Upsets, Oh My!

FIRST General Forum
1

For the 1st and 2nd week of the FIRST, there was quite a bit of interesting stuff going around what with the upsets at several regionals.

There have only been 19 events that happened so far, but by some crazy chance, the lower seeds seem to be making quite a name for themselves. While the first week had 5 1st seed winners out of 8, 2 more wins were by the third seed. While this might not seem so much of an upset, the first real one came in Central Valley where 840, 1678 and 295 pulled off a 6th seeded win. Congratulations to that team for kickstarting quite an interesting set of upsets in the second week.

Overall, the average seed for the first week to win had a value of 2.125, or the second seed. Not much of an upset, but onto the second week we go!

This week had a lot of interest in the upset category. While some events had some dominant 1st seeds cleaning up the competition and setting some records (1114, 2056, 1325 at GTE; 987, 254, 1138 at San Diego are two examples), there were some more… interesting results. There were 6 1st seeded wins out of 11 for the second week, with the 3rd seed of 2169, 3061 and 525 getting the win in Northern Lights. On a side note, the 4th seeded alliance at the same regional were finalists (269, 2826 and 1259). Beating the odds is still notable when you’re a 4th seed!

The interesting part?

The other 4 regionals had 6th, 7th and 8th seeds winning.

WPI - The 6th seeded alliance starts us off as 3280, 3044 and 2168 (some big numbers compared to winners like 118 and 148) in a sizeable upset. But it doesn’t stop there.

NYC - 694, 4797, 3137. The 5th seeded alliance–complete with a rookie and a team on the newer side–won the regional, which in its own right isn’t as formidable as the WPI win. Of course, that’s not to downgrade their win; it’s just that their competitors in the finals, 1660, 4122 and 3171, were the 7th seeded alliance. 5th seed and 7th seed in the finals; can you ask for a better upset?

Yes, yes you can. But before we get to that regional, I’ll show you THIS regional.

Orlando - While their competitors weren’t a low seed like in NYC, 744, 1772 and 79 had all of the odds against them. The reason?

They were the 8th seed. They took the odds, pulverized them into an indiscernible mess and beat the highly favoured 1st, 5th and 2nd seeds in that order. Now, what regional could top an 8th seeded upset?

Lake Superior - The winners were the 7th seeded alliance, 2052, 4011 and 2062 won and smashed the odds like the other upsets did; their opponent, though, was doing the exact same thing. 3276, 4539 and 3277 were the 8th seeded alliance that moved up to the finals, eventually losing out despite their cinderella run.

Those are some crazy odds some teams beat out, but there are many more weeks for this to keep happening. This is only 1/4 of the action, folks.

A huge congratulations to the upsets shown above, both the upsetting winners and the finalists. As for the ones who fell to them, you are still an amazing alliance, and your work was unfortunately cut short by a cinderella run; it happens to everyone.

That’s all for the first 2 weeks. As a point of interest, the average seed for the second week was 3.36 to give a 3rd seed average, but I can’t help but include something for the low seeds that made it to the finals. So, for interest’s sake, let’s say the second week had a success rate of a 4th seed. That’s quite something right there.

See you next week!

2

Most of the time when this sort of thing (low seeds winning) happens, people immediately start crying about the seeding system.

I think it is quite the opposite.

The seeding system is quite good this year, it is simply that almost overnight, many regionals have gotten much deeper. Deeper by virtue that teams seeded 8th through 16th actually have a chance at winning the event. Where in many years, the first second and third seed are orders of magnitude better than “the pack” in the 20th seed and beyond, this year they are only slightly better.

I think a major factor of this change is Robot in 3 Days. It showed teams who might have made mediocre robots an easy way to make a simple and good robot. We can discuss if this is a good change (competition up, innovation down) but I think it was a major factor. It certainly made the events more fun to watch.

3

Well, a big reason why people complained about it last year was because of the coopertition points. Not having the points in control when you’re an elite team leads to a lot of them ending up as a little lower on the seeding scale. That’s why this year, low seeds winning seems like a much bigger accomplishment.

To touch on what you said, I find it extremely admirable that teams are able to adapt to a regional so well in a couple days of an event, what with adjusting their robot from a 20th seed skill level to a top 5 seed skill level. That level of adaptability coagulates into a nice, interesting matchup when they get to the plate and win. The regionals get much more interesting, the low seeded teams’ despair over a low seed vanishes (and feels amazing at the same time; before our regional chairman’s win this year, 781’s 8th seed upset was the best moment of my FIRST career, passing even our world final appearance in 2011) and the higher seeds–most of which are elite teams winning at multiple regionals–get to put their skills to the test and actually enjoy a hard fought, fun match where losing is almost as entertaining as winning.

4

Michigan seems to be somewhat immune to this trend so far:

Week 1
-Traverse City: 1 defeats 3
-Kettering: 1 defeats 3

Week 2
-Gull Lake: 1 defeats 2
-Waterford: 1 defeats 6

So far, the #1 seed is 4/4. #2 fares far worse, making it to the finals only once so far.

5

Yeah, Ri3D was what gave us our shooter. They helped us, 4228, a lot since our budget does not allow for much prototyping. Our shooter was one of the more successful ones through the first two days of competition racking up a lot of points for us…it is a shame that one of the tires popped without us noticing until it was too late on day 3 -.- (Too late meaning after we lost our Qual match by 9 pts and our shooter had only about 30% accuracy that match)

Overall, I think Ri3D had a positive impact on a lot of teams in FIRST, innovation did not go down because of it, most teams that I saw used an aspect from Ri3D and put it on their robot…but not all of it.

