normalizing times between courses
09/17/2016 12:25:07 PM
User
Joined: Nov 2009
Posts: 594
Normalizing everything to the Carrollton Orthopedic Invite, I get the following adjustments: Fulton: -30 Benson: +15 Saucony-ATL: +20 Spike/Spur: +39 Berry: +39 Gwinnett: +48 Cobb: +49 Starrs Mill: +52 ATL-classic: +56 Warpath: +64 Blastow: +64 North GA: +87
Normalizing everything to the Carrollton Orthopedic Invite, I get the following adjustments:

Fulton: -30
Benson: +15
Saucony-ATL: +20
Spike/Spur: +39
Berry: +39
Gwinnett: +48
Cobb: +49
Starrs Mill: +52
ATL-classic: +56
Warpath: +64
Blastow: +64
North GA: +87
09/20/2016 8:23:35 AM
User
Joined: Oct 2007
Posts: 47
Could you add Nash Farms?
Could you add Nash Farms?
09/20/2016 8:44:12 AM
Coach
SUBSCRIBER
Joined: Feb 2011
Posts: 163
[quote=LHSPanther]Could you add Nash Farms?[/quote] @LHSPanther I think that is "Sacuony-ATL".
LHSPanther wrote:
Could you add Nash Farms?


@LHSPanther

I think that is "Sacuony-ATL".
09/20/2016 10:33:01 AM
Power User
Joined: Aug 2006
Posts: 835
What about normalizing year over year to increase sample sizes?
What about normalizing year over year to increase sample sizes?
09/20/2016 11:24:30 AM
User
SUBSCRIBER
Joined: Apr 2006
Posts: 62
Would you mind adding Roper Park (Jasper) / Pickens Preview, Sharptop XC Invit.?
Would you mind adding Roper Park (Jasper) / Pickens Preview, Sharptop XC Invit.?
09/20/2016 12:35:51 PM
User
Joined: Nov 2009
Posts: 594
@MaristXCcoach Normalizing year-to-year becomes challenging because both athletes and courses change, perhaps by a lot. One key assumption in the "within year" normalization is that an athlete's ability is relatively constant from one meet to the next. Here are some of 2015 normalization factors: Fulton: -51 Carrollton: 0 Benson: 0 Berry: +22 Gwinnett: +35 Cobb: +40 Blastow: +45 Spike/Spur: +45 Warpath: +56
@MaristXCcoach

Normalizing year-to-year becomes challenging because both athletes and courses change, perhaps by a lot. One key assumption in the "within year" normalization is that an athlete's ability is relatively constant from one meet to the next.

Here are some of 2015 normalization factors:

Fulton: -51
Carrollton: 0
Benson: 0
Berry: +22
Gwinnett: +35
Cobb: +40
Blastow: +45
Spike/Spur: +45
Warpath: +56
09/20/2016 12:51:00 PM
User
Joined: Nov 2009
Posts: 594
@nibornnud I get Pickens at +22 seconds this year (+20 last year).
@nibornnud

I get Pickens at +22 seconds this year (+20 last year).
09/20/2016 10:24:06 PM
Power User
SUBSCRIBER
Joined: Aug 2013
Posts: 371
Interested to see how the Wingfoot Course will compare since last years times are irrelevant due to weather. Looks like good weather for this weekend too!
Interested to see how the Wingfoot Course will compare since last years times are irrelevant due to weather. Looks like good weather for this weekend too!
09/21/2016 1:43:16 PM
Power User
SUBSCRIBER
Joined: Dec 1969
Posts: 113
Can you normalize the Lamar Murphy Park please?
Can you normalize the Lamar Murphy Park please?
09/21/2016 2:01:42 PM
Coach
Joined: Jun 2005
Posts: 233
Lamar Murphy is North GA. He did that already.
Lamar Murphy is North GA. He did that already.
09/21/2016 2:43:39 PM
Power User
SUBSCRIBER
Joined: Dec 1969
Posts: 113
Sorry, didn't realize that. I appreciate the correction. I saw the name and didn't notice the +87. My mind went back to the Athens Christian Course when Coach Cummings would name that race the North Georgia Championships and we would attend.
Sorry, didn't realize that. I appreciate the correction. I saw the name and didn't notice the +87. My mind went back to the Athens Christian Course when Coach Cummings would name that race the North Georgia Championships and we would attend.
09/21/2016 8:41:36 PM
Coach
SUBSCRIBER
Joined: May 2006
Posts: 243
@dataholic How did you make the determination of the difference between NGA and Carrollton? I looked at all of the Varsity boys that ran N. GA and raced in the Championship or 6A/7A races at Carrollton and not one ran 87 seconds slower at Carrollton than at N. GA. There were 3 runners out of 22 that ran +84, but 2 of them are 20-21 minute runners. 14 of the runners were between +19 and +60, 3 more were between +66 and +68, the final 2 were between +74 and +78. An average taking out the +19 outlier yields +60. The median is +58. If you used data from the 1A/3A race, you need to take into account the significant temperature difference between the Championship race and the 1A/3A race. It was 15-17 degrees warmer in the 1A/3A race which would lead to much slower times in that race over the same course. I would rate NGA as no more than a +64 and more likely a +60.
@dataholic

How did you make the determination of the difference between NGA and Carrollton?
I looked at all of the Varsity boys that ran N. GA and raced in the Championship or 6A/7A races at Carrollton and not one ran 87 seconds slower at Carrollton than at N. GA. There were 3 runners out of 22 that ran +84, but 2 of them are 20-21 minute runners. 14 of the runners were between +19 and +60, 3 more were between +66 and +68, the final 2 were between +74 and +78. An average taking out the +19 outlier yields +60. The median is +58. If you used data from the 1A/3A race, you need to take into account the significant temperature difference between the Championship race and the 1A/3A race. It was 15-17 degrees warmer in the 1A/3A race which would lead to much slower times in that race over the same course.
I would rate NGA as no more than a +64 and more likely a +60.
09/22/2016 8:44:44 AM
User
Joined: Nov 2009
Posts: 594
@deadsnail I don't just look at those two races. I have a spreadsheet that currently includes 300 runners and 15 meets. Each runner has a median time (across all meets he has run). Each meet has a differential for each runner between the runner's time in that meet and that same runner's overall median time. It is then a trial and error process to minimize the median differential for each meet.
@deadsnail

I don't just look at those two races. I have a spreadsheet that currently includes 300 runners and 15 meets. Each runner has a median time (across all meets he has run). Each meet has a differential for each runner between the runner's time in that meet and that same runner's overall median time. It is then a trial and error process to minimize the median differential for each meet.

You must be logged in to comment.

Click Here to Log In.