Proposal for Wildcard Reform

[Kevin Leonard]

Originally Posted by SoftwareBug2.0

How often do you estimate this will generate results that are different than the current system? It seems like the finalist captain and first pick would still usually be the first two teams.

Do you know of any cases where the altered order would have made a difference? What's the largest number of wildcards that's ever been given out at an event?

At the 2016 Tech Valley Regional, the entire finalist alliance, including a backup robot made it to the championship event via the wildcard system. This included 1665, a competent backup robot whose defense was necessary to make it to the finals, but I doubt anyone would say they were more worthy of a championship bid than, for example, the captain of the third seeded alliance, 5236, or their first selection (and winner of the quality award), 2791.

Similarly, at the 2016 Finger Lakes Regional, the #1 seed was upset in the semifinals, despite having the two best scoring robots at the event, and 2791 yet again did not get a bid to championships.

Now I'm using the example of 2791, because I know them well, and they're a team that this regularly happens to. 2791 has missed championships by the tightest of margins at two regionals in 2012, one in 2013, and two in 2016, while also fielding a tremendously competitive robot in 2014.

Essentially this would come into play whenever a regional would generate enough wild cards to bring along the finalist alliance's third robot, or whenever a wild card slot is wasted, it could instead just go to the next team with the most points.

[bdaroz]

Including here as it's relevant... Here's the breakdown of wildcard slots at each regional from 2016 (excl Ontario):

https://drive.google.com/open?id=10P...hkP4sZFtrV zg

[SoftwareBug2.0]

Originally Posted by bdaroz

Including here as it's relevant... Here's the breakdown of wildcard slots at each regional from 2016 (excl Ontario):

https://drive.google.com/open?id=10P...hkP4sZFtrV zg

It's good to see some data. Looking at the 48 events in your spreadsheet it looks like there were 57 wildcards given out, so it's already averaging more than 1 per event. That's really interesting. My team has never been to an event where there were any wildcards.

[Caleb Sykes]

Originally Posted by SoftwareBug2.0

It's good to see some data. Looking at the 48 events in your spreadsheet it looks like there were 57 wildcards given out, so it's already averaging more than 1 per event. That's really interesting. My team has never been to an event where there were any wildcards.

Perhaps your experience reflects the fact that you are in the district system?

[SoftwareBug2.0]

Originally Posted by Caleb Sykes

Perhaps your experience reflects the fact that you are in the district system?

Certainly that's part of it. But wildcards were around before we were in a district. Also, there aren't any nearby original or sustaining teams or hall of fame teams.

[blueyoshi256]

Some sort of further wildcard distribution could be very useful in Minnesota. In 2016, both of the events 2823 attended had 2+ wildcards. Lake Superior had 2 (thanks 359!), and 10k had 3 (and possibly could have had 4 if they had awarded rookie all star). We were very fortunate to qualify off of the waitlist to champs after losing in semis (to 2052) at both regionals. Had we not been so fortunate, we would not have gotten to see our robot run at peak performance, win 10 qualification rounds, and have an amazing trip. With the extra wildcards added already, there will might be less impact in Minnesota than other regionals without some sort of further distribution rules.
That said, District Points may be unnecessarily complicated for this. I can't think of a simpler way, but it might exist.

[bdaroz]

I've updated the Google Drive Spreadsheet earlier in the thread to include a calculation of district event points for the Sacramento event.

I picked Sacramento as it's mid-season, and all 3 members of the finalist alliance already had bids to CMP.

Two caveats:

1. Google Sheets doesn't have an inverse error function, so I had to approximate it using a not-as-complicated formula as I'd like. As a result there are some minor errors in the values, but it doesn't appear to affect anything but the last placed team (got a 3 instead of a 4 -- I checked several other values with Wolfram Alpha).
2. I did not add the "season-wide" district points for rookie and 2nd year teams. I believe the intent here is to evaluate the play on the field, and those points were not listed on the proposal, but I did add judged award points as listed on the proposal (excluded Chairman's RAS, EI).
3. The DP tiebreaker is not factored in the sort on the spreadsheet

On a point basis... some interesting things:

• Winning alliance took positions 1, 2, and 5 in the DP ranking
• Finalist alliance took 3, 4, and 13th in the ranking. (3rd team was qual rank 26/60 with no awards)
• After all automatic and previously-awarded bids to CMP are factored in, 701 (sf 1st pick) would get the wildcard for the 2016 rules. Team 3250 (qf 1st pick) the additional under 2017 rules.
• The next teams, should other WCs had been generated would be 3669 (sf captain), 1056 (sf captain), 4094 (qf captain), and 4543 (qf 1st pick)

All in all I'm surprised the alliance captains didn't get more of a boost, and that award points played a significant result. Without them 701 and 3250 would not be anywhere close to in the running for a wildcard slot.

