probability Penn State will make the NCAA Tournament as of today. All of that will sort itself out over the next 10 weeks. What we're trying to point out here is how important Penn State's 12-1 non-conference will be if Penn State is able to win nine Big Ten games and put themselves right in the bubble mix at 21-10.

The biggest hurdle for this program has always been making the NCAA tournament, because an 18-game Big Ten gauntlet must be managed in order to have a shot. No power conference team can make the NCAAs based solely on their non-conference résumé. If a team's 7-11 or 8-10 in one of the major leagues, past bracketology suggests they don't have a legitimate chance without accumulating some more wins in the conference tourney.

However, the Nittany Lions have been on the other side of the coin before, when a putrid non-conference schedule prevented a rare chance at national glory. Everyone fondly remembers the discussion around the 2009 team, one that earned an 11-9 Big Ten record but was ultimately left out because of its 70th-ranked RPI rating and 307th-ranked NCSOS. They also could've just defeated a Todd Lickliter-coached Iowa team in the regular season finale as well, but why force yourself to do more work than necessary in the rugged Big Ten?

Penn State is the kind of program that has to take its scheduling very seriously, because it needs every break possible to qualify for the sport's pinnacle event. If they are able to sneak a few extra wins in the Big Ten any given year, it's imperative to parlay the overachievement into an NCAA bid for the obvious boost it could provide in recruiting and program perception.

Some may be concerned when parsing over the profiles of Penn State's 2014 non-conference opponents when they discover that seven of the 13 are expected to finish with RPIs rated 221 and above, per RPIForecast. That valuable website also projects Penn State's non-conference SOS at a dangerous 286 (and rising), not far off from last year's 295 rating of SMU's schedule. The Mustangs were the most recent annual example of how a weak non-conference could keep you out of the NCAA tournament.

But before the schedule is decried for its lack of quality opponents, it's worth understanding how the RPI works and why Penn State's overall RPI rating will not be jeopardized by its paltry NCSOS. Pat Chambers -- along with Ross Condon, the team's director of basketball operations and the man who put the schedule together -- has shown an admirable understanding of the RPI system and has managed to give Penn State its best chance at the real postseason for the first time since 2001.

THE RATINGS PERCENTAGE INDEX

The only reason anyone cares about this flawed system is because the NCAA Selection Committee swears by it when determining their at-large selections for March Madness. All common criteria when reviewing a team's profile are determined by the RPI system: top-50 wins, top-100 wins, 200+ losses, SOS, NCSOS, etc. There's not one thing your KenPom rating can do for you on Selection Sunday.

So why is the RPI system flawed? Not only is it easily manipulated, but the equation has no way of accounting for a team's margin of victory in its ratings. The RPI thinks it is irrelevant to know how much a team wins by when evaluating the quality of teams. A 2-point home win over Duquesne would have the same impact on a team's overall RPI as a 35-point home win over Wisconsin. It may have been the right system to implement back in 1981 when point shaving scandals were grabbing headlines, but it's proved its limitations time and time again.

Before one can understand how the rating can be manipulated, they must understand the formula from which these ratings are derived. The RPI rating is calculated from three components: a team's adjusted winning percentage (AWP), a team's opponents' winning percentage (OWP), and a team's opponents' opponents' winning percentage (OOWP). The overall RPI rating is weighted as 25% AWP, 50% OWP and 25% OOWP. Meanwhile, a team's SOS is calculated only using their OWP and OOWP at a 66% OWP and 33% OOWP weight. Looking at the big picture, the only statistical factor that drives the RPI model is simply the number of wins by a team and its opponents.

HOW TO EXPLOIT THE RPI

There are ways to manipulate your potential RPI rating with the most common practice being Luke Winn's described "scheduleball". Winn explains one way to inflate your RPI is to schedule beatable teams that are expected to win in mid-to-low level leagues. While everyone in the world can accept that not all 22-11 teams are equal, the RPI says they are. Playing a 22-11 team like Akron will have the same impact on your OWP as playing a 22-11 Syracuse, even though many would argue there's a clear difference in difficulty between those two programs.

Penn State implemented this strategy in 2011, when DeChellis rigged his schedule with quality RPI opponents like Fairfield and Lehigh. The Lions didn't do well enough in the non-conference after finishing just 7-4, but their schedule strength was deceptively great by the RPI computers. Their 11-game slate was ranked 58th in NCSOS, but just 178th in KenPom's Pythagorean SOS. If Penn State was able to go 10-1 or 9-2 as opposed to 7-4 with the bad loss to Maine, the Lions would've been a lock for the NCAAs at 9-9 without the three wins that team needed in the Big Ten tourney.

