The state existing before the war
I’ve marked this as opinion, as I include stuff we don’t track statistically.
A technical upset is when a lower rated team defeats a higher rated team. The first day of the War includes so many closely rated opponents that it is incredibly unlikely we will see every match predicted correctly. In our Gonzalez system’s current, straight success rate, the higher seeded team will win four of five games. However, this is less likely to apply as the rating gap between teams increases. Any of GVSU’s scheduled games are unlikely to result in a technical upset. CSU vs Akron is much fuzzier; each team has near identical ratings, being only 0.157 difference.
That said, these are the kinds of matches that are a test to the system. The wins and losses and overtimes (8.19% chance of occurrence) at the end of the day will better determine each team’s rating going into Nationals. We’ve never had this many great, even match games for Nationals before. I’m excited to see the results.
|Exchange if Upset
Alright, we know GVSU’s not going to gain statistically much from Saturday. All three of the matches are the minimum value, including MSU whom they have about 10 points on. If GVSU is not deserving of this, they have the chance to lose up to 6 rating points. I don’t think people really give GVSU credit on how good they are rated compared to the rest of the League. Even to the top 2-6. The 2-6 are competitive, close in strength. But none of them, even #2, match the state of GVSU. This is objective data…
Moving on. There’s plenty of factors that aren’t directly represented in the system that always play a part. We aren’t predicting life. The first two matches will affect the rest of the day.
Cleveland v Akron (home): 0.157 Gap
Cleveland State is a new team taking on a home team. New teams tend historically not to win a lot in their first 6 matches, effectively most of the first season they participate in. They also tend not to bring a full roster, and it is more likely the home team has a fuller roster. I really wish we kept track of that sort of thing. It would be interesting proof that this home/away mechanic tends to be true. In any case, this is an even match going in. CSU is the technical favorite but their straight chances are lowered due to how close the rating gap is. Upset: 5th percentile
MSU v VCU: 2.985 Gap
VCU is currently rated at a precarious #10 at 42.489 and have tenths of a point between #9-12. MSU is sitting at a comfortable #7 at 45.474, although I’d probably gather they think it more precarious. This match is MSU’s middle of the road match, where they are slated to win but a VCU upset is not out of statistical relevance. This rating gap falls within that 1 standard deviation (68%) of all technical upsets. To borrow some phrasing from Hunter in an earlier article, this is a medium risk, high reward match for VCU. Upset: 61st percentile
GVSU v Ohio: 17.889 Gap
When GVSU played Ohio last year, it was the largest, most lopsided rating gap in history (55.458 vs 31.004). The 24.454 rating gap was the League rating spread at that point in time. Top team vs low team. This shouldn’t ever happen in Nationals round robin play. When GVSU faces Ohio, it’ll still be the 14th largest rating gap but doesn’t seem as bad as Nationals 2016. Regardless, it is interesting to see these two play. Anyone can play who they want to play. Upset: 100th percentile
Cleveland v BW: ~1.419 Gap
Another close game, this will be a nice test for both ratings. Should Cleveland lose in their first match, this second will have an altered rating and bump BW as the slated winner. Neither teams have a deep run of games played this season, so that is primarily why this game is a good rating test. People might say a fresh, first match of the day BW might have an advantage over CSU, having played a match just prior. I’m not sure that second match matters as much, I like to think that the third match is what really tires you out. With distance traveled being equal, BW could have this potential advantage. Upset: 34th percentile
VCU v Ohio: ~4.497 Gap
With almost 5 point rating gap, this match looks like many of the earlier matches of the League. We didn’t have this the League wide parity in older seasons. As it stands, and not ignoring the potential VCU upset in the earlier round, VCU has a larger statistical chance of keeping to that 4 in 5 prediction success rate. An upset here would fall outside that first standard deviation, and would be significant. Upset: 77th percentile
MSU v Akron: ~9.423 Gap
This falls under MSU’s “easy” game, although I would never underestimate a home team or a middle of the day match. This match would rank around the second biggest upset in history, but those numbers are dependent on the earlier performance of each team in earlier rounds. Upset: 99th percentile
DePaul v BW: ~1.477 Gap
Maybe Akron was gracious enough to give DePaul extra time in the morning, though they’ve been slated to Officiate the first round. DePaul’s first match will depend on BW’s performance earlier in the day as well as CSU’s match with Akron. These prior matches are high value, so it’s an example how predictability falls apart later in the day. Who gets the technical upset is up for grabs. Upset: 36th percentile
CSU v DePaul: ~0.806 Gap
Now we’re getting silly. CSU will have played two games by now, and DePaul one. These potential ratings can fluctuate by zero to three. But, this will be Cleveland’s last match going into Nationals. Wherever the ratings fall, we’ll have a much better evaluation of a new team than we’ve had in many years. Upset: 19th percentile
GVSU v MSU: ~10.370 Gap
This one is tough to predict. I would probably do best to go with the numbers. That MSU will win against VCU, Akron, and end up here against GVSU. We’ve never seen the max exchange of 2 hit by a technical upset under the revised Gonzalez system. If anyone has a chance to top GVSU in this situation, it’d be the current lowest rated Michigan team. History has shown that 4th MI rated SVSU has topped 1st MI rated GVSU on occasion. An upset like this has never happened, but 1 in 1172 is still a 1 in 1172 chance. Upset: 100th percentile
Akron v BW: ~0.947 Gap
It’s hard to say anything concrete about this game. Akron has more experience; BW’s rating will depend on their performance against similar rated opponents earlier in the day. This match will probably end up being competitive between two very similarly ranked opponents. Quality dodgeball. Upset: 22nd percentile
GVSU v VCU: ~12.872 Gap
This pair had the 2nd largest rating gap in the history, 23.541 at Nationals 2015, but on day 2 in the tournament bracket. Now, the gap has closed so incredibly dramatically that it proves this VCU is similar only in name. Those 2015 days are so statistically insignificant to VCU’s current rating, and the same is true for GVSU’s. Still, GVSU is damn good. The numbers favor them by a wide margin. But our kind of dodgeball is never solely about numbers. VCU will have a chance to take it to GVSU for only the second time in club history. I’m excited to see how it turns out. Upset: 100th percentile
We also have spots for potential unscheduled matches. Let’s hope we see more dodgeball at 5:00pm!
Stay tuned on Saturday night, where I hope to release the interbellum Gonzalez standings after this day’s all said and done.