District Championship Projections

[PVCpirate]

Last year, when I was running district point calculations for New England and posting rankings before the official ones came out, I realized I could estimate the amount of points it would take to qualify for the District Championship using the proportion of championship capacity to total teams. Since it ended up being a decent estimate for the final cutoff by week 3 or so, I've developed the method more and applied it to all 5 districts for this year. I'll be happy to answer any questions about how I got the numbers. I'll be updating this thread each week with new projections.

Here are the numbers I came up with for the final district score of the lowest ranked team to initially qualify for the district championship:

WEEK 2 PROJECTIONS
Michigan: 74 points
Indiana: 44 points
PNW: 70 points
MAR: 58 points
New England: 70 points

The team counts I used to calculate these were 351 for Michigan, 53 for Indiana, 149 for PNW, 120 for MAR and 167 for New England. I'm not entirely sure about the accuracy of the team counts, so if anyone has better or updated numbers, I would appreciate it if you could share them here.

DISCLAIMERS: These are not official in any way. Teams may or may not qualify for district championships regardless of the numbers I post. Please don't alter your competition strategy based on my projections, they are rough estimates at best.

EDIT: I reran the numbers with the team counts Mr V posted, Indiana drops to 38, while PNW and New England go up to 72.

[Mr V]

Thanks for doing this, I do get questions from people wondering how many points the will likely need to qualify for DCMP

Current team counts
152 for PNW
175 for NE
121 for MAR
346 for FiM
49 for IN

[rwodonnell]

EDIT: I reran the numbers with the team counts Mr V posted, Indiana drops to 38, while PNW and New England go up to 72.


I'd be interested in your method. I of course am hoping the number for NE drops to at least 69 (team 2876's total)! :-)

It surprised me that when you found there were more teams in NE than in your earlier projection, that the number went up. I would have thought that, with more teams to chew up points, it might go down from your initial estimate.

Also, are you factoring in the teams that are competing in multiple events, along with those coming in from outside the district? I realize that's a bit of work, but just curious.

[PVCpirate]

I'd be interested in your method. I of course am hoping the number for NE drops to at least 69 (team 2876's total)! :-)

It surprised me that when you found there were more teams in NE than in your earlier projection, that the number went up. I would have thought that, with more teams to chew up points, it might go down from your initial estimate.

Also, are you factoring in the teams that are competing in multiple events, along with those coming in from outside the district? I realize that's a bit of work, but just curious.

Let's see... When you add more teams to a district, a lower percentage go to the DCMP. So the cutoff moved higher up the list, increasing the score. You can see for Indiana, decreasing the number of teams makes it easier to get to the DCMP, so the cutoff moves down, decreasing the score.

So basically what I do is apply the percentage of teams that make it to the DCMP to the number of teams that have competed, and the team at that rank in the standings becomes the "cutoff". I then adjust the cutoff up one spot for each spot taken up by a Chairman's winner below the cutoff. If the team at the adjusted cutoff has played 2 events, their score becomes the point cutoff I reported. If they have only played one, I assume they will get the same score at their next event(big assumption, probably the biggest flaw), and add that to their score to get the cutoff.

Extra plays don't go into the standings, so I don't have to worry about that. Interdistrict plays aren't being handled correctly by the ranking system last time I checked, so for example 316 is incorrectly included in the New England standings. Now that I'm on spring break, I'll have time to correct for that.

[rwodonnell]

Okay, at least I think I understand the way you are attacking the problem and why the number went up. I know it's not much more than a hope on my part, but I think the number will be closer to the 68-70 range, and that we are on the razor's edge.


Extra plays don't go into the standings, so I don't have to worry about that. Interdistrict plays aren't being handled correctly by the ranking system last time I checked, so for example 316 is incorrectly included in the New England standings. Now that I'm on spring break, I'll have time to correct for that.

The current standings have fixed that problem. I believe the standings we're seeing now include only the NE teams, and there are indeed 175 of them.

[PVCpirate]

Yeah, should be around 70 by the end of the season, and remember that some teams usually can't go and so teams that just miss the first round of invitations can still end up going.

Yep looks like they took the MAR teams out of the NE standings. I'll have new projections up on Monday!

