Reducing Luck in FRC Competitions

Like, say, climbing this year over shooting power cells. The problem is how am I supposed to stand out as a team that picked the, arguably harder, game objective of shooting if climbing is what’s been winning matches?

1 Like

The beauty of an alliance system is that it allows teams to bring more to the table than just what their robot can do in terms of scoring game elements/individual performance.

Maybe Team A isn’t the best at scoring one game element. That’s okay because in a system where you’re challenged to play as an alliance, they can still find meaningful ways to contribute. Maybe they’re really good at shuttling game elements to their partners to maximize alliance score, or maybe they’re a beast at defense. All of those things can be a huge factor in getting a win, but there’s no way to measure that in an individual performance-measuring system. FIRST gives teams with all different levels of skill an opportunity to shine in their own way and be a part of an alliance. Ranking teams by a rigid “scored X points, did X actions” standard isn’t conducive to that, because only your quantitative aspects are measured, not your ability to contribute to a win, which is what competitions are about.

2019 was a great example of this. #1 seeds were upset left and right. This wasn’t because they made bad scouting picks or necessarily did anything wrong in their situation. The quantitative data was on their side. They were upset because lower ranked teams who had statistics stacked against them came together to create a whole greater than the sum of its parts. And imo that is the ultimate strategy.

also + to Gary’s post about FIRST often incentivizing certain aspects of the game. Often times these RP incentives can also be gained directly through individual performance, meaning a slightly less equal schedule has less of an impact.

9 Likes

I do not agree, just the opposite. There are examples of individual sports statistics in every sport, all of them keep individual statistic values. These individual statistics are used in pre-game evaluations . I remind you, this even goes for all of the children that are working for grades, individual scores are very important. This is just a given… except when it comes to schduling teams for qualifications.

Not to put too fine a point on it but if you’re doing a task that isn’t worth as many points, do you really deserve to be picked over a robot that can score more points and win more matches? Obviously it depends on the alliance captain’s strategy and the traits they’re looking for in an alliance partner. But sometimes the easier task is the more profitable one (climbing in 2020, 2019, 2017, gears in 2017, etc). This is why game analysis and strategy at the beginning of the season is so important.

16 Likes

That’s true, but you’re arguing for using the individual stats in lieu of wins and loses and using them for re-seeding/scheduling during qualification matches. No team sport does that. FRC is a team sport. Other posters have mentioned other robotics competitions that are not team oriented.

Are we going to get any real world examples of a top 10 robot finishing in the bottom 10 due to scheduling, or a robot that can do everything losing 7 of 12 qualification matches? You have said this happens at event after event so it shouldn’t be too hard to pull up multiple examples (or even one). Some concrete evidence of what you’re talking about would go along way towards bringing people around to your argument. Otherwise I’m just going to have to assume you’ve pulled these examples out of your covid.

To preface this, I am not super familiar with baseball/other sports stats so I could be entirely wrong about this and I apologise if I am. However I still 100% believe my other points still hold up even if I am totally wrong about what I am about to say.

I googled MLB stats, and got this fantastic list of stats tracked on each player by the MLB. I am not super familiar with baseball like I said, but after a brief peruse of the stats, it seems most (if not all) of them are records/averages of objective actions a player or team did. This is a professional league with near infinite resources to track these things at their disposal, and they are still only counting objective, concrete, physical actions by players or teams.

My point was that there are so many actions teams can take during and before a match that contribute to a win that are not objectively measurable, such as defense as an example(insert dpr joke). If a professional organization such as the MLB with millions of fans doesn’t track things that aren’t objective, it’s likely because it is near impossible to capture those things if at all. FIRST especially has so many different non-objectively trackable elements that add up to wins and losses, and for that reason, if those can’t be tracked, then the individual performance ranking system as a whole is severely flawed

5 Likes

Baseball uses a few stats to evaluate defense and in the public sabermetric discussion, I believe that no one defensive statistic is commonly accepted as a catchall, and some people and many teams still use qualitative observations (the eye test).

edit: this is just to add to the discussion, not agreeing or disagreeing

3 Likes

And ultimately the nature of competition is to win matches as an alliance. Ranking teams based on how much they win and how many individual RP objectives they can complete is a perfectly reasonable way to rank teams.

The measure of a great team is not how many fair fights they won. I can say without a doubt that none of the teams people consider to be shining examples of excellence got to where they are by winning a lot “equal” matches.

2 Likes

Match schedules should be designed for equality – if you switch any two teams’ places in the schedule it should not drastically affect their ranking. This isn’t a controversial idea, and has been implemented before, like Chezy Champs using pre-generated schedules and the introduction of RPs.

Match schedules should not be designed for equity – rankings will not correlate exactly to robot performance, but they should provide a method for higher performing robots to win matches and rise.

FRC is alliance based. The goal of each robot should be to maximize their alliance’s chances of winning the match and/or RP. This can be, but isn’t always the same as what maximizes each individual robot’s individual contribution. If two robots have the same auto, the one that gets it more consistently will probably run it, even if the other robot runs it in other matches.

