[TullyRunners Home Page]

Speed Rating Note 2018

Bill Meylan (Oct 8, 2018)


Katelyn Tuohy had a speed rating of 175.7 at the Ocean State Invitational in Rhode Island on September 22, 2018 and 177.0 at Holmdel Park, NJ on October 6, 2018.  These are the 3rd and 4th highest speed ratings I have derived for high school girls ... the top two are Amber Trotter (180.3 in 2001, Foot Locker best in Orlando, FL) and Melody Fairchild (177.9, Foot Locker Balboa Park course record).  Whenever speed ratings get some national attention like this, there are several coaches and viewers in California  that get really upset because they hate speed ratings.  I get some "interesting" e-mails and various message boards get messages highly critical of speed ratings and anybody that pays attention to them.

High speed ratings at Foot Locker, and especially Amber Trotter's 180, are the focus of criticism ... They don't believe this value and other relatively high values for both boys and girls at Foot Locker.  For those really interested ... in this article below, I'm going to show the process of deriving Amber Trotter's 180 speed rating.

And Now ... People seeing the awesome performances of Katelyn Tuohy can't believe any high school girl was ever capable of having a speed rating higher than Katelyn Tuohy ... So something is wrong with speed ratings! ... One group in New Jersey loves to criticize me and they consider Holmdel Park as "holy" ground, and nothing could be better than Tuohy's performance at Holmdel.

And this from the Letsrun message board ... "it's just his personal opinion (Meylan), but when you get a website with a large following like Milesplit using them as if they are fact and start considering them when picking wildcards for NXN then it's a real problem." ..... Well Mr. California coach, Bill Meylan has some good news for you ... A number of years back, Rich Gonzalez offered me a spot on the NXN Selection Committee, but I turned down the offer ... The NXR regional qualifiers were coming into play and I have always thought that head-to-head competition was the best selection process and that regional qualifiers give everybody the opportunity to toe-the-line and qualify.  NXN is Nike's party and I believe Nike should make up the guest list.

Some Speed Rating Basics

My XC Speed Ratings are a statistical out-growth of Speed Ratings & Figures from thoroughbred horse racing. Speed Ratings in horse racing were developed by two guys at Harvard University (Andrew Beyer and a mathematics doctoral student) ... Once developed, Andrew Beyer quit Harvard to become a full-time horseplayer, book author and columnist for the Washington Post ... the Beyer Speed Figures are included in the past performances of the Daily Racing Form and are commonly noted on horse racing TV shows (Triple Crown races and others).  Since high school XC races have many more "runners" than horse races, Bill Meylan simply extended the concept by applying additional statistical methods.

What are Speed Figures Used For ?

Speed Figures in horse racing relate the speed of horses to other horses with numbers that indicate how close they finish to each other on a relative basis ... they tell you how fast a horse has been running in its past races with a single number.

Speed Figures are a handicapping factor used for betting purposes ... The race time of the horse is irrelevant ... the numbers lengths he gets beaten by or beats other horses on a relative basis is the important factor ... Knowing the relative speed as lengths between horses helps you to bet real money.

Speed Ratings - Cross Country

Relative Time Differences between runners is the important factor in cross country (corresponds to lengths in horse racing)... I developed XC speed ratings with that in mind ... The single number tells you the relative speed compared to all other runners ... and comparing two numbers tells you the relative time difference (for example, comparing a 150 and 145 speed rating indicates at 15 second time difference since 1 point equals 3 seconds).

The entire process is designed so that a 150 speed rating anywhere equals a 150 speed rating anywhere else ... A speed rating is NOT a power rating that tells you how good a runner is ... a speed rating indicates how fast a runner ran relative to other runners.

To derive speed ratings, final times and the time differences between runners are correlated with the known speed ratings of the individual runners and known speed of specific races AND groups of runners (of known quality and speed) ... graphical and statistical comparison are made to determine the speed of the particular race being evaluated ... subjective estimations are commonly necessary in the early season, and speed ratings typically improve in accuracy as the season progresses ... Statistically, it is partly an iteration process. Good historical data is applied and helps when needed.  Statistical iteration involves guessing, but experience helps to make better "guesses" to arrive at an acceptable solution.

