Winningest FRC team of all time??

FIRST General Forum
#31

Coming from the context of sports, I always thought “winningest” meant “having the most all-time wins”, period, full stop, regardless of pretty much any other metric.

If you do an Internet search for “winningest programs”, every single article on the first page will be about which teams have the most all-time wins. This means that Ohio State University’s football program is higher on “winningest” lists than Boise State, even though the latter has a higher winning percentage, because OSU’s program started in 1890 and BSU’s started in 1932; thus, OSU has more wins.

Taking the “winningest” definition of sports, the winningest FRC program would be the one with either the most match wins or most event wins - to me it isn’t clear which one of these it is. Missing from the equation would be any attempt to reconcile the fact that some good teams are younger than others.

(Also, Michigan Football hasn’t been as good recently, so as a fan I’m glad they won a bunch of games in the early 1900s. Michigan has both the highest win total (winningest) and win percentage in college football :D)

#32

“Winningest” is a highly subjective metric. The most common interpretation observed in the thread is the most blue banners, or the highest banner/year rate. However, any team that sets out to change culture and ends up doing so may be interpreted as “winning” as well, as in perfectly meeting the goals of FIRST.

Success can be held to different metrics. Some teams view the goal of FRC to get to champs, some to build a robot and have fun, some to win on Einstein, some to share the experiences they’ve had and set out to change the world. All in all, the one who wins the most is society from the amazing things we can accomplish together in the grand alliance of life.

#33

The argument of the most winningest team of all time in my opinion is a mix of blue banners won and the time they have been around. Defining the most winningest team as the most blue banners is like saying Bill Russell is the GOAT because he has 11 rings, the most rings. It is isn’t fair to teams that didn’t have good alliances or hard competition, even teams that haven’t been around for that long.

That being said it’s hard to argue against 254 48 blue banners, the most, and 3 world championship wins, tied for 2nd most.

#34

Lots of good conversations on the topic.
Perhaps the title should be changed to include what people define as “Winningest.”

My personal favorite is Team 254.
Having been introduced to FRC by their rookie robot in our library at our school in Hawaii in 1999, we have had some sort of relationship with them, including Team 368 TKM.

The most impressive feat to me is that from their start in 1998 to present, they have always built one of the best robots for competition.
They switched schools, had several turnovers of key mentors, lost some key NASA sponsorship and shop space, but still manages to recruit the best mentors and alumni while building world-class robots.
Many other greats of the past have had their ups and downs, but this team hasnt had it yet.

Teams like 67, 987, 1114 and 2056 are close runner ups in terms of the consistency of running at an elite level throughout their entire history. Perhaps 2056 is the exception, but lets see what happens in 10 years or so. I dont believe they have lost any key people, where they had to adjust and rebuild.

#35

There will always be teams that dominate FRC in their respective regions but in different time eras. Fore example, one could say 71’s lasting 4 championship wins (1997, 2001, 2002, 2004) along with its 27 blue banners could make it the “Winningest FRC Team” while another could say 254 with their 48 blue banners and 3 championship wins (2011, 2014, 2017).

However, maybe we could assign teams ELO ratings for each season where each team would start with a base ELO of 1000. This could be used just for elimination matches and/or could include qualification matches. The system would add together each alliance’s total ELO and compare against the opposing alliance’s total ELO. The losing alliance has each team share an equal amount of lost ELO while the winning alliance would have each team share an equal amount of gained ELO. This would more accurately estimate a team’s success in a season as teams would lose less rating against tougher opponents and by the same token gain more rating against tougher opponents.

We could estimate a team’s total success as “Winningest FRC team of all time” by taking the average of every season’s ELO into a total all time ELO.

#36

There will always be teams that dominate FRC in their respective regions but in different time eras. Fore example, one could say 71’s lasting 4 championship wins (1997, 2001, 2002, 2004) along with its 27 blue banners could make it the “Winningest FRC Team” while another could say 254 with their 48 blue banners and 3 championship wins (2011, 2014, 2017).

However, maybe we could assign teams ELO ratings for each season where each team would start with a base ELO of 1000. This could be used just for elimination matches and/or could include qualification matches. The system would add together each alliance’s total ELO and compare against the opposing alliance’s total ELO. The losing alliance has each team share an equal amount of lost ELO while the winning alliance would have each team share an equal amount of gained ELO. This would more accurately estimate a team’s success in a season as teams would lose less rating against tougher opponents and by the same token gain more rating against tougher opponents.

