Month: January 2018

A recent post prompted an out-of-band exchange regarding scheduling games between classifications. The effect that scheduling itself has on rankings is a subject of interest to me. It is complex and requires a lot more time that I can devote during the season when, on some days, I spend five or more hours getting the previous day’s games into my database. But since I have that database I can ask it questions. It takes some time to formulate just how to ask, and understand just what that answer is telling me.

One question prompted by that exchange was just who schedules non-league games with teams in other classifications. League games are massively scheduled between teams in small subsets of the approximately 380 schools fielding varsity basketball teams. There really isn’t any point in using a ranking algorithm to determine who is the best team in a league when everybody plays everybody else one, two, even three times. That’s what the win-loss record is for. It is very easy to understand. The schedule is balanced, in most cases. The schedule for non-league games is not at all designed to even have the appearance of randomness. Thank goodness. I don’t want to see a random set of games selecting a 3A team to play a 1B team. I know that the only way Chief Kitsap girls will beat Bethel would be for the Tacoma Narrows bridge to collapse again while Bethel is driving across. The stat geek in me would find an experimentally designed schedule appealing. The fan in me would not. We get the non-league games that happen. They form the critical schedule structure for driving rankings of teams in different leagues and districts.

For starters, I looked at girls data for 2006-2007. That is the first season with the six current WIAA classifications. I did a lot of this by hand so hope I didn’t mis-transcribe. The inter-classification W-L table looks like

Looking at the yellow cells, where the classification difference is at least two, there are 81 games. The higher classification team won 47 and the lower classification team won 34. This season is also interesting for how badly 3A fared against adjacent classifications, but I didn’t look at those. Examining the 81 games was enough for me this week. Leave that for another day. Here is the list of the 81 games.

Thirty 4A teams scheduled down two or more classifications (maybe some teams count more than once) In so doing 4A went 14-16. Using 2006-2007 TeamBrunnhilde points rating, eight of these were in the top half of the classification, 22 in the bottom half, 16 in the bottom quarter. Clearly not a representative slice of 4A teams doing battle with 2A, 1A, and 2B teams. Of the 28 2A teams that scheduled up by two classifications (all against 4A), half were from the top half of that classification, 7 in the top 10. 2A fielded a lot better lineup going up against 4A.

Look at 3A. Eleven teams scheduled down. Three in the top half (#24 twice and #27), eight in the bottom half and six of those in bottom quarter.

2A scheduled down 14 times. Only three of the 14 were from the top half. Four were from the bottom three 2A teams. Klahowya, #54 (last), an epically dreadful team in the midst of a four-year period in which it won one (1) game, managed to lose twice to #60 1B team, Quilcene. Quilcene, though, rated 3 points ahead of Klahowya, so these were not upsets.

Fourteen of 1A teams scheduling down came from the top half of the classification. More than half. But where do those 1A teams live? Four games for Colfax, others from eastern Washington. People may be sparse; good girls basketball teams aren’t. Next town over might be 1B but the team is good. Seattle Christian (18) and Forks (21) were the only top-halfers west of the mountains. Okanogan (17) lost to Entiat (1B #6) twice: Entiat a six-point favorite anyway.

Overall for the 81 games, 28 down schedulers came from the top half of their classifications (4A, 3A, 2A, 1A); 45 up schedulers came from the top half of their classifications (2A, 1A, 2B, 1B). This is what you would expect. Coaches don’t want to schedule a slate of mis-matches.

If you had two hats, one with names of teams, from say 4A and the other from 2A, and pulled a name from each hat: who would win? Knowing nothing else, I would guess the 4A team. But if you had a hat with the games that were actually scheduled and asked who would win, that’s a different question: 50-50 for this season. It isn’t a random schedule–not for any of the cross-classification comparisons (2 or more differences). In theory, giving extra credit to the lower classification team winning or even playing the game seems logical. In practice it doesn’t necessarily work out.

But that’s not the end. What about the adjacent classification games? Funny that I picked 2006-2007 because that looks really interesting. What about next year, and the next? Just because something shows up one season doesn’t mean it applies to others. Now that there is the RPI incentive to making schedule arrangements, how has that changed scheduling? Then there is geographical asymmetry. Good teams are not evenly spread across the state, but most games are close by, posing difficulties for making cross-state comparisons. Lots of questions. But I’ve got a game to catch this evening.