Compatition Ranking

[Skyehawk]

With the announcement of Wins/losses no longer be tracked through qualification matches we have a situation to figure out.

Traditionally teems have been awarded qualification points via wins, ties, and losses (2-1-0 respectively), this year however the first order of sorting is not number of qualification points, but rather something called Qualification Average. The orders of seeding (2-7) are what you would expect; co-op score, auto score, etc.

I have no solid evidence on how Qualification Average works, the rule that this refers to is 5.3.4 in the game manual.

Any Ideas?

[cgmv123]

Read the paragraph before Section 5.3.4 starts.

[George A.]

With the announcement of Wins/losses no longer be tracked through qualification matches we have a situation to figure out.

Traditionally teems have been awarded qualification points via wins, ties, and losses (2-1-0 respectively), this year however the first order of sorting is not number of qualification points, but rather something called Qualification Average. The orders of seeding (2-7) are what you would expect; co-op score, auto score, etc.

I have no solid evidence on how Qualification Average works, the rule that this refers to is 5.3.4 in the game manual.

Any Ideas?



You take the average score of all your matches and that's your qualification average (QA). That's what's used for ranking all the teams. IF by some large mathematical improbability teams are tied with the same QA then they are sorted by the tiebreakers referred to in 5.3.4

[GaryVoshol]

Why use the average? Why not just the total accumulated points? Every team is scheduled for the same number of matches, so why do we do the (extra) math?

[Skyehawk]

Makes sense, so I guess best form of defense is to score as much as possible...

[ATannahill]

Why use the average? Why not just the total accumulated points? Every team is scheduled for the same number of matches, so why do we do the (extra) math?
This has irked me for years.

[Skyehawk]

Why use the average? Why not just the total accumulated points? Every team is scheduled for the same number of matches, so why do we do the (extra) math?

This is not true, in some events the total number of teams is not divisible by 6, this leads to some teams playing "Surrogate Matches" (rule 5.3.2)

That being said (since the surrogate matches do not count towards the teams qualification average) why not use the total number of points scored? I believe the answer to this is just to have smaller numbers. But this in turn introduces another problem: the average is rounded to n decimal places. Therefore co-op points, auto points, etc, come into play.

[ATannahill]

Per page 52. The total number of MATCH Points earned by a Team throughout their Qualification MATCHES, divided by their number of assigned
MATCHES (excluding any SURROGATE MATCHES), then truncated to two decimal places, is their Qualification Average (QA).

We are not counting the surrogate match, so we would not add the points to their total.

[Joe Ross]

Why use the average? Why not just the total accumulated points? Every team is scheduled for the same number of matches, so why do we do the (extra) math?

Because in the middle of competition not all teams have played the same number of matches.

[Skyehawk]

Because in the middle of competition not all teams have played the same number of matches.

But the rankings don't matter until alliance selection.

[Marc P.]

Why use the average? Why not just the total accumulated points? Every team is scheduled for the same number of matches, so why do we do the (extra) math?

I think it is to normalize the "perfect storm" random chance alliances, where 3 super powerful robots are paired by the all mighty randomized partner algorithm. It dilutes that one magic match where the stars align and the score ends up 3 times higher than your other matches, mainly because of a particular mix of robots and not so much the individual robot's performance. The average gives an adjusted and IMO more accurate portrayal of actual robot contribution over multiple matches, rather than cumulative score, where one magic match can boost a team's ranking beyond their typical performance.

[Bryan Herbst]

But the rankings don't matter until alliance selection.

They don't directly impact any aspect of the competition until alliance selection, but it is nice to know where your team lies prior to that.

It is also useful information when talking to other teams about alliance selection.

[Anteprefix]

I think it is to normalize the "perfect storm" random chance alliances, where 3 super powerful robots are paired by the all mighty randomized partner algorithm. It dilutes that one magic match where the stars align and the score ends up 3 times higher than your other matches, mainly because of a particular mix of robots and not so much the individual robot's performance. The average gives an adjusted and IMO more accurate portrayal of actual robot contribution over multiple matches, rather than cumulative score, where one magic match can boost a team's ranking beyond their typical performance.

In what way does the average adjust the robot's contribution? The QA is just the cumulative score divided by a factor.

[XaulZan11]

Because in the middle of competition not all teams have played the same number of matches.

If that is the reason, why change it now? FIRST has always done total qualification points (or whatever the points were called in 2010).

Perhaps it is a way to normalize scoring across events for district points?

[aditya29]

But the rankings don't matter until alliance selection.

But mid-tournament rankings can affect strategy in upcoming matches. In a very simple (and naive case), if you're quickly trying to determine if you still have a chance at making the top 8, you may want to see the relative strengths of the teams ahead of you, and tracking the averages is more helpful than total scores because it helps normalize across # matches played. (Clearly you should be keying in on the robot's abilities rather than just their total points scored w.r.t your match strategy, but ranking certainly is an important indicator as well).

[Skyehawk]

I think it is to normalize the "perfect storm" random chance alliances, where 3 super powerful robots are paired by the all mighty randomized partner algorithm. It dilutes that one magic match where the stars align and the score ends up 3 times higher than your other matches, mainly because of a particular mix of robots and not so much the individual robot's performance. The average gives an adjusted and IMO more accurate portrayal of actual robot contribution over multiple matches, rather than cumulative score, where one magic match can boost a team's ranking beyond their typical performance.