Reproducibility ... Anybody using my speed rating process with my baselines and libraries of past races will derive speed rating values very similar or equal to my own ... This assumes a user has a knowledge of application and some experience (absolutely required) ... Speed Ratings are not precise numbers, they are best fit approximations ... Humans and horses are imperfect and variable data points which means variable margins of error ... Some races (most with good historical data) are fairly easy to speed rate and reproducibility amongst different users will be excellent (equal or within a point or two) ... Some races are "uncertain" due to a lack of data,  so good "guesses" come into play (experience is real helpful here).

I treat XC speed ratings as if I were using them to bet upcoming races (for real) ... I want them to reflect what happens in a race relative to all other races, otherwise the utility is greatly diminished.  I tried my best to bet real money on Kelsey Chmiel at the Great Edinburgh Scotland last January because her speed rating was close enough to Claudia Lane and I knew Claudia had been sick ... in addition, Claudia was running her first XC race ever outside of California, was crossing multiple time zones to race, would be running in real mud for first time and would probably be wearing long spikes for first time ... Looked like a great betting opportunity and the United Kingdom allows betting on these types sporting events ... Unfortunately, there was betting on the Senior races but not the Junior races.

Girl's Foot Locker Race 2001

Background Information ... I encourage anybody interested in the Foot Locker 2001 races to watch the Fox Sports Net replay (start-to-finish) ... it is available on YouTube at: https://www.youtube.com/watch?v=VMrOOeMhPTA ... dominating performance by Amber Trotter and a great competitive boy's race.

It was warm and very humid at the Orlando Disney World golf course in 2001 (about 80 degrees at race-time for the girls) ... you can see dew on the grass ... the TV commentator noted the course is rather slow with many rolling hills and the turf is is soft and "spongy" with the humidity making it a strength and endurance course ... The girls ran before the boys and one newspaper article noted the girls' may have made the course faster for the guys by "squeezing" the dew moisture out of the grass making the spongy areas drier (but who knows ... the final race times tell the story for Meylan).

Amber Trotter's coach (Jerry Drew) was interviewed before the race and said Amber was healthy and "everybody knows about the anorexia problem" ... Trotter was forced to miss the spring track season because of her physical condition caused by the eating disorder ... other articles noted Trotter was improving her physical condition throughout the cross country season and was hoping for an all-out effort at Foot Locker.

Speed Ratings ... the speed ratings were determined by two methods ... (1) the known speed ratings of the individual girls were correlated with the final times and time differences between runners to "fit" (derive) a race adjustment to calculate the final speed ratings ... this is my "reference runner" method ... How fast any individual runner runs is determined by how how fast everybody else ran ... the actual final times are irrelevant (this really annoys some track coaches) ... the time differences between runners is fitted with their prior speed ratings and projected speed ratings ... as normal in statistics, outlier data needs to be excluded ... (2) the race is evaluated as a "group" based on comparison to previous years results at Foot Locker (individuals are ignored) and the race as-a-whole is compared graphically and statistically to the historical data ...Method 1 is the preferred method, and is the determining method for Trotter's 180 speed rating ... But Method 2 gets the same result when done correctly.

