top 25 teams by category

[Caleb Sykes]

Using the data from my recently released database (http://www.chiefdelphi.com/forums/showthread.php?threadid=145892), I thought it would be fun to share the top 25 teams by category.

Subtracted tower strength
rank team subtracted tower strength
1 2056 7.18
2 233 6.64
3 2122 6.28
4 148 6.15
5 610 5.41
6 359 5.40
7 25 5.38
8 118 5.34
9 5254 5.27
10 4039 5.26
11 1986 5.07
12 3683 5.07
13 2619 5.00
14 6086 4.99
15 1114 4.95
16 708 4.91
17 2481 4.85
18 4982 4.80
19 3528 4.77
20 238 4.75
21 254 4.74
22 4334 4.73
23 2486 4.73
24 1648 4.68
25 2655 4.66


Auto points
rank team auto Points
1 2056 20.66
2 1114 19.78
3 1747 19.65
4 987 18.58
5 1986 17.45
6 4451 16.99
7 234 16.79
8 1736 16.50
9 3310 16.38
10 1619 16.08
11 3754 16.07
12 2481 15.75
13 359 15.67
14 2877 15.55
15 179 15.06
16 2013 14.64
17 1678 14.58
18 2046 14.44
19 5038 14.38
20 1288 14.16
21 3339 14.08
22 2122 14.04
23 4039 13.96
24 2537 13.85
25 5137 13.82


Teleop boulder points
rank team teleop Boulder Points
1 2056 31.33
2 148 28.34
3 118 25.02
4 4334 24.32
5 330 23.73
6 359 23.60
7 4982 22.57
8 1024 22.33
9 254 21.77
10 3238 21.74
11 2502 21.14
12 2771 20.51
13 4564 20.19
14 2481 19.89
15 876 19.79
16 5038 19.75
17 2393 19.71
18 2468 19.44
19 1986 19.35
20 133 19.04
21 180 18.92
22 359 18.85
23 1241 18.69
24 1024 18.63
25 1501 18.36


scale points
rank team teleop Scale Points
1 1876 16.13
2 1418 15.00
3 337 14.12
4 3937 13.94
5 2200 13.20
6 16 12.08
7 3683 11.81
8 2081 11.68
9 5413 11.09
10 4573 11.02
11 708 10.98
12 74 10.97
13 5509 10.96
14 1876 10.93
15 21 10.89
16 1983 10.87
17 217 10.64
18 1094 10.58
19 3990 10.37
20 71 10.27
21 3009 10.23
22 1662 9.49
23 876 9.40
24 1533 9.39
25 5172 9.23


2056's results are astounding, they are at the very top of 3 of these lists. That caliber of play in so many aspects of the game is completely unmatched.

There are very few consistent climbers. A 50-50 climber puts you in the top 50, and you can probably count on one hand the teams that have climbed in every qual match.

There are very few consistent boulder scoring autos out there. Some of the teams on this list may have never even scored boulders in auto. It is important to recognize that although some of the teams with boulder scoring autos may be scoring 20 autonomous points, they often are contributing far less than this if they require the low bar, because they are taking that low bar spot away from other teams who might not otherwise get a cross.

[JohnFogarty]

Are teams with 100% breach rates common?

[Caleb Sykes]

Are teams with 100% breach rates common?

That is an excellent question that my current database is ill-equipped to satisfactorily answer. Here's my best guess, maybe in the future I will have the tools to do better.

The average breaching contribution per team is .193. That is, three average teams together would average about 57% of a breach. Thus, teams with breach contributions of 1-(2*.193) = .614 could be expected to breach every match. 54 teams meet that criterion.

Obviously, this method has some major flaws, but it should give a decent ballpark estimate.

[carpedav000]

Congrats to 1024 for being in the same list twice! :D

[jajabinx124]

Congrats to 1024 for being in the same list twice! :D

I'm guessing one of them is 1023.. typo maybe.

[Caleb Sykes]

Congrats to 1024 for being in the same list twice! :D

I'm guessing one of them is 1023.. typo maybe.

No typo. 1024 contributed 18.63 boulder points per match at the Tippecanoe district event, and 22.33 boulder points per match at the Walker-Warren district event.

[carpedav000]

I'm guessing one of them is 1023.. typo maybe.

Now the question is.... Which one is 1023? Ima guess the lower one because 1024 would be above 2771

[carpedav000]

No typo. 1024 contributed 18.63 boulder points per match at the Tippecanoe district event, and 22.33 boulder points per match at the Walker-Warren district event.

Then wouldn't you average the two?

[Caleb Sykes]

Then wouldn't you average the two?

That would be one way to do it.

I was originally going to list both the team number and the event in these lists, but the code tags provided by CD's text editor were being annoying, so I just dropped the event information.

[dirtbikerxz]

You should do a category of bots with the highest defense (defense breaching) points.

[Caleb Sykes]

You should do a category of bots with the highest defense (defense breaching) points.

