![]() |
Re: 2013 Championship Newton Division
I used the preliminary match schedules to simulate the matches using OPR and then calculate the standings. I used each team's best OPR from Ed Law's spreadsheet for calculating their contribution to a match. I also used each team's world Auto, Teleop, and Climb OPR from Ed Law's spreadsheet for calculating the rest of the standings.
Everything appears to match MechEng83's data. Code:
Rank Team Wins Losses Ties Matches QP AP TP CPCode:
Match Red 1 Red 2 Red 3 Blue 1 Blue 2 Blue 3 RScore BScoreCode:
Match Red 1 Red 2 Red 3 Blue 1 Blue 2 Blue 3 RScore BScoreCode:
Match Red 1 Red 2 Red 3 Blue 1 Blue 2 Blue 3 RScore BScore |
Re: 2013 Championship Newton Division
Quote:
|
Re: 2013 Championship Newton Division
Wait the match lists came out?
|
Re: 2013 Championship Newton Division
I'm surprised Shaun, you're usually more on top of this...
http://www.chiefdelphi.com/forums/sh...hreadid=116335 |
Re: 2013 Championship Newton Division
Quote:
It was released in today's first FRC Blog post: http://www.usfirst.org/roboticsprogr...atch-schedules |
Re: 2013 Championship Newton Division
Quote:
At least I'm predicted to lead the Division in Autonomous. :cool: If that prediction holds true then every victory we steal moves us to the head of the next win/loss category. Might make some noise here yet. ;) |
Re: 2013 Championship Newton Division
Quote:
|
Re: 2013 Championship Newton Division
Quote:
Are you sure you used the highest OPR for each team, and not just the first? I know that in the case of Team 116, your table is showing the lower OPR of our two events. -dave . |
Re: 2013 Championship Newton Division
Anyone feel like doing a Monte Carlo, should the schedule change? Simply swapping 4-6 teams per simulation and recalculating scores may be enough for more accurate predictions overall. I've been able to do one simulation manually, and the point spreads are close enough with the 2 schedules that I have that we could potentially go 6-2 instead of 3-5 (given our expected increased 24-pt teleop disc scoring ... well, that's my expectation...).
It would be interesting for each team to do this on an internal basis so they can see how their efforts leading up to champs may potentially change the outcomes. |
Re: 2013 Championship Newton Division
Quote:
While I can't speak specifically to Joe's calculations, they match up with mine. Even using MaxOPR of 42.6 (which I see for 116 was a vast improvement from your first regional) the predicted win-loss is 1-7. With so few matches and such a large field of teams, I think that luck in schedule will be more of a factor than performance in seeding. |
Re: 2013 Championship Newton Division
Quote:
Or, are you suggesting using some statistical variation on the OPR values of each team using the current schedule? If so, what variation coefficients would you use? I think this might be a very interesting simulation, but I'm not sure what the valid variation would be. I might need to dig further into the OPR calculations to find some sort of confidence interval. |
Re: 2013 Championship Newton Division
Quote:
So yes, while the average would approach OPR, we only play a single schedule which is likely to be helpful to some teams and disadvantage others. Your other interpretation is more interesting. With some small tweaks the OPR calculation can also be used to generate mean/variance parameters for a maximum likelihood estimate of each team's performance. That's a better basis for real Monte-Carlo simulation, but we'd still need to simulate over a large number of potential schedules. |
Re: 2013 Championship Newton Division
Quote:
Such simulations can also be used as motivation in the decision making process -- do we keep practicing what we're doing, or do we try a different approach for this match? How would that decision change if we have 3 hours of consecutive pit time prior to that large-spread match vs. if we only have 30 minutes? If the point spread is large, with little predicted chance for us to win, we may take more risks for those matches. Simulations may be more accurate for prediction than bean-count scouting early on in a schedule. Yet once we've hit everyone's 3rd or 4th match, I'd much rather have the disc-by-disc replay than a regression stat with few data points. Behavioral analysis also helps. |
Re: 2013 Championship Newton Division
Quote:
It's going to be exciting! Would simulating with a Monte-Carlo over a large number of schedules be nearly identical to relying on OPR for those same schedules? |
Re: 2013 Championship Newton Division
Quote:
For the values used in the calculating the Auto, Teleop, and Climb in the rankings, I used Ed Law's world ranking, because that was the only convenient place to get those values. Because of 116's performance in North Carolina, those values are significantly lower then the best OPR. 116's best OPR is comprised of 11.2 Auto 8.8 Climb 23.1 Teleop. Their World OPR is 12.3 Auto, 7 climb, and 3 Teleop. Since Teleop was where the major discrepancy is, and it is the last tiebreaker, there is little difference in 116's ranking whether best or world OPRs are used, although it could affect other teams. If someone had a convenient list of best OPR event broken out by Auto, Climb, and Teleop, I would have used that. |
| All times are GMT -5. The time now is 01:29. |
Powered by vBulletin® Version 3.6.4
Copyright ©2000 - 2017, Jelsoft Enterprises Ltd.
Copyright © Chief Delphi