This will be the sixth Ohio Dodgeball Cup on record, continually held at slightly rotating locations since 2011. Some people might draw comparisons to the earlier held Michigan Dodgeball Cup, I get it. But I don’t think that comparison is really fair. This is a different set of teams, a different set of traditions, a different caliber of dodgeball with a similar name. It was two Ohio teams that claim the first intercollegiate dodgeball match in history, touted as the Battle for Ohio Dodgeball Supremacy back in March of 2003. But I’m from Illinois; the teams could tell me off and that’d be fine. Let’s look at the Gonzalez exchanges for this event:
|Time||High Seed||Low Seed||Rating||Rating||Predicted Exchange||Exchange if Upset|
League Rating Average: 42.211 and a spread of 28.172
ODC Rating Average: 38.346 and the largest spread is 9.561
ODC Average Exchange: 0.538 and an average spread of 4.618
Working up the pairs
I had the privilege of being asked by Akron to help prepare matchups for this ODC. I started first by using our handy tool to determine what opponents each team played this season. The Ohio teams have played each other a fair amount, twenty matches among all competitors (not including OSU games).
I sorted these potential matchups by number of occurrence. Akron / Kent and BW / Ohio occurred four times, so those are nixed for variety reasons. BGSU hasn’t played their fellow Ohio teams at all, so they get four matches. As it happens, these matches are all fairly close in rating and there’s not one match that will fall below the minimum exchange value. This event is essentially the average status of the League. The median rating is going to be close to 40 (the initial rating given to all new teams) and the current League Median Rating is 39.265.
So you can direct whining to me on the matchups. I’m honored to have helped, and I’d do the same due diligence for anyone else who needs advice.
How do we evaluate this data?
In hard terms, Akron receives a three point boost for home advantage, so they take the high seed of this event. The rest of the teams aren’t that far, and I expect competitive matches all around. Do not take these predicted exchanges as set in stone; I wouldn’t take this day to include some upsets just like the closely paired MDC.
The whole purpose of the Gonzalez system is to evaluate a team’s strength over time and adjust accordingly. We talk about technical upsets, but what is really happening is a team’s rating is being adjusted to be stronger or weaker compared to the performance of that game.
So what about upsets?
A few people enjoyed the potential Upset Exchanges ranked in history from the MDC post, so here they are for the ODC. The latest count is 162 matches of 883 ranked matches (18.34%) have resulted in a technical upset exchange. A technical upset is defined as a lower rated team defeating a higher rated team in regulation or in overtime. In regulation, that exchange will be greater than one. But if the game was decided in overtime, the exchange would be greater than 0.5 as OT halves the exchange.
|Low||High||Exchange||Rank / 139|
If we have 883 ranked matches, what if we take two standard deviations from the mean? 95.45% of all matches would be 842.8 which leaves about 40 matches. So maybe at this moment, the top 20 upsets fall in the 3rd standard deviation from the mean and might be found statistically significant. That could represent that 12.4% of all recorded technical upset matches are statistically significant. In terms of standard deviations. That could be a misuse of the principle. I really wish we had a statistician or mathematician around to help interpret our data, and the rest of ranking systems.