[PVCpirate]

I just ran the numbers post-Week 3, and the new projections are ready! I believe all the district teams have now played at least one event, so the cutoffs are just the DCMP capacities. This is basically the projection after everyone's first event. I'm not really sure why MAR's number dropped so far. The jumps for New England and PNW were mostly due to low ranking Chairman's teams. Next week is probably the first really accurate projection, as a significant amount of final scores will come in, giving me a much better idea of how teams perform over 2 events.

The above is mostly wrong, I did the projections without realizing there were still teams that hadn't played yet. I reran the numbers and found the projections correctly, shown below:

WEEK 3 PROJECTIONS
Michigan: ↓ 72 points
Indiana: ↑ 46 points
PNW: ↑ 76 points
MAR: ↑ 60 points
New England: ↑ 78 points

[Jscout11]

I just ran the numbers post-Week 3, and the new projections are ready! I believe all the district teams have now played at least one event, so the cutoffs are just the DCMP capacities. This is basically the projection after everyone's first event. I'm not really sure why MAR's number dropped so far. The jumps for New England and PNW were mostly due to low ranking Chairman's teams. Next week is probably the first really accurate projection, as a significant amount of final scores will come in, giving me a much better idea of how teams perform over 2 events.

WEEK 3 PROJECTIONS
Michigan: ↑ 76 points
Indiana: ↑ 46 points
PNW: ↑ 74 points
MAR: ↓ 50 points
New England: ↑ 76 points

There are still ~25 (20%) MAR teams yet to play their first MAR event.

[Mr V]

Per the FRC leader board there are 7 teams in the PNW District that have not played their 1st event yet.

[PVCpirate]

I know what happened, previously the leaderboards were only showing the teams that had played an event, now they're showing all teams. I only refreshed my Excel data queries without looking at the bottom of the standings pages. I'll redo my math, thanks for the heads up!

[MikeE]

I feel these suggested points thresholds are rather high.
To take the New England District as an example, 32 teams have now played both their scored events. If we just rank those 32 teams and cut the list at the same proportion that will go to District Champs (60 of 175) then we see that the 11th team has 70 points.
So if we assume that teams completing their two scoring events by week 3 are performing as a whole close to average, then the District Champ invitation threshold should be around 70 points.

However, as pointed out in previous years, teams that compete in their second event early tend to do much better than average since they are at a distinct advantage over other teams competing for the first time. We see this trend again, since the average score for all NE teams in their first event is 26.5pts, whereas the average for 2nd event teams is 36.0pts. This pattern is seen in all Districts at the end of Week 3. (There is also an argument that better teams compete early, but it has the same effect on the District Champs point threshold.)

District Event 1 Event 2
#teams #teams Average #teams Average
FiM 327 303 25.3 49 31.0
IN 49 49 27.4 15 33.6
MAR 121 98 26.3 11 35.9
NE 175 168 26.5 32 36.0
PNW 152 145 28.5 43 33.1


By the end of the season the averages for 2nd events are actually expected to be a little lower than 1st event averages, since teams are often also competing against non-point-scoring 3rd event teams at later events which reduces the points pool. So 2nd event scores will become progressively lower on average from Week 4 onward.

Therefore we can conclude that 70pts is highly likely an overestimate of the points needed to qualify for the NE District Champs. My current estimate is 63pts.

[Brandon_L]

There are differences in the amount of district points teams are getting according to the blue alliance and the district ranking website. Am I correct to assume that the district ranking website would have the correct point value?

Example - 2495 has 21 points at chestnut hill according to http://frc-districtrankings.usfirst.org/MAR but according to TBA has 24 points.

[Jscout11]

There are differences in the amount of district points teams are getting according to the blue alliance and the district ranking website. Am I correct to assume that the district ranking website would have the correct point value?

Example - 2495 has 21 points at chestnut hill according to http://frc-districtrankings.usfirst.org/MAR but according to TBA has 24 points.

The district ranking site has the correct values.

[Nathan Streeter]

Based on the current rankings, projecting points earned Week 5 based on past performance, and the fact that there are 9 teams who will be competing who won't be acruing points, I'm going to guess that the threshold for New England qualification is going to be 67pts, if there are 54 teams going to NE Championships. Obviously if several teams decline attendance that might come down.