For starters, here are the Girls Foot Locker 2001 results with speed ratings:

Girls Footlocker 2001 Results (with speed ratings)

1   Amber Trotter  (12)       West        Redwood Valley      CA   16:24.07   180.3    180 
2   Erika Odlaug (12)         Midwest     Bannockburn         IL   17:04.30   166.9    167
3   Natasha Roetter (12)      Northeast   Lexington           MA   17:05.14   166.6    167
4   Molly Huddle (12)         Northeast   Elmira              NY   17:19.59   161.8    162
5   Laura Stanley (12)        South       Fletcher            NC   17:34.54   156.8    157
6   Felicia  Guliford  (12)   West        Gallup              NM   17:35.67   156.4    156
7   Jackie  Zeigle  (12)      West        South Jordan        UT   17:41.84   154.4    154
8   Valerie Lauver (12)       South       Allen               TX   17:45.40   153.2    153
9   Amber Harper (11)         West        Orem                UT   17:46.10   153.0    153
10  Heidi Lane (12)           West        Great Falls         MT   17:46.93   152.7    153
11  Julie Allen (12)          West        Corona Del Mar      CA   17:50.51   151.5    151
12  Sally Meyerhoff (12)      West        Tempe               AZ   17:51.40   151.2    151
13  Tracey Brauksieck (12)    Northeast   Cuyler              NY   17:51.87   151.0    151
14  Megan Kaltenbach (11)     Midwest     Aurora              CO   17:52.68   150.8    151
15  Kara Scanlin (12)         South       Ponte Vedra Beach   FL   17:53.19   150.6    151
16  Kristina Roth (12)        Midwest     Celina              OH   17:53.83   150.4    150
17  Mindy Sullivan (12)       South       Lubbock             TX   17:56.74   149.4    149
18  Angela Homan (12)         Midwest     Delphos             OH   17:57.27   149.2    149
19  Liz Gesel (12)            Northeast   Manchester          NH   17:58.19   148.9    149
20  Zoe Nelson (9)            West        Kalispell           MT   18:00.88   148.0    148
21  Jessica Gall (12)         Midwest     West Lafayette      IN   18:01.25   147.9    148
22  Laurel Burdick (12)       Northeast   Manlius             NY   18:03.34   147.2    147
23  Stephanie Madia (12)      Northeast   Wexford             PA   18:06.15   146.3    146
24  Jackie Mulrooney (12)     Midwest     Hartland            WI   18:08.67   145.4    145
25  Nicole Bohnsack (10)      Midwest     Rockford            MI   18:14.01   143.7    144
26  Jessica Cickay (10)       Northeast   Newtown             PA   18:14.64   143.5    143
27  Lindsay Van Alstine (11)  Northeast   Midland Park        NJ   18:16.79   142.7    143
28  Julia Cathcart (12)       South       Starkville          MS   18:23.88   140.4    140
29  Amber Smith (10)          Midwest     Ishpeming           MI   18:34.19   136.9    137
30  Carly Matthews  (12)      South       Winston-Salem       NC   18:39.23   135.3    135
31  Jessica Hellender (12)    South       Cooper City         FL   18:51.63   131.1    131
32  Brooke Stewart (11)       South       College Station     TX   19:09.99   125.0    125

Going into the race, I had comprehensive speed rating data for the three New York girls ... and tentative speed rating data for Foot Locker regionals and a few State meets outside NY ... I re-evaluated the race at a later time after collecting more data (and better data outside NY) and adjusted the speed ratings.

Some Individual Data

Molly Huddle ... the top NY girl who finished 4th ... Molly was not one of the three favorites in the race despite being really good ... I had her projected to be 4th and that's where she finished ... Here is Molly's speed rating record:

  Footlocker Nationals    12/8/2001    4    17:19.59   161.8    162
  Footlocker NE-Champ     11/24/2001   2    17:59.3    158.2    158
  Federation Meet         11/17/2001   1    18:02.5    156.2    156
  NY States - Class D     11/10/2001   1    18:24.5    149.8    150
  Section 4 - Class D     11/1/2001    1    17:35.0    160.3    160
  IAC Championships       10/27/2001   1    18:16.0    156.3    156
  Conwicke-Stock Inv      10/6/2001    1    16:56.0    157.0    157
  McQuaid Inv Var AA      9/29/2001    1    17:04.3    156.6    157
  Corning Invitational    9/15/2001    1    18:25.2    153.6    154	

Going into Foot Locker Nationals, Molly was solid upper 150s to 160 speed rating ... now she would be running with a group of girls faster than her, so projecting a speed rating a few points higher is within the realm of possibility ... Her speed rating of 161.8 at Foot Locker is consistent with what she had done and what she might do.

