Last bit of analysis on Elimination Round analysis. Prior to Turning Point, VRC Tournaments’s Elimination rounds were to be at most 8-alliances of 3 teams each, thus a Quarter Final, Semi Final and Final round. For Turning Point and beyond, the format shifted to 16-alliances of 2 teams each, with a Round-of-16, followed by QF, SF, and F.
After analyzing win rates for Red for Tower Take Over, there was demand/interest for a multi-year comparison. I’ve completed this data run (thanks especially to VexDB.io for making this possible).
The data:
Round
SS R Wins
SS R W Rate
ITZ R Wins
ITZ R W Rate
TP R Wins
TP R W Rate
TTO R Wins
TTO R W Rate
4YR R W Rate
Qualifiers
10,062
50%
11,840
50%
10,956
50%
12,828
50%
50%
R16
106
70%
176
67%
1,897
76%
1,752
75%
75%
QF
4,088
76%
4,596
76%
2,016
76%
1,746
74%
76%
SF
2,104
68%
2,382
71%
928
70%
832
69%
70%
F
1,016
64%
1,159
67%
431
64%
383
63%
65%
Total
17,376
20,153
16,228
17,541
17,825
The per-round win-rates across all years with more than about 400 match results appear to converge quite nicely, as the last column shows the 4 year composite win rate.
As noted, StarStruck and In The Zone seasons did not sanction tournaments with a R16. The results for these rounds may represent mis-entered data, tournaments with non-conforming elimination structures, or something else entirely.
So basically there is NO statistically significance difference in Red win rates between BO3 and BO1 with the assumption that it is rare that blue is the higher seed. (May happen on the rare occasion in TM).
Locally, I have each round broken down by seed (though it simply assumes the higher seed advances from the QF to SF to F…e.g. I don’t show cases where the 4-seed makes the finals) for each of these years. The win-rates of 1v8, 2v7, 1v4, etc. do not change significantly year-over-year. There is some small variance, as one might expect.
For comparison to other sporting events, here is the stats from college basketball converted to red wins %. The data is from here. When coaching I liked to complain that there were never enough qual matches to sort teams properly. But it does appear that on the EP side we run enough to be as good as college BB rankings.
In md table format:
Seed
R16
Q
S
F
1
99.30%
16
0.70%
85.30%
8
50.00%
14.70%
9
50.00%
77.90%
4
79.40%
22.00%
13
20.60%
51.50%
5
65.40%
48.50%
12
34.60%
67.60%
2
94.10%
32.30%
15
5.90%
63.20%
7
61.80%
36.80%
10
38.20%
58.90%
3
84.60%
41.20%
14
15.40%
53.00%
6
63.20%
47.10%
11
36.80%
R Wins R Wins R Wins R Wins
74.73% 63.25% 68.40% 67.60%
I was actually surprised that data wasn’t behind a pay wall. But is is funny. And, it didn’t take too much work, just what could be accomplished while listening for my name during a teleconf.