[rwodonnell]

Based on the current rankings, projecting points earned Week 5 based on past performance, and the fact that there are 9 teams who will be competing who won't be acruing points, I'm going to guess that the threshold for New England qualification is going to be 67pts, if there are 54 teams going to NE Championships. Obviously if several teams decline attendance that might come down.
I hope this is much closer to the real cutoff (we're at 69 points) and I really think it is. Not only are there 9 teams competing whose points won't count, as you said, but there is one team (236) who is already at 73 points, and another 4 teams who have won Chairman's awards. That's really 14 teams (11 in Hartford alone) whose points don't matter for those teams trying to reach the unknown threshold (assuming that threshold is under the 73 points team 236 already has.)

By the way, I thought there were going to be 60 invitations this year.

[Nathan Streeter]

I hope this is much closer to the real cutoff (we're at 69 points) and I really think it is. Not only are there 9 teams competing whose points won't count, as you said, but there is one team (236) who is already at 73 points, and another 4 teams who have won Chairman's awards. That's really 14 teams (11 in Hartford alone) whose points don't matter for those teams trying to reach the unknown threshold (assuming that threshold is under the 73 points team 236 already has.)

By the way, I thought there were going to be 60 invitations this year.

Certainly there are several factors that may bring that cutoff down below the 70pt mark...

I have no idea what the invitation count will be... I've tried looking around for it but have come up empty-handed. Battlecry has gone up above 54 teams in recent years, but I figured keeping my guess based on a 54-team list would be safe. If it bumps up to 60 teams, then that obviously brings the point threshold down by probably 1-3 points from whatever it would be.

[gafftron]

Certainly there are several factors that may bring that cutoff down below the 70pt mark...

I have no idea what the invitation count will be... I've tried looking around for it but have come up empty-handed. Battlecry has gone up above 54 teams in recent years, but I figured keeping my guess based on a 54-team list would be safe. If it bumps up to 60 teams, then that obviously brings the point threshold down by probably 1-3 points from whatever it would be.

The invitation counts are actually posted in the Admin manual:

7.4.3 District Championship
Number of teams attending District Championship
FiM .................................................. .................................................. ......... 102
IN................................................ .................................................. ................ 32
MAR .................................................. .................................................. ......... 55
NE .................................................. .................................................. ............ 60
PNW............................................... .................................................. ............ 64


So it looks like WPI will be hosting 60 teams for the NE Championship.

[Nathan Streeter]

The invitation counts are actually posted in the Admin manual:

So it looks like WPI will be hosting 60 teams for the NE Championship.

Oops... I usually pride myself on familiarity with the manual too! :-/

In that case, I'm expecting the cutoff to be 65 points... +/- 1 pt. We'll see on Sunday, I guess!

[Mr V]

The "official" projection to qualify for the PNW DCMP is 58 points.

[PVCpirate]

Here's what I got using my method:


WEEK 4 PROJECTIONS
Michigan: ↓ 53 points
Indiana: ↓ 42 points(complete)
PNW: ↓ 47 points
MAR: ↑ 70 points
New England: ↓ 56 points

Honestly I think these are pretty useless this week, Michigan, PNW and New England all look too low and MAR looks too high. Figured I would do them anyway for continuity or something.

[PVCpirate]

There are differences in the amount of district points teams are getting according to the blue alliance and the district ranking website. Am I correct to assume that the district ranking website would have the correct point value?

Example - 2495 has 21 points at chestnut hill according to http://frc-districtrankings.usfirst.org/MAR but according to TBA has 24 points.

I noticed this too, I think the issue is that TBA isn't calculating the ranking points correctly. During the UNH district, they were reporting 12 ranking points for teams at the bottom who should get something like 4 points.

[Link07]

Here's what I got using my method:


WEEK 4 PROJECTIONS
Michigan: ↓ 53 points
Indiana: ↓ 42 points(complete)
PNW: ↓ 47 points
MAR: ↑ 70 points
New England: ↓ 56 points

Honestly I think these are pretty useless this week, Michigan, PNW and New England all look too low and MAR looks too high. Figured I would do them anyway for continuity or something.

I haven't run any numbers or projections myself, but I think I can say with confidence that 70 is too high for MAR.

[Orthofort]

I did a projection just for Michigan by giving each team that haven't played 2 events the same score they got in the first event and I got 64-66, depending on how good the teams are that haven't played any events yet. Because of this, I estimate for Michigan it will be 64, maybe 65. FYI I did this using the numbers on the official FIRST ranking page, because I noticed TBA wasn't calculating qualification points correctly.