Laurel Burdick (Fayetteville-Manlius) and Tracey Brauksieck (Homer) ... these are the other two New York girls ... both schools are within 20 miles of Tully, so I was very familiar with them ... Both girls had solid upper 140s to low 150s speed ratings going into Foot Locker Nationals ... Laurel finished 22nd with 147.2 rating and Tracey finished 13th with a 151.0 rating ... Both totally consistent with what they had been doing (and slightly lower than Foot Locker Northeast) ... Here are the ratings for Foot Locker Northeast:

1 NATASHA ROETTER      12  LEXINGTON      MA     17:48.4    161.9    162
2 MOLLY HUDDLE         12  ELMIRA         NY     17:59.3    158.2    158
3 TRACEY BRAUKSIECK    12  CUYLER         NY     18:00.4    157.9    158
4 LAUREL BURDICK       12  MANLIUS        NY     18:13.6    153.5    153
5 LIZ GESEL            12  MANCHESTER     NH     18:19.4    151.5    152
6 JESSICA CICKAY       10  NEWTOWN        PA     18:20.5    151.2    151
7 LINDSAY VAN ALSTINE  11  MIDLAND PARK   NJ     18:24.5    149.8    150
8 STEPHANIE MADIA      12  WEXFORD        PA     18:26.7    149.1    149

The other Northeast girls (places 5-8 at regionals) had slightly lower ratings at Nationals (but it turned out those ratings were more consistent with their seasonal ratings leading to Foot Locker Northeast which is not uncommon) ... It is common for some runners to have their best speed rating at qualifiers because that's how you make it to Nationals ... getting there is Job #1.

Natasha Roetter ... she beat Molly Huddle handily at Foot Locker Northeast ... the times at Foot Locker Northeast were relatively slow because it was cold and rainy (but time differences are still time differences) ... Natasha finished 4th at Foot Locker Nationals in 2000 (155 speed rating and the top returner) ... she was undefeated in 2001 and had been running significantly faster times at meets than the previous year ... many observers thought she was the true favorite ... she was already a 162 speed rating, and now she would be running competitively with other girls as fast as her ... certainly possible her rating might rise ... BUT let's see how fast she runs compared to other runners (e.g. time differences) ... The time gap between her and all other Northeast got bigger, including Huddle (which indicates a possible higher speed rating for Natasha ... and that's exactly what happened).

Ericka Odlaug ... ran a very smart race at Foot Locker ... let Huddle and Roetter pace her, and out-kicked Roetter right at the end ... Broke 12 course records in 13 races (she did not break the Footlocker Midwest course record; it was completely rain-soaked and muddy) ... The Illinois girls were racing a 2.5-mile distance in 2001 ... Erika Odlaug had shattered that Detweiller Park record earlier in the season by over half-a-minute ... She won the IL State Meet by 45 seconds ... Yes, Meylan had to "guess" (I dislike 2.5-mile championship races), but she was at least in the upper 150s to low 160s range ... The previous outdoor track season, Erika Odlaug won the Golden West 2-Mile (10:11.31), was 2nd at the adidas Outdoor Nationals 2-Mile and then won the Junior National 3000 Meters over bunch of college runners ... So she was pretty fast ... and she was helped by getting pulled to fast time at Foot Locker ... her 166.9 speed rating was not a surprise to me ... BUT probably unbelievable to some California coaches because Meylan's Foot Locker ratings are all inflated.

Other Runners (except Amber Trotter) ... Making a long story shorter ... lots of consistency with their Foot Locker ratings versus other race ratings!

Amber Trotter ... followers of speed ratings can recognize this - The race adjustment and speed ratings of the race has already been determined, and I haven't even considered Amber Trotter's performance ... Amber Trotter's speed rating has been determined by the other runners in the race ... How fast they ran relative to each other determined the race adjustment ..... The exact same principal determined Katelyn Tuohy's ratings at Holmdel and Ocean State.

