Galileo 2012!

[Gary Dillard]

Hmmmmmmmmmm SPAM was #1 seed at both of their regionals, finalist in one and winner in the other, 2nd highest OPR & World ranking in the division, Joe's simulation seems to like them, yet noone posting seems to think they are a contender. They must be flying under the radar. I have a feeling that they may have kept a few tricks in their bag just for the Championship. It's been 10 years since their last Einstein appearance; could this be their year?

[George1902]

Originally Posted by Gary Dillard

They must be flying under the radar.

I don't mind that too much, actually...

Good luck, everyone!

[Rashek]

I just wanted to post what my research has showed me about the teams on Galileo. I came up with this data by going down the entire listof the field. Next I looked up these teams on FRC Spyder. Anytime that scored 200+ in any of the categories I wrote down. After I figured out those teams I copied the information onto an excel spreadsheet.

The numbers I used were a mix from each of the competitions. For example: Say a team went to three competitions. At Competition A they had their best teleop points. At Competition B they had their best hybrid. At Competition C they had their best balancing. So I would used the teleop from A, the hybrid from B, and the balancing from C.

Next I added all the numbers together in order to come up with a grand total of points. I then arranged the grand totals in descending order. I came up with the following:

Team # HP BP TP Grand Total
2054 227 200 289 716
33 195 130 380 705
1918 235 140 269 644
25 175 120 329 624
2337 234 150 239 623
1718 166 150 260 576
2016 205 140 209 554
3322 166 100 285 551
180 175 120 209 504
2169 155 30 317 502
103 136 110 223 469
48 154 70 240 464
1323 132 50 240 422
801 116 90 215 421
399 93 110 206 409
868 148 50 211 409
772 137 50 204 391
148 97 50 208 355

I am well aware of the fact that FRC Spyder doesn't count the elimination matches, and therefore this spredsheet isn't 100% accurate. This is my first time doing anything like this so any feedback woul be greatly appreciated.

[flippy147852]

Originally Posted by Rashek

Next I added all the numbers together in order to come up with a grand total of points. I then arranged the grand totals in descending order. I came up with the following:

Team # HP BP TP Grand Total
2054 227 200 289 716
33 195 130 380 705
1918 235 140 269 644
25 175 120 329 624
2337 234 150 239 623
1718 166 150 260 576
2016 205 140 209 554
3322 166 100 285 551
180 175 120 209 504
2169 155 30 317 502
103 136 110 223 469
48 154 70 240 464
1323 132 50 240 422
801 116 90 215 421
399 93 110 206 409
868 148 50 211 409
772 137 50 204 391
148 97 50 208 355

I like the Michigan bias, first of all (6 of the top 10 are Michigan teams ) That being said, BP isn't a very accurate metric without adding some sort of coopertition compensation. I don't have the data in front of me right now, but I am sure that the rankings would change with coopertition.

[David Smellie]

Originally Posted by Rashek

I just wanted to post what my research has showed me about the teams on Galileo. I came up with this data by going down the entire listof the field. Next I looked up these teams on FRC Spyder. Anytime that scored 200+ in any of the categories I wrote down. After I figured out those teams I copied the information onto an excel spreadsheet.

The numbers I used were a mix from each of the competitions. For example: Say a team went to three competitions. At Competition A they had their best teleop points. At Competition B they had their best hybrid. At Competition C they had their best balancing. So I would used the teleop from A, the hybrid from B, and the balancing from C.

Next I added all the numbers together in order to come up with a grand total of points. I then arranged the grand totals in descending order. I came up with the following:

Team # HP BP TP Grand Total
2054 227 200 289 716
33 195 130 380 705
1918 235 140 269 644
25 175 120 329 624
2337 234 150 239 623
1718 166 150 260 576
2016 205 140 209 554
3322 166 100 285 551
180 175 120 209 504
2169 155 30 317 502
103 136 110 223 469
48 154 70 240 464
1323 132 50 240 422
801 116 90 215 421
399 93 110 206 409
868 148 50 211 409
772 137 50 204 391
148 97 50 208 355

I am well aware of the fact that FRC Spyder doesn't count the elimination matches, and therefore this spredsheet isn't 100% accurate. This is my first time doing anything like this so any feedback woul be greatly appreciated.

The primary problem with this method is that FRC Spyder calculates alliance scores not individual robot scores. As such this will produce a bias towards teams in deeper regional events (ie MSC and MAR). The top ten appears to confirm this bias as the top eight teams are from these two regions. When looking at this data, be sure to consider this fact.

[Rashek]

I am aware of both of these problems. This was just a last minute thing I put together.

