Saucony National Rankings vs. NY Speed Ratings

 

   
Though creating any type of cross country rankings is tough, and very much open to vigorously dissenting opinions, the task of ranking teams from around the nation only magnifies these problems. Most of the generally recognized US elite teams do not race against each other, and if they do, it often happens in early October when some squads have not reached very close to the level that will eventually make them top candidates for the national championship come December.
 
Clearly the complexities of cross country races preclude the fairly neat ratings that are available for track races, despite minor differences there.  Cross country courses vary hugely in terrain, surface, intricacy, trail width, and dozens of other factors. Even when they are supposedly the same approximate 5K or 3 mile or 2.5 mile distances, the "give-or-take" can add up to a lot of giving and taking. Race day conditions such as heat, precipitation, wind, sponginess and traction level of the surface, size of the starting field, etc. can also have big impacts on race dynamics.
 
Even for rating individual runners, raw times for personal-best performances do not necessarily provide an accurate sorting of the best runners based on a season's work. True, the top two runners for the girls as of Week 10 in the US Milesplit Saucony Rankings, Alexa Efraimson and Sarah Baxter, are #1 at the 5K and 3 mile distances respectively that they run the most, but you will not see the boys 1-2 of Cerake Geberkidane or Tanner Anderson in the top two of any of the 25 XC distances of MileSplit's stats lists, and some runners with top 5 times are not listed among the top 25 ranked individual runners. So the circumstances behind a time can matter big time, and rankers base their decisions off many factors. Plus, a time is just for one race, and if it was run back in mid-September, it may not have much relevance anymore. Keeping up with the status of runners in your own state is tough enough, but a check of the individual rankings shows that tracking runners from around the nation is a monumental task that sometimes leads to errors.
 
At the end of it all, rightly or wrongly, the results from Nike Cross Nationals sit as judgment on all of the seemingly profound national ranking decisions made by XC experts throughout the year. True, you may have a #1 ranked team composed of five Mary Deckers who can blast that dry hard 3-mile circuit and leave other teams in the dust, but then in Portland maybe your team loses its shoes, eats enough mud to fill a supertanker, and files in 22nd. The rankings may not have been wrong, but the NXN results sure makes it look that way.
 
Amidst all the crazy chaos of XC, there has been one rock of sanity that high school XC analysts can cling to if they want, something called speed ratings, or in New York as Tully Runners (TR) ratings. A quick description of the TR system and its application to an example course here in NY are provided in the two sections below, and then we wrap up with a review of how the speed ratings for top national teams in recent meets compares to their most recent rankings on the USMS Saucony lists. Be aware though that like most rating and ranking systems, the TR system is focused on how teams and runners will run under standard fairly normal conditions that do not include rivers of mud, freezing rain, and the occasional rugby scrum for boys' races. While somewhat rare, when these conditions do occur every December or so at a place out west, anything can happen, and usually does.
 
Now, if you already think you know more about TR speed ratings than you'll ever want to know and just want to check the national rankings vs. speed ratings charts, click here. If you do wish to read more about how speed ratings work and how nonstandard race conditions can affect them, click on the two headers below to expand out the text sections. On the other hand, if you want to access the big trove of articles on the subject on the Tully Runners site, click here.
 
   

 

Understanding TRs  [click to toggle]
Back at the start of the millennium in upstate New York, where all the towns have Roman or Greek classical names due to the interests of a late 1700s-era land surveyor, a team from the town of Tully began to excel in XC. When two members of the Lost Boys of Sudan, Lopez Lomong and Dominic Luca, were placed in Tully and started to put their running prowess to work, the team suddenly became very good. Lomong and Luka achieved the unprecedented feat of finishing 1-2 at their state championship race for two years in a row, and they led the team to its second state title in 2003.
 
A local engineer, statistics maven, horse-racing fan, and self-described crazy person (not unusual in XC, and crunching thousands of race results will do that to you) named Bill Meylan took an interest in the team beginning in the late 1990s, and by 2000 he was pumping out stats-based analyses of XC and track action on a site he called Tully Runners. From the beginning, the scope of the site only encompassed the local Section 3 action centered around Syracuse, while also touching upon the state action for those moving  through. By at least 2002, TullyRunners was covering the entire state in depth, and in later years it would inexorably expand to devour XC results from the Northeast and Mid-Atlantic regions.  Finally, the rankings were tackling the entire nation, based on the need created by a new national championship started in 2004 by Nike.
 
