Galileo 2014

[JohnSchneider]

Originally Posted by Aileen

What are outliers?

People will much easier or much harder schedules relative to the majority of teams.

Welcome to the outliers club

[Aileen]

Originally Posted by animenerdjohn

People will much easier or much harder schedules relative to the majority of teams.

Welcome to the outliers club

Awesome. The exact reply I was hoping not to receive

[Ben Martin]

Originally Posted by Aileen

Awesome. The exact reply I was hoping not to receive

It's a fun club

Get your strategies ready, it's going to be a bumpy ride.

[Caleb Sykes]

I have created an alternate strength of schedule calculation. Essentially, it is the minimum OPR required of a team to go 8-2 in qualifications.

For me personally, I don't really care about rank unless the rank is <12. Since teams with records 8-2 or better will likely exclusively make up this group, I created a metric to see how hard it would be for a given team to achieve this record. Unsurprisngly, if this value is less than a team's OPR, then they are predicted to go 8-2 or better. If this value is greater than a team's OPR, then they can know approximately how hard they will have to work to go 8-2 or better.

team #	OPR required to go 8-2
1610	35.2
3138	48.1
5148	48.6
2980	52.8
3309	54.5
353	55.7
4945	58.7
494	60.5
4719	63.7
1310	67.1
4967	68
1318	70.1
45	70.3
3098	71.1
4965	71.2
3480	71.5
365	72.5
1108	73.4
4982	73.7
3504	77
3683	77.3
1266	78
4488	78
4991	78
714	80
1683	80.4
973	82.3
5098	82.8
3191	83.6
4060	85.1
3103	85.5
5012	85.6
176	85.8
228	88.5
862	89.6
1515	90.6
4176	91.5
869	93.8
1730	95.7
1218	95.8
2471	96
67	97.6
1775	98.3
79	98.6
2642	98.9
1717	99.7
5145	100.1
836	100.6
2122	100.7
4979	100.7
558	100.9
1334	101.2
188	101.7
2052	102
3008	102.4
126	104.9
3360	105.7
604	106.1
4940	108.1
2481	108.3
2665	108.5
217	108.9
2337	108.9
2424	109.2
884	109.5
1756	110.3
4985	110.4
488	110.8
2974	111.8
70	114.7
1023	118
4063	119
1011	119.1
5122	121.4
766	123
2177	123.1
4476	123.9
3316	126.5
2135	126.6
5137	126.9
2363	127.2
4917	127.2
955	128.1
857	130.5
4288	131
771	131.8
1153	132.2
3937	133.3
5320	133.3
193	133.4
225	135.5
179	137
303	138.3
148	141.4
3310	142
337	143.4
4256	144.8
4536	145.7
1885	150.1
384	154.6

average = 100.8
standard deviation = 26.5

[Laaba 80]

Originally Posted by inkling16

I have created an alternate strength of schedule calculation. Essentially, it is the minimum OPR required of a team to go 8-2 in qualifications.

For me personally, I don't really care about rank unless the rank is <12. Since teams with records 8-2 or better will likely exclusively make up this group, I created a metric to see how hard it would be for a given team to achieve this record. Unsurprisngly, if this value is less than a team's OPR, then they are predicted to go 8-2 or better. If this value is greater than a team's OPR, then they can know approximately how hard they will have to work to go 8-2 or better.

team #	OPR required to go 8-2
1610	35.2
3138	48.1
5148	48.6
2980	52.8
3309	54.5
353	55.7
4945	58.7
494	60.5
4719	63.7
1310	67.1
4967	68
1318	70.1
45	70.3
3098	71.1
4965	71.2
3480	71.5
365	72.5
1108	73.4
4982	73.7
3504	77
3683	77.3
1266	78
4488	78
4991	78
714	80
1683	80.4
973	82.3
5098	82.8
3191	83.6
4060	85.1
3103	85.5
5012	85.6
176	85.8
228	88.5
862	89.6
1515	90.6
4176	91.5
869	93.8
1730	95.7
1218	95.8
2471	96
67	97.6
1775	98.3
79	98.6
2642	98.9
1717	99.7
5145	100.1
836	100.6
2122	100.7
4979	100.7
558	100.9
1334	101.2
188	101.7
2052	102
3008	102.4
126	104.9
3360	105.7
604	106.1
4940	108.1
2481	108.3
2665	108.5
217	108.9
2337	108.9
2424	109.2
884	109.5
1756	110.3
4985	110.4
488	110.8
2974	111.8
70	114.7
1023	118
4063	119
1011	119.1
5122	121.4
766	123
2177	123.1
4476	123.9
3316	126.5
2135	126.6
5137	126.9
2363	127.2
4917	127.2
955	128.1
857	130.5
4288	131
771	131.8
1153	132.2
3937	133.3
5320	133.3
193	133.4
225	135.5
179	137
303	138.3
148	141.4
3310	142
337	143.4
4256	144.8
4536	145.7
1885	150.1
384	154.6

average = 100.8
standard deviation = 26.5

Wow. The difference from top to bottom is insane. Could you provide any more info about your process to get these numbers? When you are looking at team X, do you use max/world OPR for opponent Y's OPR, or is Y's OPR also based off what it takes to reach 8-2. This is a very interesting way to think about things, thanks for sharing it.

[Caleb Sykes]

Originally Posted by Laaba 80

Wow. The difference from top to bottom is insane. Could you provide any more info about your process to get these numbers?

I found it very interesting as well. All of the data that I used came from the Galileo predictions document found here. Essentially, for each match in a team's schedule, I subtracted out said team's contribution and calculated their alliance's winning margin without them (which were largely negative). I then ordered these winning margins in ascending order. My value is simply the negative of the third number in this ordered list.

