pic: 2016 Championship Alliance Selection Rates by Qualification Type

[AGPapa]

[cdm-description=photo]43940[/cdm-description]

[ATannahill]

[cdm-description=photo]43940[/cdm-description]

Do these numbers include the alliance captain at the Championship or not? The first sentence implies that they are excluded.

[AGPapa]

Do these numbers include the alliance captain at the Championship or not? The first sentence implies that they are excluded.

They include everyone who made it to the playoff rounds at the championship. The captain, 1st pick, 2nd pick and sub (even if the sub wasn't used).

[Brian Maher]

These are some really neat stats, thank you for sharing!

Question: did you pull your data from the FMS API or TBA, and how did you differentiate between waitlist and points teams in districts? As someone who uses the TBA API, I've found this to be problematic.

If it's not too much trouble, I'd love to see the breakdown by:

level of playoffs reached
captain / first / second / third picks

I think it could be interesting to compare with overall alliance selection rates.

[AGPapa]

These are some really neat stats, thank you for sharing!


Thanks! I found it interesting that the 1st picks did better than the captains.


Question: did you pull your data from the FMS API or TBA, and how did you differentiate between waitlist and points teams in districts? As someone who uses the TBA API, I've found this to be problematic.


I took the list of qualified district teams from http://frc-districtrankings.usfirst.org/ and removed all of the teams that had otherwise qualified. Then I took the top X teams (where X is the number of points slots for that district) and called those the district points teams and everybody else was waitlist. There was a pretty clear cut off from when the points spots end. It's a pretty tedious process so I just uploaded the spreadsheet as a whitepaper (https://www.chiefdelphi.com/media/papers/3280?) if you want to use if for something else.


If it's not too much trouble, I'd love to see the breakdown by:

level of playoffs reached
captain / first / second / third picks

I think it could be interesting to compare with overall alliance selection rates.

Here's a chart (http://imgur.com/gallery/486p0) showing where these teams were picked. Some of the group sizes were small and created a noisy graph, so I removed every group that had fewer than 10 teams in it.

[IKE]

Thanks! I found it interesting that the 1st picks did better than the captains.

...snip...

So, it makes sense if you look at the probability of the two roles. You can end up in the #1 spot as a decent, but not the top performing team. In fact you may be not even a top 10 team at an event.
That first pick is almost always the top or 2nd from the top (when top team is #1).

I know when doing the initial points proposal for FiM, being the "first pick" at an event had the highest correlation with being most competitive and/or most likely to win during a season.

What would be really interesting is to see this chart from 2009 through about 2012. The district Points section would have been extremely high correlation through that time period.

[Chris is me]

So, it makes sense if you look at the probability of the two roles. You can end up in the #1 spot as a decent, but not the top performing team. In fact you may be not even a top 10 team at an event.
That first pick is almost always the top or 2nd from the top (when top team is #1).

I know when doing the initial points proposal for FiM, being the "first pick" at an event had the highest correlation with being most competitive and/or most likely to win during a season.

What would be really interesting is to see this chart from 2009 through about 2012. The district Points section would have been extremely high correlation through that time period.

Another factor to consider is that this year, being an alliance captain required a slightly different skill set than being the consistently highest scoring team at the competition. Being able to guarantee breach and capture inflated the ranks of particular classes of robots, particularly highly mobile low goal scorers, so more of them were top seeds. This is in addition to the normal variance caused by strength of schedule, strategy, reliability, etc.

[Basel A]

Another factor to consider is that this year, being an alliance captain required a slightly different skill set than being the consistently highest scoring team at the competition. Being able to guarantee breach and capture inflated the ranks of particular classes of robots, particularly highly mobile low goal scorers, so more of them were top seeds. This is in addition to the normal variance caused by strength of schedule, strategy, reliability, etc.

As seeding has gotten more accurate over the years (largely because of more matches?), 1st picks have gotten less good compared to their captains. This effect you mention is probably the only reason they were better this year.

https://www.chiefdelphi.com/media/papers/2843

[Chris is me]

As seeding has gotten more accurate over the years (largely because of more matches?), 1st picks have gotten less good compared to their captains. This effect you mention is probably the only reason they were better this year.

https://www.chiefdelphi.com/media/papers/2843

As long as even some teams are carried, this gap will always exist. Every first pick* by the top seed is one of the top 3 robots at the event; but not every alliance captain is. Maybe that is offset by non-carried captains who declined from a low seed and won the event, I dunno.

*2014 is an exception to all of this because picks were very role based.