Amber Trotter was a work-in-progress during the 2001 XC season ... Her best speed rating entering Foot Locker Nationals was roughly 164-167 achieved at Mt SAC ... very comparable to Sarah Baxter and Claudia Lane ... It was obvious the Mt SAC course ran slow for Amber Trotter compared to current years ... Based on her 2001 ratings entering Foot Locker, I would project her at mid to upper 160s (she had a 153.0 rating at Foot Locker 2000) ... But Something else happened (Wow did something else happen).

She broke the course record by 31 seconds, but speed ratings don't care about that ... speed ratings care about the time differences between runners of known ability ... Amber Trotter beat Molly Huddle by 55 seconds (or 18.3 speed rating points) and she beat Ericka Odlaug and Natasha Roetter by 40 and 41 seconds ... and all the other runners of known speed ratings ... And collectively, the other runners in the race determined Amber Trotter's speed rating of 180.3.

For those believing Amber Trotter's speed rating is greatly inflated ... If I dropped her rating at Foot Locker by 10-15 points, I would need to drop everybody else's by 10-15 points and that obviously make no sense at all ... I think that would be disrespecting a lot of talented females.


Some Stats of Interest

Here are the official race splits ( http://footlockercc.com/2001/ ):

Amber Trotter's Splits:      Boy's Splits (Tim Moore):
  800m        2:30             ---
  1 Mile      5:04            1 Mile      4:40
  2 Mile     10:24            2 Mile      9:32
  2.5 Mile   13:05            2.5 Mile   11:59
  3 Mile     15:45             ---
  Finish     16:24            Finish     14:50

Trotter  had a 23 second lead at 1-Mile and 33 second lead at 2-Miles ... she averaged 5:15 per mile for 3 miles ... Her 13:05 time at 2.5 miles is 16.8 seconds faster than Katelyn Tuohy's 2.5 mile course record at Van Cortlandt Park.

The Boy's 2.5 Mile split is 4 seconds slower than Edward Cheserek's course record at Van Cortlandt Park ... and the Boy's final time is 3 seconds faster than Edward Cheserek's course record at Holmdel Park of 14:53 (Cheserek's speed rating was 208 and Tim Moore's rating at Foot Locker was 207) ... New Jersey guys thought Cheserek's rating was a bit low ... and the California contingent thought his 210 at Foot Locker was inflated.

Katelyn Tuohy ran 16:21 at Holmdel ... Amber Trotter ran 16:24 at Foot Locker ... as noted in the New Jersey Shore Coaches Holmdel results, Holmdel ran about 9-10 seconds fast compared to Cheserek's record race, so adding 9 seconds to Tuohy's time yields 16:30 to make it comparable to Cheserek. 


One Graph to Annoy Those Who Hate My Graphs

The speed rating course adjustments for the Girls for Foot Locker 2001, Foot Locker 2017 and NXN 2017 were not that far apart ... Just for fun, I merged the results from Foot Locker 2017 and NXN 2017 ... I merged them because NXN did not exist in 2001, and in 2017, the talent was divided between NXN and Foot Locker ... Levels the playing field for comparing the 2001 race to the 2017 races (in terms of quality) ... I used the top 26 from the merge for the comparison ... The graph is simply the raw data plotted with some labels:

The graph displays a very nice consistent match in quality with runners in finishing positions 6 through 23 ... Another nice matching comparison involves Molly Huddle and Kelsey Chmiel ... Molly Huddle's speed ratings are posted above ... New Yorkers will recognize that Huddle's and Chmiel's ratings look like clones of each other ... and in the graph above, their data points are very close to each other (Chmiel was 159 at NXN and Huddle was 161.8 at FL, and Chmiel did a 161.7 this season); just another indication of relative comparison.

Trotter's 180 is nicely displayed in relation to Tuohy's 172.