Individual tracking affecting rankings is inherently flawed because it promotes individual contribution above the alliance. FRC likes alliance based games. We’ve tried the opposite: 2015 had no WLT record, almost no interaction with alliance partners or opponents, and rankings very closely matched individual robots’ contributions – and it was one of the most disliked games.

FIRST tries make an exciting competition that balances scoring individually, working on an alliance, and competing against an opposing alliance. Some games skew too much one or the other (2015, 2017 (climbing), 2019 (defense)), and the best balance all three. If you don’t win 50% of your matches, or aren’t picked, or underperform your expectations, that’s on your team. Look at how well your robot plays all 3 aspects – do you consistently score points? do you work well on an alliance (climb second, shoot in auto from multiple locations, etc.)? do you hold up against opposing alliances (play under defense, play defense)?

Luck will always have some role. But instead of complaining about what you can’t control, learn to make your own luck.

17 Likes

“Luck is what happens when preparation meets opportunity.” - Lucius Annaeus Seneca

I refuse to believe that in every single match in every single season, a team has been declined opportunity.

Why should you care? If you’re doing your job as an alliance member people are going to pick up on that. And there’s only so many common factors within a trend of lost matches - we may as well prove we’re not one of them.

2 Likes

I’m a bit confused here. It sounds like @tmichals wants teams to be individually scored regardless of their alliance’s performance. Why would the way matches are scheduled need to be changed if a team’s alliance partner’s performance won’t affect their ranking?

Basically every reader going though this thread…

image

25 Likes

Opening CD for the first time this afternoon:

Except in very rare circumstances*, both sides of the field are always at the same disadvantage. You have the same likelihood of getting “that one ref” who thinks the robot is too fast and misses defense crossings, or the slightly misplaced airship that borks your auto, or the fuel bin with a sticky release, or the hatch panel that doesn’t stick quite as well, or the champs FMS lag that makes you drive backwards off the HAB across the course of the competition.
I will also take this and point out that more quals matches = more statistical “smoothing” and reduced risk of a particular team being singled out by all these undesirable circumstances. plz gib Cali districts already

*those are typically due to major malfunctions that get caught quickly and almost always warrant a replay, usually sensors not doing their job recently.

11 Likes

I’d argue 2015 is disliked despite the more accurate or individual rankings and not because of it.

3 Likes

all this thread shows is the importance of properly scouting and looking at the schedules to determine if rank accurately reflects performance. Teams may be ranked low due to bad schedules, or ranked high due to a good one. And the best of the best will never leave it to a schedule, its why teams build things like buddy climbs, to lower their reliance on partners.

5 Likes

There is another way to reduce luck. Unfortunately, some in this thread seem like they want to use it.

That particular way is to be a person/team that cannot be worked with. That doesn’t play well with others. That is a “my way or highway” person/team.

Teams with that reputation may find themselves left out in the cold after alliance selections time after time after time. And I’ll remind everybody that if you have a team number on you CD username, what you say can reflect on your team. I personally try to avoid that generalization–but I can’t say I’m altogether successful.

By the way, in pro (and collegiate) sports, individual stats don’t make a difference in the games played or the rankings of the teams. Not. One. Bit. The collection of the individual stats in a game for a team matters (because that’s what helps you win and all), but a player’s individual stats over the season are mainly worth it for negotiating the player’s next contract (AKA, getting paid more).

12 Likes

Let’s take the NFL as an example, Ya, the worse team gets the first pick in the draft… after the current season is over in an effort to help them improve for the coming season… They however don’t get the bye week and home field advantage in the current season, which is what you should be comparing to here in regards to alliance selections.

Top tier teams in the NFL earn their way to the bye week and home field advantage in the playoffs
of the current season by being the highest ranked team at the end of the current season, and FRC teams earn their way to the #1 seed and having the first pick at alliance selections, by being the highest ranked team at the end of the current event.

I’ve seen you use so many incorrect comparisons throughout this thread, please actually do some research in how different things work before claiming its the same thing.

4 Likes
11 Likes

Are you saying that the individual statistics have an effect on the teams’ rankings, or on their current year’s schedules? Like the baseball team with the home run king gets a golden ticket to the playoffs?

I thought you wanted the later rounds in qualification to have team matchups based on the current team statistics. Or have you changed that position?

An option teams have is that if any individual event is “too hard” because it attracts too many “powerhouse teams”, the team can choose a different event. Teams choose the events they want to attend (within limits); they are not arbitrarily assigned.

4 Likes

No, it isn’t.

Qualification points are rarely used by teams that scout to pick other teams for the playoffs. This means that in reality, a team’s qualification record doesn’t matter all that much. If a team is getting downhearted because they aren’t winning a ton of qualification matches, then the mentoring group isn’t doing a good job of teaching how the system works.

If you are calling the concept of being picked for your robot ability artificial, then I really don’t know what to say to that. It’s a competition and someone has to lose. And the picking is almost never ‘artificial’. With our scouting we can usually predict with high accuracy what robots are going to be picked. And we know why. It’s based on a robot’s performance. Not their win/loss record.

20 Likes