When Meylan first became obsessed with HS running stats, particularly those of XC, he tried to wrap his brain around the question of how you can compare the performance of different runners in different races. The left side of his brain reviewed the lists of results with all of the bunching and gaps in the list of times.  Meanwhile, the right side put the pieces together to come up with the most accurate and logically based system for rating XC runners and teams. Rather than relying on a system which assigned values to the difficulty of a race course, accurately measured distances, race day conditions, and all of the thousands of variables that affect a race, Meylan took the easy way out and decided to let the runners do all the work for him. Building a database of thousands of runners' results from a myriad of races, Meylan assembled a record of each runner's performances.  His speed ratings would then become based on the race, not the runners, where time differences of the participants and a median time value for a race became key. A runner's combined speed ratings, along with certain other adjustments provided a season's score. The XC slogan, "Pain is temporary, but speed ratings are forever" soon became a given in NY, but the workings of the TRs were very difficult to grasp for many coaches and runners.
 
Meylan inevitably receives queries from puzzled runners along the lines of, "Hey, I won the race up Mt. Doom today while battling the lava flows and nasty flying things and you gave me a 180 TR, while Jimmy Silverspoon beat a nothing field at Flatfast Meadows and he got a 182. No way!" Well, actually way, because the difficulty of a run only indirectly affects a speed rating; everything comes down to how much faster you ran than that median TR value for a race that all the runners have a hand (plus two feet) in deciding. If a hundred other moderately fast runners fought off the Nazguls as well as you, and finished on your heels, you're not going to get an elevated TR. And if you had a fast easy course but pushed yourself to the max and left your fast group of pursuers way behind, you're going to get a lot of love in the TRs.
 
For the standard 5K or 3 mile race, every 3 seconds difference in time equates to 1 TR point, with the lower the time the higher the TR score. The top boys in the nation run somewhere between 200 and 210 for top TRs, and the girls national leaders hit anywhere from 159 to 170. The TR average for the top 5 guys on the national elite teams is usually around the mid 180s, while for the girls it will be somewhere in the 130s. Shorter races can cause some issues with standard TR factoring, so the ratio used for 2.5 mile races such as the ones at Van Cortlandt Park in the Bronx is 1 point per 2.67 seconds.
 
As Meylan has repeatedly pointed out, speed ratings are an approximate estimation built on both science and art. What he seeks for (and occasionally race by race if conditions seriously deteriorate during the hours of a meet) is the best fit, the sweet spot, that most accurately reflects how all the runners at a meet were running based on their past history. The more runners, the merrier, and results will generally be most accurate for meets with large numbers of runners, rather than a small league championship involving 50 less-documented runners. TR scores early in the season will be less accurate, but by the end of the season the scores are pretty precise. Scores for New York and Northeast-Mid Atlantic runners are likely to be the most accurate, but in recent years Meylan has built up an impressive archive of results from around the country that are looked at to help determine the national rankings too. And for major postseason events such as the NY state championship and NXN, Meylan uses the season's TRs to run regression-based analyses to project scores for the upcoming action.
 
Of course the system is only as good as the analyst who runs it, and the TR system could potentially crank out badly skewed ratings if used by the wrong person. After grinding through thousands of races, Meylan is not only adept at finding the best TR fits under even the worst race circumstances, but he is also a reluctantly hardened pro at resisting entreaties of NY coaches who want the speed ratings of their team's last race fattened up a little. If anyone was ever suspicious that Meylan might have too generous a heart toward the local teams in Section 3, the Fayetteville-Manlius girls have shown with seven straight national titles that he may be even slighting them a bit when they win by 144 points at NXN. But going forward, as other XC analysts adopt speed ratings programs, the question of how the numbers are being cranked out could become more of an issue.  As such, it is not wise to compare speed ratings composed by separate people.  They are in turn, two separate measurements.
 
In XC, to paraphrase Tina Turner, nothing is ever nice and easy, and the same is true for the issues that can affect standard TR scores, which are dealt with in the next section. And it is important to remember their purpose.  They only reflect a performance, they do not predict a performance.  However, what has been done in the past, is a pretty good indicator of what will be done in the future.  Reflective, not predictive.  But the speed ratings are still by far the best tool available for evaluating runners, teams, and races.  As a long time devotee of TR crunching, I have been able to to chalk up a 90% success rate for projecting the winners of NY state qualifier races off of Meylan's obsession, even when faced by legions of those most horrifying of creatures, the toss-up-and-anybody-can-win-this type competitions. And as a betting odds enthusiast, Bill Meylan has surely already built a dozen or so chalets on Lake Tully from his winnings off the TR scores and related hunches about the premiere XC events.
 