But this is not a fool-proof strategy, as you're still projecting which teams will be good any given year. Penn State's 2014 non-conference schedule was never expected to be overly challenging, but it also wasn't expected to potentially be ranked 272nd either. Some of the Lions' foes like Fordham, Duquesne, Bucknell, Morgan State and Drexel were not expected to be this bad just two months ago. Jon Severe has been a complete non-factor for Fordham after leaving the team. Duquesne was expected to move up in the A-10 in Jim Ferry's third year, but that clearly hasn't happened. Bucknell and Morgan State have their worst teams in years, while Drexel already lost two starters to injuries before the season began. Even Akron has lost its best player from the program, as well.

These unforeseen events for Penn State's opponents clearly impact their potential winning percentage, and if Pat Chambers knew those teams would be facing those challenges, he probably wouldn't have scheduled them. However, there's no perfect methodology to forecasting which teams will finish with winning marks any given year. Therefore, it's important to have safeguards in place in the event your opponents do tank like Penn State's have so far this season.

To understand how Chambers smartly scheduled Penn State into potential NCAA consideration, we break down each three components of the overall RPI rating.

ADJUSTED WINNING PERCENTAGE

RPI Weight: 25%
Potential Range: 0.0000 (winless) - 1.0000 (undefeated)
Hows It's Calculated: Only D1 games count. Home win = 0.6, Neutral win = 1.0, Road win = 1.4 and vice versa for losses
How It Can Be Inflated: Schedule as many winnable road/neutral games as possible

Here's where Pat Chambers hedged Penn State's RPI all along. The Lions played six of their 13 games away from the BJC this year, tied for the most combined road/neutral non-conference games with Wisconsin, Illinois and Michigan State in the Big Ten. While the PR spin for games like Drexel at the PPL center and Bucknell in Lewisburg revolved around the benefit for the fan base, don't think for a second there wasn't an RPI strategy behind it, as well.

With the weighted factor for road wins versus home wins, any team will earn more in their adjusted win column from winning at Marshall (worth 1.4 wins in this metric) than they would have by defeating both Arizona and Kentucky at home (worth 1.2 wins in this metric). That's how valuable road wins are to this part of the RPI equation. Penn State was able to inflate their AWP with road wins at Marshall and Bucknell and their 3-1 neutral court record.

How much of an advantage did the 12-1 record with 5 road/neutral victories and no home losses give Penn State? Let's compare to the rest of the bubble teams currently in the Big Ten, based solely off non-conference performance:

Final AWP Projection

I projected what each team's AWP would be assuming they finish with a bubblicious 9-9 conference record including 6 home wins and 3 road wins. Since everyone's conference slate is balanced with the same number of road/home games (9), the only way to make up ground in this metric is to win 10 or more conference games. Those teams are almost always locks for the tournament anyway, so we are not concerned about them in the bubble discussion.

The column titled "PSU RPI Adv" signifies the overall RPI advantage Penn State would have over their conference brethren from this weighted factor, if they were all to finish with identical .500 conference records. While it's important to keep in mind all of these teams will make up some ground on Penn State's overall RPI with their advantage in SOS, it is worth understanding that a 0.0136 overall difference in the RPI usually accounts for at least 10 spots in the overall standings.  For example, Pittsburgh last year finished 39 in the RPI with a rating of 0.5895, while Kansas State finished 51st with a rating of 0.5772. That is a difference of 0.0123 in their RPI.

Penn State conceivably has a 10-30 spot advantage in the RPI over all of these Big Ten bubble teams based on their inflated AWP that was caused by smart scheduling and 12-1 execution. To further prove this is part of Penn State's scheduling formula, they are already booked for three road games in 2015 against Duquesne, Drexel and George Washington, in addition to another likely road game in the ACC challenge. Throw in a couple of neutral court affairs from whatever multi-team event they participate in, and Penn State once again has plenty of AWP potential in next year's non-conference slate.