I'm interested to see what the rookie / 2nd year points boosts do. If someone wants to know, either post here, or PM me the team # ranges for rookies and 2nd year teams for 2016. I'm just too tired to go looking right now.

[bdaroz]

Originally Posted by Kevin Leonard

At the 2016 Tech Valley Regional, the entire finalist alliance, including a backup robot made it to the championship event via the wildcard system.

I updated my wildcard spreadsheet to add info about "Burned" wildcards. These are Wildcards generated that could not be given out because the finalist alliance had already received a bid to CMP. (Either by earlier event, other award, or getting a WC)

Only 4 events generated burned WCs under the '16 rules, not surprisingly mostly later events. There were, however, two events that generated 4 wildcards, and at Lone Star two of those were burned.

All in all under the '16 rules, we used 57 WCs and burned 5.

Under the '17 rules (thus far) we would have used 88 wildcards and burned at least* 22.

(* - If a team would have gotten a WC under the 2017 rules in an earlier event that is not reflected in the burned total. Thus, this is a minimum number without going through and recalculating WC generation regional-by-regional for the 2017 rules. The spreadsheet calculates the 2017 WC generation as a simple +1.)

What's clear is that the 2016 rules did not have a material effect on the number of teams that "lost" out due to burned wildcards (92% of WCs used), but if we used the 2017 rules that would change with only about 80% of WCs being used.

Put another way, ~41% of the newly generated WC slots would be burned under the 2017 rules as they are.

[Jon Stratis]

Originally Posted by SoftwareBug2.0

How often do you estimate this will generate results that are different than the current system? It seems like the finalist captain and first pick would still usually be the first two teams.

Do you know of any cases where the altered order would have made a difference? What's the largest number of wildcards that's ever been given out at an event?

If you want to look into this more, you can use the MN State Championship rankings, which are based on district points (for the teams first event): http://mnfirst.org/docs/2016/States_2016.pdf

Combine that with the wild card slots, and you'll at least have something to look at. The only thing it doesn't give you is the points for a team's second event, which may throw things off a little.

Originally Posted by bdaroz

Including here as it's relevant... Here's the breakdown of wildcard slots at each regional from 2016 (excl Ontario):

https://drive.google.com/open?id=10P...hkP4sZFtrV zg

[MasterMentor]

Originally Posted by bdaroz

I updated my wildcard spreadsheet to add info about "Burned" wildcards. These are Wildcards generated that could not be given out because the finalist alliance had already received a bid to CMP. (Either by earlier event, other award, or getting a WC)

I think looking at "Burned" wildcards is contrary to the point of the exercise. As Frank pointed out, the goal of adding an additional wildcard to each regional is to increase the participation of Regional teams at the Championship event. As shown in your awesome spreadsheet, 31 more Regional teams will be able to move forward that previously would not have. So what if the wildcards are burned - maybe that's factored into the reasons why FIRST added it (maybe they wanted ~30 more Regional teams and not ~50 more Regional teams). It also, in my opinion, increases the "fairness" of the wildcard system for early Regionals that would not have otherwise benefitted from wildcard-generating teams from previous performance in attendance - at least, in the circles I run in we typically scoff at the Wildcard system because the later Regionals in the season are the ones that benefit from it mostly. #WildcardBlues #EarlyRegionalsMatter

-George

[bdaroz]

Originally Posted by MasterMentor

I think looking at "Burned" wildcards is contrary to the point of the exercise.

Not directly, no, but an interesting view of what will happen under the 2017 rules as listed at this point.... It's also something that came up at NYTVR when all 4 members of the finalist alliance were announced as eligible for CMP. (It wasn't clear until we did some digging as to why, and initial thoughts questioned if there should have been more.)

It's somewhat more relivant under the proposal here because this would ensure a distribution of all wildcards under the points model and not artificially cap the number of potential wildcard recipients at three.

If the rest of your argument truly reflects FIRST's intentions, they want more representation, but not that much more, than they could tweak the proposal here to cap at x wildcards per regional.