   

 

Sinking Your Teeth in a TR Case Study  [click to toggle]
So why isn't life and XC easy, with a nice, smooth flow of everything that is needed to give us clear and consistent speed ratings numbers? Well, sometimes the weather changes and dumps rivers of precipitation on running trails, and sometimes we make XC courses modeled on Himalayan scenes. There are other times when we pack our races with 250 or more runners in places where the recommended occupancy level is more like 100. Many other factors can obstruct the route to a nice and happy TR score. Speed ratings that are, on average, generated from six to twelve races during the year for a runner under more normal conditions, can look very different from those that are posted for a late-season race held under less sane circumstances. At state qualifiers, the speed ratings for a few lucky teams that have no strong competition may well plummet as they decide to take leisurely jogs and save their energy for States.
 
Let's take a hypothetical situation at a place on the eastern side of NY called Sunken Meadow, which as its name implies has been the source of many dashed dreams and painful journeys on Long Island. The legendary course has a gauntlet of terrors that runners must navigate before they crawl over the finish line. Even the starting line is deceptively evil, as I have seen numerous runners pitch forward for an opening somersault as their feet spin out on a slick patch of filmed-over dirt. There are also such obstacles as a narrow little bridge with a sharp turn early on that bottlenecks packs of runners. Along with a short but very steep thing called Cardiac Hill , it is clear that Sunken Meadow is not the same type of course as Elma Meadows or Queensbury, two other places where States have been held during the last six years.
 
On a nice day, with a standard group of 70-100 or so runners, the speed ratings for a race at Sunken Meadow will look pretty normal for the participants, as they plow through the gauntlet without much impedance. But what happens when you take a bigger group of 126 runners in a state qualifier race, where the average time might be more than a minute faster than for an invitational race run there in early October? Throw in some unseasonably warm temperatures for the tighter up-front pack of runners and you are not likely to see the roadrunner conditions that some of the runners enjoyed at the Brown University Invitational a few weeks earlier.  There are also a number of potentially unknown psychological and strategic issues that could be factors in a highly competitive race for seven or eight top teams that are all running for a title, but who possibly do not want to go out too fast with the battle up Cardiac Hill looming.
 
The conditions in a big race at Sunken Meadow can easily add up to a "mud-run" situation like the one that happened at the Vernon-Verona-Sherril States in 2011. Though the race day was clear with pleasantly cool temperatures, the preceding day's heavy precipitation turned the soft spongy course into a long river of mud fairly early on during the races. All of the runners were slowed down, but the problem for speed rating is that not all of the runners were slowed at the same rate. Top runners at 2011 States such as Nick Ryan, Dan Lennon, and Brendan Smith for the guys and Jillian Fanning, Samantha Nadel, and Mary Cain for the girls ran far out in front for much of their races, ran the firmer-footing margins, and were not subjected to bumping from packs of bogged down runners slipping and sliding around them. A runner like Ryan may have been able to run 20-25 seconds faster without the mud, but the guy in 20th could likely have run 45-50 seconds faster under standard conditions.
 
So what's a speed rater to do when faced with the mud bath conditions? He can give the top runner like Ryan an incredible boost to something like a 210 TR level to keep some of the guys who floundered nearer their standard numbers, or Ryan can be kept close to his standard level (the choice taken) and many runners are going to look like they had bad days in the mud, which is reality. Coaches and runners may curse the speed ratings on the day of the packed-in state qualifier at Sunken Meadow or the mucky States meet, but speed ratings are a tiny blip in all of the very important things about the two races.
 
A final point about congested conditions in huge races that I will not delve into too deeply for fear of being tarred and feathered. From cursory analysis of a few races, constricted races appear to affect the girls teams more than the guys. Perhaps this is due to the guys' being better able to do the quick lateral movements required inside crowds, a product of the knee and hip and general muscle structure, but likely many other factors are at work here. As the saying goes, "Three's a crowd, and 255 you're dead (in the water)."
 
   

Stacking Speed Ratings Against the 2014 National Rankings 

XC fans can be forgiven if they get confused about how both state and national rankings work. In a fluid sport like XC, having a crushing loss in mid September means less than nothing once mid November rolls around. In college football, on the other hand, a big early-season loss will likely eliminate any shot at a national title. Viewing the top of the national XC girls rankings, Unionville fans could reasonably complain (and they do) that their team beat #1 ranked Fayetteville-Manlius in both of their meetings, so what gives? Well, an astounding seven-year run as national champs is a table-setter for FM, but the fact that the Hornets are always late-season bloomers and that their speed ratings have soared over the last month to near the top of the US charts are factors that push results from before mid October lower down on the scale of importance.
 