OPPONENTS' WINNING PERCENTAGE

RPI Weight: 50%
Potential Range: 0.3500 - 0.6500
Hows It's Calculated: Your opponents raw winning percentage against D1 teams, minus the results of the games played versus team in question
How It Can Be Inflated: Schedule teams that are expected to have winning records in mid-level leagues

Here's where scheduleball can have its biggest effect, but it's also important to realize how much playing in a top league like the Big Ten will boost PSU's overall SOS. While there's a significant range of potential OWP values across basketball, Penn State's potential range any given year is much smaller due to its power conference boost. Even with the infamous 2009 schedule, Penn State's OWP was still good nationally at 0.5229 (ranked 118th). That is why you will still see the worst major conference teams remain in the top two-thirds of the overall RPI.

However, we're talking about bubble competition and Penn State's mark was not even close to some other major conference teams on the bubble in 2009. One of the last teams selected for the field that year was Minnesota. The Gophers held about a +0.0400 advantage in OWP over PSU. Considering Penn State played in the same conference, Minnesota sure didn't earn that significant edge with their Big Ten slate. The sadder thing is the Gophers' NCSOS ranked just 187th that season, far from anything remotely challenging.

When that overall margin factored into the RPI rating at a 50% weight, the Gophers held a +0.0200 advantage in overall RPI. The Lions did not have as good of an AWP that year, even with their sparkling 22-11 record, and Minnesota's 28-spot advantage in the RPI was ultimately what separated them from Penn State in the committee's eyes.

Unfortunately, Penn State will be giving up ground here to plenty of other potential bubble teams, but it's complicated to tell exactly how much until the conference picture comes into clearer focus (when we get to find out if any of this is worth it, anyway). The Lions don't seem to have the massive RPI anchor like they did with 1-30 NJIT in 2009, so at least that's a positive. It's also worth pointing out that Penn State will have one of the toughest rated Big Ten schedules, since four of the top five teams in the expected RPI projections will face the Lions twice this year. More on this shortly.

OPPONENTS' OPPONENTS' WINNING PERCENTAGE

RPI Weight: 25%
Potential Range: 0.4400 - 0.5600
Hows It's Calculated: The average of your opponents' OPW against D1 teams.
How It Can Be Inflated: Have a winning season and avoid scheduling teams with weak schedules. Since you don't know other team's schedules ahead of time, just avoid teams from the weakest leagues.

This weighted factor is almost irrelevant in the RPI picture. While the notion to avoid teams from weak leagues is definitely true, this factor is far more forgivable as there is limited variance between major conference teams. Thanks to Jerry Palm's archives at the now defunct CollegeRPI.com, we can look at the Big Ten's final OOWP for the 2013 season to illustrate its limited RPI impact.

Final Big Ten 2013 OOWP

I included San Diego State (1st) and Grambling State (last) from that year to show the maximum RPI advantage that can be gained (roughly 0.0300 any given year). That's fairly sizable, but the overall RPI difference between power conference teams on the bubble is much closer to a range of 0.0025-0.0100, as seen with the Big Ten's distribution. That's only about a 2-5 spot difference in the overall RPI standings.

However, OOWP is weighted as 33% of your overall SOS, so it is important to keep this metric in mind since the committee values those rankings, too. Where the philosophy has its biggest effect is on NCSOS, when the security of the Big Ten boost can not be applied.

Since Penn State is the most common opponent on all of their opponents' schedules (this confusing concept is explained well here on Wikipedia), the Nittany Lions helped themselves already by going 12-1. They will also help themselves after avoiding some of the real dregs of college hoops from leagues like the Big Sky, SWAC, Southland, Ohio Valley and NEC. There was only one bad game on this year's schedule in violation of this guideline, as Morgan State hails from the MEAC.

Thanks to RPIForecast, we can see the impact all Penn State's opponents' opponents will have on their RPI at the end of the season. When you deliberately schedule a MEAC school, you're guaranteeing the rest of the teams in basketball's 32nd-best league will get to impact your RPI (since Morgan State will obviously play all MEAC teams in their conference schedule). That's not wise, but sometimes it's unavoidable when trying to fill late game slots. Nearly everyone played opening night, so Penn State might not have had a better option for November 14th.

Since there are no other bottom-league opponents on Penn State's schedule, Chambers limited the effect college basketball's worst teams could have on their schedule, even if their faltering opponents' WP continues to tumble as they play more games.