We could estimate a team’s total success as “Winningest FRC team of all time” by taking the average of every season’s ELO into a total all time ELO. This would most likely/indefinetly still rank the powerhouse teams like 25, 27, 33, 67, 125, 118, 148, 180, 195, 217, 330, 359, 469, 610, 624, 971, 987, 1114, 1477, 1538, 1678, 1986, 2056, 2481, etc (so many others) to name the least still on top but in a more defined and orderly way.

#37

https://www.chiefdelphi.com/media/papers/3306

#38

Using that spreadsheet, if you take team’s average ELO since 2008, the list looks like this:

254
1114
2056
1678
118
148
987
971
1986
195
330
33
2481
2122
16
359
67
1806
4488

If you take team’s average ELO from the past 4 years, the list looks like this:

1114
2056
254
987
67
118
148
330
1986
469
33
1678
16
971
359
4488
624
111
525

In both lists, the top 3 spots are populated by 254, 1114, and 2056.
This would be a “winningest team” metric relative to match wins.

#39

ELO is designed to figure out who would win, but “winningest” is an intentionally naive statistic that measures who has won the most. This is like how FPI is a better measure of how good a football team actually is than wins, but again, we’re talking about an intentionally naive statistic.

I’d lean towards individual matches more than events because that’s the smallest individual unit. So I think somebody needs to do thousands of TBA API queries so we can figure out the answer.

#40

Or use the data dump repo :stuck_out_tongue: save them some bandwidth.

#41

One thing I love about FRC is seeing younger teams rise to power house status. I call them New Tech, and the teams who have been consistently good for at least a decade are Old Tech.

Some Old Tech Teams would be (There are more, these are just examples)
33
67
118
254
330
1114
2056

All of these teams have been fairly consistently solid throughout their lifespan as a team in FRC.

Then there are the New Tech Teams (Again, just naming a few)

2122
2481
2767
3310
3476
4488

New Tech Teams have started later, but have been building up their knowledge and are at the point where all of them can consistently perform at the high levels of play

Next there is the category of teams who have been around a while, and have done pretty well, but have really hit their stride in the last few (3-4) years.

195
971
1619
1678 (Citrus Circuits to an incredible degree, 20/22 of their blue banners were won in the last 4 years)
1690

There is also the inverse of this, which are teams that were giants, but have since fallen whether it be from loss of mentors, build spaces, or other factors. These teams are still fantastic, and are always one game/mentor/student/sponsor/etc away from having a great season.

71
111
177
217
233

Lastly there are the teams that are always good, but when they get the right combination of factors just knock it out of the park.

469
1241
1625
2338
2826

The point of this is to say that there are so many factors to determine a team being the “winningest”, that it is a poor metric to go by. FRC teams are so dependent on so many different factors that it is impossible to be objective.

(Apologies if I left out any obvious choices on the various lists, being from the Midwest I have an obvious leaning. Feel free to add any teams I may have left off)

#42

I think these lists are flipped. 1678 should be near the top of the 4-year list, not the 10-year list.

#43

Somewhat off-topic: Another concept that comes to mind in the context of “best teams of all time” is whether or not there is a “triple crown” of FRC.

That said, I personally don’t see a true triple crown as there are not three comparable high awards. We have Chairman’s (HOF) and Championship winner (lots of shared ground there, many teams to list). But the other top tier awards are personal, not team Awards, like WFA (which AFAIK the only teams to meet that plus the other two are 365, 67, 16, and 51*) and perhaps Dean’s list (Which I have not thoroughly studied to give a team list).

*- Due to the 47 and 65 merger. HOF though 47, WFA (Ken Patton) and championship win through 65. Both teams were top notch (as is 51), so I’m counting it

That said, at least in the first case (which has a small enough pool for me to comprehend), there are comparatively many teams with two of the legs, and two added this year alone, 2614 and 359 (both have a WFA mentor past or present and also are HOF).

There are other event wins which could be counted; historically there have been events considered as hard if not harder to win than championship (Midwest regional in years past, IRI and Michigan State Championship currently, and FOC this year), but since they don’t have nearly as wide of pool of eligble-to-win teams, winning such is a great accomplishment all right but IMHO not fair to count as the third leg of any triple crown definition.