Quantifying all the ratings factors for a team is very difficult to do in a brief space. The two numbers cited in the charts for the ranked teams below, the TR speed rating average of a team's top five runners and the TR gap between the first and 5th runner, provide only a limited snapshot of a team's true strength level. The numbers do not include any info about the 6th-8th group of runners, who can provide critical depth. They do not show anything about a runner's health status, and whether they may be ready to explode upward or whether a nagging calf strain may have them on a downward spiral.
 
A TR gap of 20 can cover a team that has a 190-185-185-185-170 makeup or equally a 190-170-170-170-170 one. Generally, big gaps can indicate a point of weakness for a team and close-packed teams have an advantage, but the fields for a race can also have make-ups that cause problems for pack teams. Using speed ratings from the most recent race is usually a pretty safe bet at this time of year, but there still may be cases where teams just did not have to give a maximum effort to accomplish their goals at a state championship or NXN qualifier. Finally, some of the fields in these late-season races are TR-chopping huge, such as in that 255 field at NY Federations championship, so even the winner will likely have a drop in speed ratings.
 
Despite the limited data shown here, the basic TR speed rating averages have proven to be solid predictors for projecting winners. Tully Runners has nailed a healthy 7 of the last 10 NXN champs in the last five years' projections, despite the tendency of top-rated guys' teams to take severe dives out at Portland.  So for what they're worth, here's the score. Rankings are the MileSplit Saucony Week 10 results, and the TR scores are from the November postings (or late October in a few cases) by Tully Runners through November 22. For multiple listings, the first cited meet is the most recent. Note that not all races a team may have run in lately were speed rated, so the rankings can also have factored in performances at meets not cited here. Finally, new ratings for many teams from NJ, CA (but not Madera South), UT, CO, and TX were added to the chart from the period after the release of the Week 10 rankings.
 
TR-Rank Explanation: The numbers list out top teams by the numbers, but they do not represent what we think the rankings should be.  On the whole, we think the girls rankings are spot on, minus Great Oak and Fort Collins, who should be moved up based on this weekends performance, which isn't represented here because we used last week's rankings.  On the boys rankings, I think both Liverpool and Northport should be higher, with Henderson right behind, but I would be surprised if all three didn't move on to NXN to settle out the rankings.
   
Boys Team's Rankings vs. Recent Speed Ratings Averages  
Rank Team TR Speed Rating Average (and TR gap between 1st and 5th runner) TR-Rank
1 Christian Brothers Academy (NJ) 184.6 (15), 182.4 (6), 182.8 (14) 2
2 American Fork (UT) 179.2 (18), 182.0 (20) 12
3 Arcadia (CA) 182.2 (17), 176.0 (23), 177.4 (18) 3
4 Gig Harbor (WA) 181.8 (12), 187.0 (18) 5
5 Fayetteville-Manlius (NY) 185.2 (16), 184.0 (12) 1
6 Brea Olinda (CA) 179.0 (23), 178.2 (24)  
7 Brentwood (TN) 176.2 (26)  
8 North Central (WA) 181.6 (32), 185.4 (22) 6
9 Madera South (CA) 174.4 (7), 179.0 (9)  
10 Carmel (IN) 179.8 (11)  
11 West Chester Henderson (PA) 180.0 (22) 10
12 Liverpool (NY) 180.8 (14), 180.6 (9) 7
13 Broughton (NC) 179.4 (17)  
14 St. Xavier (OH) 180.6 (24), 180.8 (14) 8
15 Central Catholic (OR) 175.4 (15), 179.2 (14) 14
16 Pembroke (MA) 174.4 (22)  
17 North Allegheny (PA) 172.6 (7)  
18 Edina (MN) 175.2 (9)  
19 Hinsdale Central (IL) 174.4 (10), 177.8 (10)  
20 Cardinal O'Hara (PA) 174.0 (24)  
21 Wayzata (MN) 175.6 (22), 178.2 (19),  
22 Corona Del Sol (AZ) 178.6 (22)  
23 St. Xavier (KY) 176.4 (12), 180. 0 (24) 13
24 Northport (NY) 180.6 (23), 180.8 (18) 9
25 Nathan Hale (WA) 175.8 (8), 178 (24)  
Other
Top TR Teams
Warren (CA) 182.8 and175.0,  Don Bosco (NJ) 180.0, Dana Hills (CA) 179.4, Loyola (CA) 178.0, Southlake Carroll 177.8 and 173.4, Downer's Grove (IL) 177.6, O'Fallon (IL) 177.6, Bozeman (MT) 177.4, Columbus North (IN) 175.4 and 178.2, Bishop Hendricken (RI) 176.6 and 172.4, Jurupa Hills (CA) 176.4, Saugus (CA) 176.4, LaSalle (RI) 175.4 and 172.6, Ridgefield (CT) 175.4, and 176.0, Kamiakin (WA) DNS at NW regional and 178.2, Davis (UT) 175.0, Sioux Falls Lincoln (SD) 174.4, Mt. Tabor (NC) 174.6, Mt. Spokane (WA) 174.2, Saratoga Springs (NY) 170.8 and 175.2, West Windsor-Plainsboro South (NJ) 172.0 and 174.0  
   