6

Orlando reporting in — I honestly have no idea how our 8th seed alliance formed. We got into the scouting meeting and realized all three of us were solid shooters with 10 point hangs. Set strategy to be an offensive strategy (79 bouncing between harassment and scoring but focus on scoring) and ran that through.

79 hasn’t won a regional since 2009 (hasn’t won Orlando since 2006), 744 and 1772 had never won a regional. I understand my own team, but how 744 and 1772 have never won is beyond me. They were solid teams and were a genuine pleasure to work with.

Edit: I’ve been asked to also draw attention to the fact that not a single one of our teams was in the top 10. Our 8th seeded alliance was made up of a 13, 15 and 19 seed.

7

And that was a well deserved win. They had to go through the 3 alliance, the 2 alliance, and (us) the 1 alliance. It goes to show how the game this year is so dynamic, your strategy can really change how well you perform. The finals matches were especially back and forth, and definitely the most fun I’ve ever had in FRC so far. I’ll look forward to evening the score at SVR.

8

I think it’s quite interesting that this is happening in a year when there are no “tricks” to the ranking system at all, no coopertition, no points based on opponents’ scores, nothing; just QP and autonomous points as a tiebreaker. One might think it would work the other way around…

9

^The game is already the most strategically deep game in the modern era (in my opinion), so I can understand why the seeding system (not ranking system) is the most coherent one in recent memory.

I remember posting last year that anyone who built a strategy to put them in a position to make educated, well-informed picks has the advantage over someone who is the 1-seed. If a regional does not have a super-elite team at the event, the eliminations are wide open. Not because last year, where the bridge acted as a “great equalizer” of alliance scores, but because robot diversity is probably the biggest it’s been in the modern era.

Not only are regionals getting deeper in quality, but also deeper in quantity. Not too long ago most regionals would cap out in the 50s with a few exceptions, now regionals with fewer than 40 teams are becoming rare birds (maybe because FRC is rapidly approaching a point where it can’t add more events because it can’t provide 6 more CMP bids while CMP is still under the 400 cap).

I know that heading into a Week 3 regional I will probably lecture to the scouting team more than the recommended dosage to make sure the team has solid data to work off of. So many good alliances are formed on lucky or well educated picks that other seeds ignored.

10

Number of matches played is a major factor in how you seed. This year, both of the regionals we are attending have ~65 teams registered. At Palmetto, we only played 8 matches. Unless the number of matches played is statistically significant, or you have a stand out robot, seeding is more a matter of chance. We found OPR to be a better predictor of a team’s performance than seed. And even better than that was individual robot scoring data obtained from scouting. Both seeding rank and OPR are influenced by your randomly selected alliance partners (in quals). And the fewer matches you play, the stronger that influence.

So “upset” is a subjective term. At Palmetto, we were part of the #1 seed, but the alliance that won, had a much higher combined OPR than our alliance. I would have considered us beating them to be more of an upset than the other way round.

11

A fun fact that isn’t recorded on the FIRST website (or The Blue Alliance):
After awards on Saturday, the winners of the Northern Lights and Lake Superior regionals faced off against each other, with the Lake Superior alliance (#7 seeded 2052, 2062, 4011) beating the Northern Lights alliance (#3 seeded 2169, 3061, 525) by a mere 3 points.

2052, the alliance captain for the winning alliance was actually ranked 9th before alliance selection.

12

Orlando was EXTREMELY exciting to watch, it was quite the nail biter!!

13

That was a really fun match to play in and very exciting when we won :slight_smile: Apparently our alliance’s strategy was a very good one since we used the same strategy against all our opponents and came out on top. Usually we have to switch it up for the semi-finals when our new opponents counter-strategize.

14

Yes I was on team 3044 at WPI. I think what this regional has showed us is that it doesn’t matter as much of your bot this game as much as your strategy.
No matter what though, playing against the teams we played again is an honor I will remember for the rest of my life. Thank you to our alliance (2168 and 3280) and to every team we faced, all of you have made our experience memorable.

15

The relative ease of scoring discs and the accessibility of the ten point hang leads to a strong middle tier. Combine that with a serpentine draft and great scouting, and upsets are to be expected. 2168 in particular capitalized on their favorable seeding position to make possibly the two smartest picks in the entire WPI draft.

16

Perhaps our rankings are more representative as every team gets 12 qualifying matches at districts as opposed to the 8-9 that teams are getting at regionals. I haven’t the time to run numbers at the moment, but more accurate seeding could be a contributing factor around here.

17

For those of us who missed the Lake Superior matches, do you mind enlightening us as to what this wonderful strategy was?

I strongly agree. This game is incredibly strategically complex. There are so many different valid strategic decisions, it makes for quite exciting matches, and a lot of potential for upsets.

18

Two shooters, one a climber one just a shooter, and one defensive climber blocking the opponents best robot and keeping the defense away from the two shooters. The shooters just went back and forth from the feeder station to the back of the pyramid to shoot so we couldn’t be defended against very well and we alternated from shooting to feeding so we didn’t get in each other’s way. Then, the climbers climbed while the plain shooter either kept shooting or protected the climbers. Worked like a charm and it was very hard to defend against.