This year, it is particularly evident that being on an alliance other than the #1 seed makes it extremely difficult to win an event. There’s been a lot of discussion around changes that would make it easier to win, or at least get an invitation to the championship, from reversing picking order to adding more slots. What about game design? I think that’s the most likely culprit for our current circumstance. I think the last time #1 seeds don’t run the field every event was 2015. Anyone want to correct me on this? Can we discuss how game design makes upsets more difficult?
4 Likes
Games that require more team play lead to more upsets imo. I totally believe this year is a pretty balanced game which allows the upper seeds to generally do well.
2012->Triple balances could outweigh basketball scoring (that plus bringing out new balls for elims hurt teams that had dialed in shooters at a lot of events).
2014->Getting more points a cycle per robot touch hurt upper seed alliances when the 23/24 robot couldn’t manipulate the ball well (until the bounce passes started happening).
2017->The climbing was so high valued that not having a climber was like a DNP. Upper seeds fell to the serpentine.
2020-> Once again the point value for the endgame was pretty high which hurt alliances that couldn’t get a 3rd climber on the backside of the draft.
6 Likes
Honestly I think the fact that 1 seeds win the overwhelming majority of the time is a function of good game design… Rapid React does a fantastic job of sorting out the top teams during quals. The RP requirements are perfectly valued to enable top teams to largely control their destiny.
I much prefer this compared with games like 2017 in which all teams were subjected to climber roulette (if you had a climb deficit it was really hard to win). In that game, it felt like you had to be in an elite tier to overcome a bad schedule.
Teams that build great robots and execute in the playoffs should win in my opinion. I don’t think it detracts from the excitement of events in a way that it’s a problem.
27 Likes
It’s also worth noting that alliance dependant games tend to produce less accurate rankings, and that the best robots at the event won’t always end up on the first alliance because of that. This produces more “upsets” per the bracket, but the effect might not be as strong as you’d think when looking at good scouting data.
This was particularly true of 2014.
1 Like
I don’t disagree; this is my favorite game since 2014 and probably my favorite game ever, and hey, my team did pretty well so far this year, so I’m not complaining. I do think that game design impacts how easy it is to have upsets, though, and there’s always that “rooting for the underdogs” thrill which hasn’t had much room to happen this year.
2 Likes
2017 had pretty low win rates for 1 seeds for most of the season. Given the scoring structure, that was a game all about break points, and unless a 1 seed could hit a 4 rotor break point more consistently than the lower seeded alliances, having a less reliable 3rd climber often put them at a relatively disadvantage compared to being the 1 seed in other years. Stealing @Super84 data from this post, thru the first 3 weeks of 2017, 1 alliances “only” won 45% of events and were finalists at less than 60%. About 1 in 5 1 seeds in the first 3 weeks of 2017 lost in the quarter-finals.
Granted, that data is thru 3 weeks and I’m far too lazy to calculate it further myself, but the general trend would hold with 1 seeds being vulnerable basically until Championship (and even at the Houston Championship, more non-1s advanced to Einstein than 1s).
4 Likes
I think the reason that upsets aren’t as common this year is because of how massive the skill difference is between the top ranked teams, and the lower ranked teams. If you look at a game like 2018 or 2019, it’s not that difficult to build a robot to play that game well. If you build a bot this year with an unreliable shooter, you won’t be competitive. Building a good shooter is just harder than building something to place game pieces. Good teams will always find a way to be as competitive as they can be depending on the game, I just think that the good teams this year are really good, which is why there aren’t as many upsets. This isn’t necessarily a bad thing. As long as the games continue to vary from year to year, I will be happy.
Also, compared to 2019, you can’t just send one of your robots out to play defense. If you try that this year, it won’t be as effective as it was in 2019.
I enjoyed watching the top ranked bots this year. Some of them are incredible.
One point that I feel is overlooked a bit is that the past two years had a very imbalanced effect on different teams.
The very best teams likely didn’t lose a significant amount of their teams’ knowledge, likely enough of their mentor base stuck around to maintain their teams’ knowledge base and they often have more robust training programs to get rookies more knowledgeable quickly.
Many teams who might be competitive with the very best suffered a lot from such a long time away. These teams could have mentors step away during the hiatus or have a strong student leadership team that lost a season where they could teach their members while building their 2021 robot.
The teams that aren’t often competitive with the very best often have weaker structures transferring knowledge between years even in a regular season, so they will not likely see a decline as severe as the second group.
All of this combines to make events look even more top-heavy than usual. If we played this game in 2019, I don’t believe that it would seem quite as extreme as it does now, though it wouldn’t be anywhere near as favorable to lower seeded alliance as a break-point based scoring year like 2017.
8 Likes
All very good points. There has been a tremendous loss of growth among many teams, which has manifested in teams folding (maybe 10-20%?), in the difficulty of meeting and building (especially in Canada), and in the weakening of culture and training. Sad events all around.
3 Likes
Unfortunately 2020/2021 messed up the available data sets, but it’s not evident to me that upsets are getting significantly less common over time. What has become more common is that wins not by the 1 seed are increasingly by a higher seed – in 2019 and 2014 only 56% of wins were by the 1st seed, but in 2019 74% of those upsets were by the 2nd or 3rd alliance, in 2014 that was only 55%.
While I think a lot can be attributed to game design – 2015 was a particularly uncommon for upsets due to the lack of defense, 2017 particularly common due to how valuable climb points were – in some areas, the dominance of a small subset of teams has become larger over time. California is a particularly striking example of this, as a state still in the regional system (few events) and a handful of very consistent powerhouse teams that have increasingly pulled away from the average team, making upsets rarer and rarer.
[1] For events that have completed, that ran with 8 alliances only
[2] For events that were held
7 Likes
Fantastic data, Rachel, thank you!
This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.