As you wish.

rank team crossing points
1 1418 26.00
2 4061 25.76
3 11 24.57
4 1418 23.87
5 1746 23.56
6 3102 23.43
7 2637 23.36
8 4931 23.36
9 3937 23.34
10 4680 23.05
11 4468 23.05
12 4384 22.91
13 3999 22.91
14 2642 22.91
15 4680 22.74
16 5632 22.36
17 5587 22.33
18 2974 21.99
19 16 21.93
20 79 21.92
21 4911 21.83
22 818 21.81
23 4935 21.80
24 1797 21.72
25 3005 21.68

Be careful interpreting these though, remember that it is much easier to get high crossing points at weaker events.

[lurker]

Does this data only include qualification numbers?

[Caleb Sykes]

Does this data only include qualification numbers?

Yes.

[Anthony Galea]

As you wish.

rank team crossing points
[snip]
10 4680 23.05
[snip]
15 4680 22.74
[snip]


Be careful interpreting these though, remember that it is much easier to get high crossing points at weaker events.

For those of you who haven't seen 4680's robot yet, I would definitely check them out. They are probably the leading candidate for MCC this year IMHO. They have been the #6 pick and #5 captain at their first two events, respectively.

[jspatz1]

Summing Caleb's four categories into a cumulative rank, where the points for each category =26-rank, the top 25 are:

2056 - 75
359 - 53
148 - 46
1986 - 43
118 - 41
1114 - 35
2481 - 35
3683 - 33
2122 - 27
4982 - 27
2122 - 27
4334 - 26
708 - 25
1876 - 25
233 - 24
1418 - 24
1747 - 23
337 - 23
254 - 22
987 - 22
3937 - 22
610 - 21
330 - 21
2200 - 21
4451 - 20

[tindleroot]

Then wouldn't you average the two?

Average would not necessarily be useful since every event is different - then again, we are trying to compare all of the teams worldwide, which have competed in dozens of different events. My personal choice for list would probably be to include teams only once, with their highest calculated performance, in order to have a unique top 25.

[Bob Steele]

How does a team average higher than 15 in scale points?

?

[Mike Schreiber]

How does a team average higher than 15 in scale points?

?

If they play X matches there are more than X scales in those matches. Doesn't matter who actually scaled, this is all alliance based.

[Caleb Sykes]

How does a team average higher than 15 in scale points?

?

These values are calculated in the same way that OPR values are calculated. The numbers shown indicate a given team's average contribution to a given scoring category, which is not necessarily the amount that they themselves score in that category. The amount greater than 15 from 1876 could just be noise, but it could possibly indicate something deeper. For example, 1876 might have scaled from the side rungs, which left the easier center rung available for other climbers.

[Spartan710]

At our event this weekend we went 13 for 15 in scaling. So to say that those teams are better isn't really correct. It just means that their alliances were better.

[Breakaway3937]

At our event this weekend we went 13 for 15 in scaling. So to say that those teams are better isn't really correct. It just means that their alliances were better.

Just remember that it depends on how you play the game. We go into every qualification match wanting to take down the tower and breach. It is hard to quantify a team this year by statistics since strategies are so different. Point in case, we are high in the rankings for defense points because we knew the value and ease of the breach ranking point so we made sure to tackle the defenses during qual matches and when we had alliance members that struggled with certain defenses, we would go and cover those thus taking our shooting scores down since we were no longer shooting. The key to this year's game is creating a robot that can fill the void where you may be missing alliance members. With this year's game, the numbers may not show how great your robot is in one specific area, but if you can cover the field in a case by case scenario, the ranking points sure will. It all depends on the quality of the regional. 2056 had a lot of opportunity to shot because of the quality of their alliance partners. If you are the best shooter, like 2056, and your other alliance partners can handle the necessary defenses, that takes a large load off of you. There are a lot of really good bots who may not be in any of these rankings due to varying their strategies match by match.

[Caleb Sykes]

At our event this weekend we went 13 for 15 in scaling. So to say that those teams are better isn't really correct. It just means that their alliances were better.

Firstly, these results only account for qualification matches. I'm not sure what your climb success rate was in quals, but the data potentially implies that your team was about 9/12.

Secondly, the calculated contribution results do a pretty good job of separating individual teams out from their random alliances given that the metric of interest is separable. Scale points are pretty separable, that is, whether or not a given robot on an alliance scales has little impact on whether another robot on this alliance scales. Crossing points are not so separable (which is why I did not originally post the crossing rankings) since a robot that can cross all defenses will almost never contribute 50 crossing points.

[Spartan710]

I was not trying to pick it a part. In quals we went 10/12 and in elems went 3/3. I love work you put in to this, so sorry if you thought i was try to pick it apart. the only reason i said that is because the top teams average was above 15 that's all.

[Caleb Sykes]

I was not trying to pick it a part. In quals we went 10/12 and in elems went 3/3. I love work you put in to this, so sorry if you thought i was try to pick it apart. the only reason i said that is because the top teams average was above 15 that's all.

No problem, I just used your post as an opportunity to clear up some common misconceptions about these calculated contributions.

The exact rankings of teams should be taken with a grain of salt. However, the general relative positions of teams should be correct.

[Bochek]

How can a team have more then 15 scaling points per match average?

[Matt_Boehm_329]

I'm interested in how you are calculating these? Some teams have negative values for boulders scored. Are you trying to run these numbers the same way OPR and CCWM are calculated?