[Cory]

Originally Posted by David Smellie

The primary problem with this method is that FRC Spyder calculates alliance scores not individual robot scores. As such this will produce a bias towards teams in deeper regional events (ie MSC and MAR). The top ten appears to confirm this bias as the top eight teams are from these two regions. When looking at this data, be sure to consider this fact.

The bigger problem is that you need to normalize to either a per match basis or a 10 match schedule, as many events got more or less than 10 qual matches.

[maverickfan138]

It would be interesting to see how that list would look with the rest of the division considered. For instance, 573 at MSC scored 234 HP, 170 BP, and 239 TP for a total of 613 points. That places them in between 2337 and 1718, and I'm sure there are other teams who would be thrown into the mix.

[Rashek]

I'm sorry 573 Somehow I must have forgot to put you guys into that list. I don't know how that happened.

[maverickfan138]

Originally Posted by Rashek

I'm sorry 573 Somehow I must have forgot to put you guys into that list. I don't know how that happened.

No worries. You did say it was a quick job so a slip up is understandable.

[John G]

Team 399 has successfully tested our "stinger" / balancing assist mechanism. Can't wait to use it. Only 5 more days!!!

[Travis Hoffman]

10 minute montage (we're gonna need a montaggggge) of Team 48 and (eventually) Team 3193 practicing long-short-long triple balancing - all the highs and the lows. Enjoy, if you dare: http://www.youtube.com/watch?v=Z1wugVzdhG0

[slijin]

Originally Posted by Travis Hoffman

10 minute montage (we're gonna need a montaggggge) of Team 48 and (eventually) Team 3193 practicing long-short-long triple balancing - all the highs and the lows. Enjoy, if you dare: http://www.youtube.com/watch?v=Z1wugVzdhG0

Looks nice!

Was there a reason you never tried reversing the orientations of one of the long bots so that the CG was closer to the center than the edge?

[Travis Hoffman]

Originally Posted by slijin

Looks nice!

Was there a reason you never tried reversing the orientations of one of the long bots so that the CG was closer to the center than the edge?

3193's weight is actually front-biased a bit, so they were in the proper orientation. Note what happened whenever we applied forward force to the center robot later in the video - which way did 3193 tilt?

They do not have a stinger or bridge tender in the rear of their bot. There is actually not much at all back there - the internal guts of their bot are laid out very differently from ours. Plus the pivoting arms let either of us shift the CG further toward the front.

They might have had a similar result driving up backwards and tilting their arm all the way back, but I think we prefer this configuration; plus it was more symmetric and therefore looked cooler!

48's CG is very slightly rear of center, but with the stinger, we kinda need to be pointing that way. As you can see, we can hang quite a bit of the rear bot off the bridge and be comfortably balanced. Tilting the arm forward helps with stability. You can be assured we'll be pushing ours as far forward as possible, into contact with the center bot during any triple opportunities - we promise we'll be gentle!

For the record, the center "bot" in practice (donated by Team 46 ) was close to a max length 28" widebot frame.

[IKE]

Originally Posted by Rashek

I just wanted to post what my research ....

Next I added all the numbers together in order to come up with a grand total of points. I then arranged the grand totals in descending order. I came up with the following:

Team # HP BP TP Grand Total
2054 227 200 289 716
...
I am well aware of the fact that FRC Spyder doesn't count the elimination matches, and therefore this spredsheet isn't 100% accurate. This is my first time doing anything like this so any feedback woul be greatly appreciated.

I really like that you are looking at the data but... this is kind of bad math gone wrong scenario. As others have stated, you have two factors contributing to a lot of possible error.

#1 12 matches at FiM & MAR versus 8-10 matches most other regionals. If you normalize per match, you will get a much better metric.

#2 Alliance partners. The metric above can be highly dependent on alliance partners. When a robot is scoring around 2X the average, in theory they are only responsible for 50% 0f the points (2X+X+X)=4X & 2X/4X=50%...). Since MSC and MAR are "qualified" events, the average strength of the participant is significantly different from most regionals.


The general trends are often reasonable (highest scorer will be highest), but trying to sort with any sort of fidelity requires additional work.

Look at some of the normalizing functions (dividing by number of matches, normalizing to event strength...). Normalizing function have their own problems as well.
In 2010, from an OPR perspective, 3 robots all at an OPR of 2 (about 2X national average) could actually combine to be worth 8-10 pts. versus the predicted 6 pts. in a match doing zone strategy.
In 2011, at really deep fields, OPR was actually driven down as 60+60+60 would often equal 120 due to the digresive scoring last year. Thus teams taht had really good partners at a strong event might actually see their OPR go down as they more frequently got into the stronger digresive scoring regions .