When you are looking at team X, do you use max/world OPR for opponent Y's OPR, or is Y's OPR also based off what it takes to reach 8-2. This is a very interesting way to think about things, thanks for sharing it.

In all of my calculations, I use the OPR from the above document (I believe this is max event OPR).

I'm not sure if I'm being clear enough. I would be happy to answer any other questions you have about this.

[Boltman]

Originally Posted by inkling16

I found it very interesting as well. All of the data that I used came from the Galileo predictions document found here. Essentially, for each match in a team's schedule, I subtracted out said team's contribution and calculated their alliance's winning margin without them (which were largely negative). I then ordered these winning margins in ascending order. My value is simply the negative of the third number in this ordered list.



In all of my calculations, I use the OPR from the above document (I believe this is max event OPR).

I'm not sure if I'm being clear enough. I would be happy to answer any other questions you have about this.

Keep in mind the value of a goalie robot, the matches in where we were matched with two good shooters we (5137) won. There are lots of really good shooters at the finals. Might be a metric you want to calculate I think defense can be as good as offense by ruining cycles of the opposing alliance or at least delaying them significantly while our own alliance scores. We had a zero points in a match because all three robots were defense and two of our other losses was a broken robot in the mix...so that affected our overall stats at the regional but we made an impact and earned the rookie all star. Problem with defense only is we rely on two good offensive robots to provide scoring when that happens good things do. I think with all the offensive firepower at the championships...we will be fine.

Looking forward to St. Louis!

[JesseK]

All of these new lists based upon OPR make me .

I have scouted 1885's 20 alliance partners using the schedule currently posted and the video that is available (at least 6 matches per). My mentors & students will scout 1885's 30 opponents over the next 2 days. We start from the team's latest video (including elims) and work backwards in order to see the behaviors we will most likely see at Champs.

Data includes:
HG shots attempted, HG shots made, Inbound attempts, inbound successes, Auto ball attempts, auto ball successes, # of HOT goals, # of truss attempts, # of truss successes, # of trusses to HP or out of bounds, # of times a ball popped out of the robot during gameplay, rough estimate of intake time in seconds (also taking into account driver skills), rough estimate of favorite shooting distance, # of time effective defense was played on other teams, # of times the robot was pushed, # of times the robot was out of offensive position due to playing or receiving defense, # of obvious penalties, # of seconds of lost comms

Trivia: HOT is the 2nd most high goal scoring robot we play with on a per-match basis, but only by < 0.25 goals per match. Can anyone guess who the #1 high goal bot of the 20 is, statistically?

If anyone is willing to trade data on Wednesday, we will have plenty. This data isn't quite free since it took many labor hours to get. Yet if a team can even provide realistic statistics of their own team using the categories above and online video, we'd be more than willing to share all of what we have. The point of the trade is mutual hedging against schedule changes. Unfortunately the schools were out on Spring Break last week, so our ability to get all 100 teams' worth just wasn't there.

[Nate Laverdure]

Maybe next season I want to develop and test a new metric for FRC scouting: Bumper Quality Rating (BQR).

BQR-5: The platonic ideal of FRC bumpers. Fabric is snug and no wrinkles are visible; no extra material in the corners. Bumpers are installed or removed in one fluid motion. Milled slots or pockets in the plywood backing, sized for minor protrusions from the robot frame perimeter, allow the bumper frame to mate tightly against the robot.
...
BQR-0: Moderate to heavy use of duct tape.

I believe BQR would outperform OPR as a predictor of on-field robot performance.

[Calvin Hartley]

Originally Posted by Nate Laverdure

Maybe next season I want to develop and test a new metric for FRC scouting: Bumper Quality Rating (BQR).....

This. This is awesome.

[mrmummert]

I think "BQR" would have worked at Chesapeake....I can't tell you
how many robots we saw that had saggy,wrinkled and poorly made
bumpers. We saw a few that had material dragging on the floor that
made me wonder how they ever got thru inspection. I recall one team
there hadn't even made bumpers until they got there.

[brandon.cottrell]

Originally Posted by mrmummert

I think "BQR" would have worked at Chesapeake....I can't tell you
how many robots we saw that had saggy,wrinkled and poorly made
bumpers. We saw a few that had material dragging on the floor that
made me wonder how they ever got thru inspection. I recall one team
there hadn't even made bumpers until they got there.

Post-regional bumpers are the best kind of bumpers.

[Mr. Tatorscout]

Originally Posted by JesseK

All of these new lists based upon OPR make me .

I have scouted 1885's 20 alliance partners using the schedule currently posted and the video that is available (at least 6 matches per). My mentors & students will scout 1885's 30 opponents over the next 2 days. We start from the team's latest video (including elims) and work backwards in order to see the behaviors we will most likely see at Champs.
.

Well done! I hope all of your legwork pays off (Except in match 31, of course ). Good Luck!

[JesseK]

Originally Posted by Mr. Tatorscout

Well done! I hope all of your legwork pays off (Except in match 31, of course ). Good Luck!

I think Match 31 is too close to call since the margin is small relative to how fast a single cycle can happen. Right now 857 & 2052's stats and play styles are almost identical, so we'll have to figure out who plays what position. Both look like great partners to have. That strategy will likely decide the match flow, and therefore the match. With the right teamwork, we could totally blow that match wide open - or not, if we fail to work together.

[Christopher149]

Originally Posted by JesseK

I think Match 31 is too close to call since the margin is small relative to how fast a single cycle can happen. Right now 857 & 2052's stats and play styles are almost identical, so we'll have to figure out who plays what position. Both look like great partners to have. That strategy will likely decide the match flow, and therefore the match. With the right teamwork, we could totally blow that match wide open - or not, if we fail to work together.

Well, here's to hoping we work well together, eh?