• Buying Better Bakery: An Analysis of Scheduling Cupcakes, Part I

    Last year, Marquette returned to winning ways by posting a 20-13 record, however they remained below 100 in the ever-important RPI ranking that trumps all come Selection Sunday. Of all the teams in the top six leagues, the only team that won 20 games and finished with a lower RPI than Marquette's 110 was Northwestern. Not surprisingly, the Wildcats had a slightly weaker non-conference SOS (Ranked 326) than Marquette (322) did.

    Why did this happen? The reason is twofold. First, of the seven buy games that Marquette scheduled, only Jackson State finished with a record over .500. The other six teams all not only finished below .500, they also finished with single-digit wins. In Division I games, those six teams (Grambling, Maine, San Jose State, Chicago State, Presbyterian, Stetson) finished with a combined record of 35-138. As the first factor in RPI is opponent's winning percentage, this is a killer.

    The second factor in calculating RPI is opponent's opponents record. The majority of any team's games will be against teams from their own conference, so if they play in a better mid-major, you are less likely to suffer a negative impact for playing them. Getting San Jose State from the Mountain West, despite their 6-22 record, actually has some value because they will benefit in that second factor. The problem is despite getting one team from a top-half league, the other six opponents were all from leagues rated 23 or worse out of 32.

    So how should we analyze the proverbial cupcakes? I've come up with some criteria that I hope to apply to a variety of different teams to analyze how worthwhile they are to play. There are four positive and four negative criteria. I'm sure that some of the criteria will seem somewhat subjective, so I am going to take as much subjectivity out of it as I can. Here are the criteria:

    Positive Criteria

    Track Record of Success: This criteria looks at the past records of teams, with the emphasis placed most heavily on recent conference games. Teams that consistently win are more likely to help RPI and schedule strength, while teams that lose consistently do the opposite.

    League Strength: This is criteria looks at how good the rest of the league is, going back to that opponent's opponents factor. The better a team's league, the more likely they come back as a positive on the resume come March.

    Coach Tenure/Success: This looks at the coach's record, his tenure at his current job, and his total coaching experience. Coaches that don't have success typically don't last.

    Non-Conference Scheduling: This will look both at how well the team tends to perform in non-conference games as well as how difficult they usually schedule. Teams that win more and play tougher opposition are the ones that will give the most valuable scalps.

    Negative Criteria

    Buy Game Performance: Playing good teams can be beneficial, but can also be risky due to the risk of an upset. This looks at the likelihood a team has of keeping the game close or even pulling off a high-major upset.

    Mercurial Star: Every now and then, one player can put up a Herculean effort that leads to a win. Not many smaller schools have individuals like that, but when they do, they can turn an expected loss into a surprising win.

    Recent Tournament Play: Teams that are accustomed to postseason play are less likely to be rattled by a high-major environment. This will look at how often a team has played in invitational postseason tournaments in the past three years.

    Roster Experience and Continuity: Non-conference games are usually early in the season, which gives a benefit to teams that have experienced players that are used to playing together.

    The Math

    TROS: League winning percentage, past 3 years + Total Winning percentage, past 3 years + Total Winning percentage, past 6 years

    Example: Belmont had a 37-11 (0.771) conference record the past 3 years, a 68-33 (0.673) total record the past 3 years, and a 151-53 (0.740) record the past 6 years for a total TROS score of 2.184.

    LS: Adjusted League Rating from past 3 seasons added

    Example: The Atlantic Sun had a Rating of 0.2942 in 2014, a rating of 0.2562 in 2015, and a rating of 0.3168 in 2016 for a total LS score of 1.173.

    CTS: 0.1 x coach's years at current job (max 0.5) + 0.1 x coach's 2 years coaching (max 0.5) + coach's total winning percentage

    Example: Dan Monson gets 0.5 points for his 9 years at Long Beach State, 0.5 points for his 19 year career, and 0.553 for his 325-263 career record for a total CTS score of 1.553.

    NCS: Non-conference winning percentage over the past 5 years + average kenpom.com non-conference SOS rating the past 5 years

    Example: IUPUI went 18-49 (0.269) during the past 5 years with an average non-conference SOS rating of 0.545, giving them a total NCS score of 0.814.

    BGP: (4 x high-major wins) + (2 x high-major <3 point losses) + (1 x high-major 4-6 point losses) / Total number of high-major road games

    Example: This looks at the past 5 years. Vermont had zero high-major wins, 2 high-major single-possession games, and zero high-major two-possession games in 7 attempts. (0+4+0)/7 = 0.571.

    MS: Only for top-340 Value-Add Basketball players. 0.1 x VAB rating + VAB NBA%.

    Example: James Webb of Boise State has a rating of (0.1 x 6.58) and 0.47 NBA percentage for a total MS score of 1.128.

    RTP: (0.5 x Number of NCAA games past 3 years) + (0.25 x Number of NIT games past 3 years)

    Example: Texas Southern has played 2 NCAA games and 1 NIT game for a total RTP score of 1.25.

    REC: Returning minutes per game average of all players for each of the past 3 seasons added together / 200

    Example:

    Cupcake Types

    Hostess: (-2 or lower) These are the opponents that are bad and bad for you. Few teams will be in a situation where they are both bad enough to have a low positive rating yet experienced and talented enough to pose a threat. Hostess are simply the worst cupcakes. They may go down easy, but they should be avoided at all costs.

    Piggly Wiggly: (-1 to +1) Your local grocery store may have a bakery department, but it's not their specialty. Just about every schedule will have a few Pigs, but they're not where you go to spend your premium bakery dollar.

    Panera: (+2 to +4) This is where we start getting to the quality opponents. Bakery is what they do, and they do it pretty well. It may not be premium, but if you make Panera cupcakes the bulk of your schedule, you're doing pretty well.

    Grebe's: (+5 or higher) The Grebe cupcake is your local specialty bakery that does exactly what you want and gives you that extra care. They are the high-value, low-risk cupcake. Few teams will likely meet this value, and they probably are similarly valuable to playing high-major opponents.