Consider this scenario:
Last match of qualifications:
Team A: Average Score:98pts/match
Team B: 100pts/match
Team C: 102pts/match

Team D: 98pts/match
Team E: 100pts/match
Team F: 102pts/match

Seems like a fair match-up right? but the stars align for team A-B-C and...
Final score:A-B-C 300 | D-E-F 100

Even tough these teams were identical in their stats team A-B-C pulled out a significant win. The average scores of team D-E-F stay approximately the same, and while the teams were tied in the standings before this match took place teams A,B, and C are now all ahead of teams D,E, and F.

My point being: since the same number of qualification matches are played (excluding surrogates) why does averaging matter?

[Kpchem]

Under this system of averaging, it is easier to compare between events that run different numbers of matches. This could be especially valuable if events within a single district region have to run different numbers of matches.

That's my guess anyway.

[Skyehawk]

My updated insight is as follows- the averaging simply allows for an easier way to compare teams that have different number of matches played. For example:
Team A has a QA of 100pts/match and 4 games played
Team B has a QA of 110pts/match and 3 games played

This is a lot easier to compare than
Team A has 400 points total, and 4 games played
Team B has 330 points total, and 3 games played

[Ian Curtis]

If that is the reason, why change it now? FIRST has always done total qualification points (or whatever the points were called in 2010).

Perhaps it is a way to normalize scoring across events for district points?

The one nice (depending on your perspective) thing about the old method was more teams got to be in 1st place (or other high ranking spots) for a short while due to the extra noise. Now since everything is normalized, there will be a lot less movement in the rankings. A good thing from a usability standpoint, but not as many teams will get a short-lived boost from seeing their names at the top of the scrolling chart.

[cbudrecki]

My big question is if we are taking an AVERAGE score, then why leave out a surrogate match. Why not have that team's average be based on 11 matches instead of 10 (or whatever the case may be)? I can see in a Win/Loss scenario, or even if we were going with total points, but going off an average?

My fear is that teams will adopt the strategy of sabotaging their surrogate matches. If it doesn't count against them, why not lower the other teams' average? Granted, definitely NOT in the spirit of FIRST, but you know teams will do it. If we did away with surrogates this year, and let a few teams have an extra match figured into their average, I don't see where it would hurt.

[cbudrecki]

My other question; Is ranking averaged match-by-match as the day goes on, or is it always calculated by the number of matches scheduled?

Example: After 3 matches, a team has 330 points. They are scheduled for 10 Qualifier matches. Is their current average 110 (330/3) or 33 (330/10)?

[dellagd]

I Initially thought it was just to make the ranking nicer to watch, but it appears this doesn't happen...

Section 5.3.3:
The total number of MATCH Points earned by a Team throughout their Qualification MATCHES, divided by their number of assigned MATCHES (excluding any SURROGATE MATCHES), then truncated to two decimal places, is their Qualification Average (QA).
The key word is assigned. It does not say matches played at that point.

This leads me to believe that the divisor of the average is constant through the whole event, actually making the whole averaging thing pointless for viewing purposes or... anything... over just a total. This seems really weird though, I'm guessing they don't mean that since, well, it doesn't make any sense.

[MikeE]

I Initially thought it was just to make the ranking nicer to watch, but it appears this doesn't happen...

Section 5.3.3:

The key word is assigned. It does not say matches played at that point.

This leads me to believe that the divisor of the average is constant through the whole event, actually making the whole averaging thing pointless for viewing purposes or... anything... over just a total. This seems really weird though, I'm guessing they don't mean that since, well, it doesn't make any sense.

I don't have any special knowledge but I expect the Pit Display will show average score throughout the competition. From a tournament perspective Qualification Average doesn't have any role before final ranking, although in practice we all want to know where we stand throughout the competition.

[MikeE]

Under this system of averaging, it is easier to compare between events that run different numbers of matches. This could be especially valuable if events within a single district region have to run different numbers of matches.

That's my guess anyway.

I hope that District events stay with the standard 12 matches for all teams. This guarantees no surrogates and for smaller events also keeps a feasible inter-match minimum.

[Skyehawk]

My big question is if we are taking an AVERAGE score, then why leave out a surrogate match. Why not have that team's average be based on 11 matches instead of 10 (or whatever the case may be)? I can see in a Win/Loss scenario, or even if we were going with total points, but going off an average?

My fear is that teams will adopt the strategy of sabotaging their surrogate matches. If it doesn't count against them, why not lower the other teams' average? Granted, definitely NOT in the spirit of FIRST, but you know teams will do it. If we did away with surrogates this year, and let a few teams have an extra match figured into their average, I don't see where it would hurt.

You wouldn't want to throw a surrogate match. Not all teams look closely enough while scouting to realize that a team is playing a surrogate match. A bad performance in match number 4 (typically where surrogate matches are inserted) may put the team that might want to pick them in a position where they said "But they did terrible in match Number 4". And thus may not consider picking the team when Saturday afternoon comes around.