What method are you using to get 53 for Michigan? I cannot see the point values being that low.

Also I could do the same for the other districts if there is interest.

[Richard Wallace]

I did a projection just for Michigan by giving each team that haven't played 2 events the same score they got in the first event and I got 64-66, depending on how good the teams are that haven't played any events yet. Because of this, I estimate for Michigan it will be 64, maybe 65.Your estimate seems good to me because it matches this (http://www.chiefdelphi.com/forums/showpost.php?p=1460868&postcount=12). Also, FiM leadership has already communicated to Michigan teams that they anticipate the cutoff will be around 63 points.

[IceStorm]

I did a projection just for Michigan by giving each team that haven't played 2 events the same score they got in the first event and I got 64-66, depending on how good the teams are that haven't played any events yet. Because of this, I estimate for Michigan it will be 64, maybe 65. FYI I did this using the numbers on the official FIRST ranking page, because I noticed TBA wasn't calculating qualification points correctly.



What method are you using to get 53 for Michigan? I cannot see the point values being that low.



Also I could do the same for the other districts if there is interest.


I do the same thing with giving points equal to first event. How do you calculate the teams that have not played at all?

[Orthofort]

I do the same thing with giving points equal to first event. How do you calculate the teams that have not played at all?

I use a range, where the lower bound is if none of the teams that haven't played make it in, and the upper bound is if a proportional amount of teams make it in from this pool. So for example, after week 3 there were 40 teams in Michigan that hadn't played, if we take the proportion 102 qualifying / 347 total teams then we can expect a maximum of 12 teams making it from this group, and a minimum of 0.

For what it's worth, I think it's fair to do this because in general, the teams that wait until the last couple of weeks to do their events aren't as experienced or might not even go to 2 events. This of course is not true for every team but I think is generally true.

[PVCpirate]

I did a projection just for Michigan by giving each team that haven't played 2 events the same score they got in the first event and I got 64-66, depending on how good the teams are that haven't played any events yet. Because of this, I estimate for Michigan it will be 64, maybe 65. FYI I did this using the numbers on the official FIRST ranking page, because I noticed TBA wasn't calculating qualification points correctly.

What method are you using to get 53 for Michigan? I cannot see the point values being that low.

Also I could do the same for the other districts if there is interest.

If you go to the first page I explained how I was doing it, I'm basically just taking the percent of teams that have competed, multiplying that by the DCMP capacity, and then using the team at that ranking spot to determine the projected cutoff. It seems to work best when most of the teams have competed once, but very few have gone twice.

[swootton]

I'd love to see your prediction for NE. WE have 66 point and know we are on the bubble. I'm expecting 65 to be the cutoff.

[SoftwareBug2.0]

Here's what I got using my method:


WEEK 4 PROJECTIONS
Michigan: ↓ 53 points
Indiana: ↓ 42 points(complete)
PNW: ↓ 47 points
MAR: ↑ 70 points
New England: ↓ 56 points

Honestly I think these are pretty useless this week, Michigan, PNW and New England all look too low and MAR looks too high. Figured I would do them anyway for continuity or something.

Yeah, I hate to pile on but the number for the PNW cutoff is not a good guess. The PNW district championship is a 64 team event and the 64th most points right now is already at 47.

[Andy A.]

For anyone who hadn't noticed yet, NEDC qualified teams are now indicated on the leaderboard at http://frc-districtrankings.usfirst.org/NE . The point cut off was 66.

Incidentally, three cheers to FIRST for the leaderboard site! It's a fantastic tool; fast to update, easy to navigate, easy to understand. After years of struggling with the typical FIRST web based tools this was a breath of fresh air.

[MikeE]

For anyone who hadn't noticed yet, NEDC qualified teams are now indicated on the leaderboard at http://frc-districtrankings.usfirst.org/NE . The point cut off was 66.

Incidentally, three cheers to FIRST for the leaderboard site! It's a fantastic tool; fast to update, easy to navigate, easy to understand. After years of struggling with the typical FIRST web based tools this was a breath of fresh air.

Last year there were several teams who declined to attend for various reasons, so I wouldn't be surprised if a few teams just below the threshold will see an invitation later this week.

[Christopher149]

The rankings leaderboard just updated for FIM, and the cutoff is 66 points (assuming no declines).