TR-Rank Explanation: The numbers list out top teams by the numbers, but they do not represent what we think the rankings should be.  On the whole, we think the girls rankings are spot on, minus Great Oak and Fort Collins, who should be moved up based on this weekends performance, which isn't represented here because we used last week's rankings.  On the boys rankings, I think both Liverpool and Northport should be higher, with Henderson right behind, but I would be surprised if all three didn't move on to NXN to settle out the rankings.  
   
Girls Team's Rankings vs. Recent Speed Ratings Averages  
Rank Team TR Speed Rating Average (TR gap between 1st and 5th runner) TR-Rank
1 Fayetteville-Manlius (NY) 132.2 (17), 128.2 (24) 2
2 Unionville (PA) 130.2 (28) 3
3 Davis (UT) 127.0 (13), 126.2 (11) 6
4 Carmel (IN) 124.6 (28), 125.2 (27) 8
5 Wayzata (MN) 124.0 (7), 126.6 (5) 10
6 Simi Valley (CA) 127.4 (50), 123.2 (50), 128.6 (45) 5
7 Monarch (CO) 124.2 (10), 125.4 (20) 9
8 Great Oak (CA) 135.0 (18), 127.2 (4), 127.6 (26) 1
9 Blacksburg (VA) 114.0 (17)  
10 Glenbard West (IL) 116.0 (30), 128.6 (18)  
11 Bozeman (MT) 123.2 (13), 125.0 (14) 14
12 Naperville North (IL) 116.6 (22), 124.4 (17)  
13 Desert Vista (AZ) 123.4 (31) 13
14 Assumption (KY) 118.8 (16)  
15 Green Hope (NC) 122.6 (31)  
16 John Jay-Cross River 113.0 (15), 113.8 (13)  
17 Saratoga Springs (NY) 113.0 (13), 121.8 (10)  
18 Palatine (IL) 114.2 (19), 116.6 (30)  
19 Fort Collins (CO) 128.0 (24), 121.4 (35) 4
20 Pennsbury (PA) 126.6 (42) 7
21 Hamilton Southeastern (IN) 120.8 (30)  
22 Bellamarine Prep (WA) 116.4 (22), 117.8 (18)  
23 Park City (UT) 121.6 (42)  
24 Columbus North (IN) 112.4 (18), 120.0 (36)  
25 Cherry Creek (CO) 115.2 (15)  
Other
Top Teams
New Braunfels (TX) 124.6, Trabuca Hills (CA) 124.4, Birmingham Seaholm (MI) 123.2, Coe-Brown (NH) 122.8, Oakton (VA) 122.8, Dowling Catholic (WA) 121.6, South Barrington (RI) 121.6,  Camus (WA) 121.4 and 119.8, Redondo Union 121.2, Davis (CA) 120.2, LaSalle (RI) 119.8 and 120.6, Summit (OR) 119.8, Saugus (CA) 119.4 and 119.0, Randolph (NJ) 119.2, Suffern (NY) 119.0, East Ridge (MN) 115.0 and 119.8, Mt. Lebanon (PA) 119.2, Hudsonville (MI) 118.8, Capistrano Valley (CA) 118.4, Woodlands (TX) 118.0, Southlake Carroll (TX) 117.6, Lewisville Flower Mound (TX) 117.6, La Costa (CA) 117.6, Mira Costa (CA) 115.8, Rapid City (SD) 117.8, Tolland (CT) 117.0