Interesting observation on stats for this year (to the end of March)

[maxnz]

I was looking over the year aggregate stats for defense damaging this year and I noticed something interesting:

The percent of successful damages to class C defenses goes down in playoffs, by about 0.58%.

It's nearly negligible, but I still found it interesting.

Qualifications:
3772/12656       29.8%

Playoffs:
776/2656         29.22%

[MaGiC_PiKaChU]

I find it interesting that there has been exactly 10000 more qualification matches than elimination matches. What a coincidence

[GreyingJay]

Probably because nobody has a very fast Class C breach. I saw a few robots try in quals at North Bay, but they're typically slow and finicky, and the floppiness of the drawbridge makes it harder. The most reliable way to go through the drawbridge and sally port is to have it opened for you from behind.

I think in elims people realize that the time spent for two or three robots to coordinate a Class C defense crossing, is better spent having those robots line up an extra few goals instead. There are generally enough low bar robots that the breach can always be obtained even while ignoring the Class C defenses.

[maxnz]

Originally Posted by MaGiC_PiKaChU

I find it interesting that there has been exactly 1000 more qualification matches than elimination matches. What a coincidence

I think you mean 10000.

Also, the total number of opportunities vary between defense classes. Probably just some data glitch. But it still is about a 10000 match difference.

Qualifications:
A - 12663
B - 12665 
C - 12656
D - 12655

Playoffs:
A - 2660
B - 2656
C - 2656
D - 2652

[Eugene Fang]

Originally Posted by maxnz

I think you mean 10000.

Also, the total number of opportunities vary between defense classes. Probably just some data glitch. But it still is about a 10000 match difference.

Qualifications:
A - 12663
B - 12665 
C - 12656
D - 12655

Playoffs:
A - 2660
B - 2656
C - 2656
D - 2652

Sometimes the FIRST API returns "unknown" for a defense. I believe this happens because recording what a defense is isn't required for a match to begin.

[Ari423]

Right now a z test of the data yields a p-value of .273 (i.e. very not statistically significant). That means there is a 27.3% chance that this is just random chance causing the gap, not an actual correlation. Until we get more data or the gap grows substantially, we can't conclude that there is any correlation rather than just random chance. Statistics ftw.

[Citrus Dad]

Originally Posted by maxnz

I was looking over the year aggregate stats for defense damaging this year and I noticed something interesting:

The percent of successful damages to class C defenses goes down in playoffs, by about 0.58%.

It's nearly negligible, but I still found it interesting.

Qualifications:
3772/12656       29.8%

Playoffs:
776/2656         29.22%

There's more boulder shooting in the playoffs, and less focus on scoring through defense crossings. Class C slows down shooting teams too much.

[Ether]

Originally Posted by Ari423

Right now a z test of the data yields a p-value of .273 (i.e. very not statistically significant). That means there is a 27.3% chance that this is just random chance causing the gap, not an actual correlation. Until we get more data or the gap grows substantially, we can't conclude that there is any correlation rather than just random chance. Statistics ftw.

What "gap" are you referring to, and to whom are you responding?

There's no quoted context for your post, and it's linked to post #5

[mjc49]

Originally Posted by GreyingJay

Probably because nobody has a very fast Class C breach. I saw a few robots try in quals at North Bay, but they're typically slow and finicky, and the floppiness of the drawbridge makes it harder. The most reliable way to go through the drawbridge and sally port is to have it opened for you from behind.

I think in elims people realize that the time spent for two or three robots to coordinate a Class C defense crossing, is better spent having those robots line up an extra few goals instead. There are generally enough low bar robots that the breach can always be obtained even while ignoring the Class C defenses.

I agree with this assessment. The class C defenses typically take a lot of time relative to the amount of point that can be attained. If a bot can't do it solo, it is not efficient to have two bots tied up to get 5 points.

[Keefe2471]

I believe the importance of class C in Elims was discussed on Gamesense this last week. What I came away with was the opinion that as alliances increase in skill and start capturing and breaching every match, regardless of defense, those 10 points will be more and more critical for securing the win. So I would expect those stats to start shifting in the opposite direction as we go forward. I wouldn't expect that championship level alliances leave those points on the field.

I do agree that doing it solo is the best way to do it, I think that teams like 118 and 1197 that can do those defenses quickly (or in auto!) will be more valuable picks than comparable teams shooting wise.

[Jean Tenca]

I think one contributing factor is that in quals you more often have to deal with a random robot getting stuck or dying on a defense, which forces the alliance to attempt the class C defense. By eliminations, failure prone robots have been filtered out and teams don't bother with class C due to time/difficulty/specialization.

[Ari423]

Originally Posted by Ether

What "gap" are you referring to, and to whom are you responding?

There's no quoted context for your post, and it's linked to post #5

I was referring to the gap between the percentages of C class crossings in quals vs elims. There just isn't a big enough difference in the percentages given the sample size to be statistically significant, even if there are many perfectly valid reasons why there may be less C class crossings in elims.

I was responding to the OP (or just trying to provide added insight in general). If my post was linked to post#5, it wasn't intentional.

[Ether]

Originally Posted by Ari423

I was responding to the OP (or just trying to provide added insight in general). If my post was linked to post#5, it wasn't intentional.

To respond to the OP (and link to it), go to the OP post and click the "Quote" button. That will link your post to the OP, and provide quoted context.

If you just click "Post reply" it will reply (and link to) to the most recent post, not the OP. With the wrong linking, and no quoted context, sometimes it's not clear what your post is referring to.

[M217]

Random guess, but it could also be that Drawbridge seems to me to be more common in Elims than Quals, which could be contributing.

No stats to back this up at all, obv.

[Caleb Sykes]

Originally Posted by M217

Random guess, but it could also be that Drawbridge seems to me to be more common in Elims than Quals, which could be contributing.

No stats to back this up at all, obv.

Here are the stats.

Drawbridge frequency in quals: 30.3%
Drawbridge frequency in playoffs: 27.2%

Source: http://www.thebluealliance.com/insig...#year-specific