That said, of the two-leg (HOF + Championship win) teams (“Double Crown” teams?), 67 a bonus for doing so at once, but since that’s practically a one-shot occurrence in the modern day, one could argue that luck is part of such since there have been several near-misses. 842, 1114, 1538, and 987 come to mind (842 and 1538 were division finalists whose elimination was not expected, the others all made Einstein with guns blazing but did not win).

Unfortunately, the double crown doesn’t narrow the list much it seems, as of the teams who seem to get mentioned a lot in this thread, 67, 254, and 1114 are all HOF and championship winners.

Perhaps there actually is no single #1 team here? Not this is at all a bad thing… lots of teams to look up to, which is a good thing (and I don’t think that’s disputable).

#44

TBA only has match data back to 2002. If someone has data going further back that would be awesome. I’m planning on summing match wins from 2002-2017 for the following teams: 254,1114,2056,67,359,125,148,20,190. Any more suggestions?

#45

With all due respect to a great team, 190 isn’t going to be close to in contention for this title.

Notable teams your list misses include 987, 469, 217, 195, 111, 71, 1678, 33, 16, 25, etc. Many of these are teams which have not had as much “flashy” recent success, but they are historically among the very best in the world.

Including match wins in 2003 will mess things up because the system for both quals and elims was not WLT based. Including match wins in 2015 similarly does not make sense.

#46

Dump to database, compute wins, sum all team wins, group by team… why not do all teams? It’s trivially easy for any year that’s not 2015 (which I propose we pretend simply never existed)

#47

Team 358’s history page has some hand-collected data in the form of excel docs for each year: http://team358.org/history/

#48

Would love to see total win results for all teams and maybe winning percentage since many teams only go to one or two events a year ( may be more helpful metric for predictive use going into events with grain of salt consideration?)

#49

2005 - Present
All events in tba-data repos (This means it includes some off season events, the data repos don’t have a handy way to skip these)
Define red win as redScore > blueScore, blue win as blueScore > redScore. Ties are not counted as wins for anyone.

“Winningest”

  team   | wins 
---------+------
 frc67   |  812
 frc1114 |  773
 frc33   |  752
 frc469  |  723
 frc118  |  699
 frc254  |  662
 frc217  |  662
 frc2056 |  651
 frc27   |  605
 frc148  |  583
 frc195  |  559
 frc1678 |  545
 frc330  |  524
 frc1918 |  520
 frc359  |  515
 frc494  |  513
 frc987  |  510
 frc125  |  509
 frc68   |  498
 frc503  |  496
 frc25   |  494
 frc1023 |  491
 frc1519 |  488
 frc70   |  486
 frc48   |  477

Most Likely To Win

  team   |     win_pct      | matches | match_wins 
---------+------------------+---------+------------
 frc2056 | 82.7191867852605 |     787 |        651
 frc1114 | 81.5400843881856 |     948 |        773
 frc254  | 80.3398058252427 |     824 |        662
 frc1678 |  79.100145137881 |     689 |        545
 frc987  | 78.9473684210526 |     646 |        510
 frc1986 |  78.348623853211 |     545 |        427
 frc118  | 76.3100436681223 |     916 |        699
 frc330  | 75.6132756132756 |     693 |        524
 frc4613 | 75.5952380952381 |     168 |        127
 frc971  | 74.9211356466877 |     634 |        475
 frc5172 | 74.6666666666667 |     150 |        112
 frc67   | 74.6323529411765 |    1088 |        812
 frc4488 | 74.1433021806854 |     321 |        238
 frc1519 | 73.0538922155689 |     668 |        488
 frc2970 | 72.9166666666667 |      48 |         35
 frc148  | 72.6932668329177 |     802 |        583
 frc1717 | 72.6027397260274 |     438 |        318
 frc469  |             72.3 |    1000 |        723
 frc1574 | 71.4723926380368 |     326 |        233
 frc359  | 71.2309820193638 |     723 |        515
 frc2122 | 70.4761904761905 |     525 |        370
 frc5553 | 70.4545454545455 |      44 |         31
 frc525  | 70.1525054466231 |     459 |        322
 frc2052 | 70.0534759358289 |     374 |        262
 frc2481 | 70.0258397932816 |     387 |        271

Code to import and recreate all views and queries is available at https://github.com/schreiaj/TBA-Data-SQL

TBA Data is loaded as a git submodule so that may need to be updated from time to time.

I used PG 9.6.

#50