Unfortunately the rest of the Big Ten can get away with playing dogs like UMBC, Nicholls State and Grambling State, since many of them are included in primetime ESPN matchups against some of the blue bloods of hoops. While NJIT was the '09 anchor, teams like Louisville and Kentucky pack tremendous computer juice for anyone's SOS. Do you think Minnesota benefitted from playing pop's Louisville Cardinals? The Gophers share the same number of dead weight teams as Penn State, but they are still projected to have an NCSOS 100 spots better. Yes, that difference can be attributed to just one game against Louisville. Another example is Ohio State, who went out of their pay to host cupcakes all non-conference season (including the 4-game 'Buckeye' classic). But since Thad's program can easily land two games against UNC and Louisville through ESPN, he is afforded the luxury to not have to care about anything else RPI-related when constructing his early season competition.

Penn State can't land those games. Kentucky-Penn State doesn't drive TV ratings and that's why you'll rarely see the Nittany Lions in Maui with Kansas and UCLA (the 2011 Hall-of-Fame Tipoff was an exception). Nor will you see Penn State and North Carolina in the B1G/ACC Challenge. While I haven't hit on it in this post, obviously the financials of these games tremendously impact scheduling, as well.

Penn State's Potential Résumé That Could Break the RPI

Everyone wants to know how many wins it's going to take for Penn State to make the NCAA tournament, and that's a hard question to answer. There's a lot to love with Penn State's potential computer profile at 9-9 but also a lot to hate. While the first red flag will come from their lousy NCSOS, a bigger concern for the Nittany Lions could be a lack of quality wins.

Generally speaking, almost all power conference teams are locks with a sub-50 RPI, especially now that the tournament expanded to 68. But thanks to the Big Ten's struggles in non-conference play, Penn State could put that rule of thumb to the test. While the RPI holds great weight with the Selection Committee, so does top-50 wins and clear evidence that the team in question can beat other teams in the field.

However, Penn State's opportunities for such scalps in conference play took a hit these past two months when teams like Michigan and Nebraska played their way out of the RPI top-50 for good, barring a miraculous turnaround. If Penn State was to somehow manage to scrape together nine conference wins, there's a decent chance a scenario exists where PSU could have only two or three top-50 home wins to show for it. I'm not a bracketology expert, but that seems unusual for a team that could end up with a solid 21-10 record out of the Big Ten (and a typically safe 44 expected RPI, per RPIForecast).

This concern would be even more dire if they had Illinois' conference draw. The Illini only play Wisconsin, Maryland and Ohio State once, and that has caused their expected SOS (70) to fall below Penn State's (64), despite their 100-spot expected edge in NCSOS. Penn State at least has eight chances against the top-4 teams in the league from an RPI standpoint (Wisconsin, Ohio State, Maryland, Minnesota). It would certainly help their case if the Nittany Lions are able to snag two wins out of that group.

But if they don't fare well in those eight games and their collection of nine wins include let's say Rutgers (2x), @ Northwestern, Michigan, Purdue, Nebraska, Iowa, @ Illinois and Minnesota, it's hard to believe that will be enough without any truly impressive wins.

The other important factor to consider is the new format for the Big Ten Tournament. If Penn State is sitting at 9-9, they're likely going to be seeded around 6-8 for the conference tourney. That would earn them a first round bye and a matchup with one of the bottom six teams in the league, the kind of late-season matchup any bubble team must have to avoid yet another red flag to the committee. Therefore, I'm hesitant to say 9 Big Ten wins will be enough, as the bubble landscape if Penn State's really on it will need to be taken into account. But 10 wins (post-BTT) should be enough in the RPI for Penn State to be safe. That's much better than 2009, who was left out with 11 Big Ten wins and 2011, who needed 12 to overcome a poor non-conference performance.

There is part of me that would love to see what the committee would do with a 21-11 (9-10) Penn State, however. Especially if they hold a 7-10 RPI spot advantage over the Iowas and Indianas that would likely have better wins and better KenPom profiles. If PSU is passed over with an RPI ranking on the good side of 50 for some more qualified major conference teams, maybe college basketball can finally de-emphasize the RPI, or better yet, fire it into the sun for good.

We can only hope Penn State will defy the KenPom odds once again set before them heading into conference play, much like the 2009 and 2011 team did. Then maybe all of these interesting scenarios can play out and Penn State can truly challenge the validity of the RPI. But if not, Penn State still has a sterling NIT profile, one that would keep them in consideration at 6-12. Either way, Penn State is positioned beautifully to improve on last season's forgettable CBI experience.