Every cross country season, we have the same debate come up multiple times. Results come out from a meet that seem out of line with reality--or at least the rest of the state--and there are cries that it is short. The kids, parents, and sometimes coaches whose kids ran there want to insist that it's accurate because they want to preserve those precious personal bests. They jump up and down about how it was wheeled no less than 10,000 times and surveyed from space via GPS.
Argument #1: I wheeled it! Are you calling me a liar?
By and large, I don't believe any meet managers go out with the intention of making a short course. Maybe some do, but most legitimately are trying to do the right thing and believe that they have. The reality is that, first of all, your GPS readings don't matter. Secondly, wheeling a course is actually a science not an art. Subtle differences in technique can make a massive difference in the end result. The intent of this is not to hash out the details of that debate, but just to say there are a lot of factors that change the numbers substantially even if you think they are irrelevant, to name a few:
- On the rail? Three feet from the rail? Or true middle of the course?
- Shortest possible route (tangents) on corners versus taking them rounded and wide
- Straight lines versus a meandering path
- Weaving wheel (artificially added distance)
- Bouncing wheel (spins faster if it gets air)
- Slippage over grass and especially sand (more spins than actual distance)
I could go on and on, but the point is... even if someone says they wheeled it... HOW did they wheel it?
Argument #2: It was perfect conditions and a fast course!
Fantastic! Good competition, course surface and terrain, perfect weather, etc... these things obvious have an impact on the times. The fact is though that impact is FINITE and measurable.
Allow me to state the obvious: even a down hill, firm rubberized track, with 65 degree weather, and a wind at the back is not going to allow a 17:00 5K runner to suddenly run a 14:30. Therefore, there is a measurable upper performance limit and diminishing returns that even the best conditions can bring. So, yes, all of those things have an impact. However, that impact has a limit and we can estimate what difference "perfect" can make both from experience and from analyzing statistics... just like we can estimate what less than perfect will mean. Experienced coaches and stat junkies have some pretty accurate adjustment factors for various difficult courses. When an experienced coach says "this course runs about 45 seconds slow" that isn't hokus pokus... it is based on experience and statistical analysis. Our huge performance database and number crunching software makes this even easier to do.
In fact, if you look at any known venue on flrunners.com we have assigned values to them based on advanced performance comparisons and other algorithims.
Argument #3: This is cross country! It doesn't matter!
If time doesn't matter to you then why are you arguing about it? Do you really think time doesn't matter to these kids? Or to college coaches? Or to the thousands of stat addicted people who visit this site every year?
If it didn't matter then there would be no reason to argue. Or even keep time in the race. The kids and parents would not be fighting to keep their ficticious personal best. The fact is that time does matter and is an important way to track progress and have an idea where you stack up versus the competition.
Obviously head-to-head is the thing that decides championships or else we'd just mail the state meet results in based on time trials on a track. And, obviously, there are different courses and conditions. Those courses have different difficulty levels and YES slightly different distances, but as much as old timers and purists might not like to admit it, those differences are also measurable.
Let me use this place to talk about course distance variation... With the different measurement techniques (which now finally has a clearly defined recommendation from the national federation) there is some expected variation. You are unlikely to get the exact same number every single time, unless you use steel tape to measure. So some argue that this course or that course is short (including my own meet) and we leave it in as a 5K in our database. Why? Because there is an amount of acceptable variation or what I often say within the realm of reality. We compare it to other known courses and other times the same kids have run. There is a huge difference between when those stats tell you it looks a few seconds fast to when those stats tell you it looks a 40 seconds, a minute, or more fast. So that is why some are left alone... not because we pick and choose, but because of what stats tell us.
Argument #4: You are out to get my son or daugther!
Incorrect. I am a fan of the sport and of Florida kids. I want to see them succeed. I am not a coach and am not allied with anyone. I probably have never met you or your kid in my life. Believe me, I have nothing against you.
Argument #5: You are so mean! Why are you trying to hurt these kids?
So we start doing time comparisons and say well this kid ran this at this place and that at this place. And so we infer that the course is substantially short and therefore it should not be counted in the rankings or as a personal best. The next thing is always that we are being mean. Course doubters are accused of "calling the kids out" or insulting them. We hear "they worked so hard to earn these times and you want to take them away... you are hurting them by taking away their new PR... they should be in the rankings.. how dare you!"
Because I love bullet points, allow me to address this in a few ways...
- If we did not uphold standards and make such judgement calls then the rankings would be useless. Our rankings are used by newspapers, pollsters, coaches, parents, fans, athletes, college recuiters, etc. If we did not uphold the integrity then they would be worthless.
- YOU are hurting the kid if YOU want to give them a mythical personal best. You may think you're helping them, but you're not. The point of personal bests is to break them. Times are there to allow you to chart progress, make goals, and improve. How do you think your kid will feel if they are working hard all year long and can't figure out why they can't repeat that 16-flat from week two? They will think they're doing something wrong in their training and possibly kill their confidence and all around make their season not as fun. Improving and setting new PRs is fun! Trying to replicate a mythical personal best from a bogus course on a legit course is not. The athlete will likely be getting fitter and fitter during the season and be completely miffed at why their times have gotten slower.
- College coaches are smart... they've been in this game a long time... they will not be fooled by an outlier time.
- Comparing times from one race to another is not calling out any kids... it's illustrating facts. It's not making any statement about the kid... it's making a statement about differences in the two courses.
Fact #1: There is truth in the aggregate
You can take any antectdotal case or subset of data and make that statistic say virtually anything you want. So whenever anyone worth a darn talks about this kind of thing it should never be because Susie ran this at one course and that at another. It should be examining the data as a whole. There are various specific techniques, but my personal preference is to take the whole and remove the extreme cases (outliers) on either end. If you have enough data points and take into account all variables then the statistics do not lie.
Susie may have had a break through race, tapered for it, learned a new strategy, been well hydrated and carbo loaded, and legitimately PRed by 1:30... it happens. But it doesn't happen for 70% of the field all on the same day.
Fact #2: History is a great predictor of the future
There is also a sort of historical accuracy that if you have immersed yourself in stats for years you just know. You know, as a whole (not anecdotally), how much certain course conditions or difficulty is likely to efffect the times. You know what weather conditions may do to it. You know how track times are likely to impact the early season cross country times. Or how certain coaching strategies, training regimines, and schedules should impact those performances and when. You know where most people start the season and how much is realistic amount of improvement... for the entire season and week to week.
We also understand the exceptions. Like if you take an out of shape kid then they are going to see more rapid improvement more quickly... where as if you have a kid who is already fit then they are not likely to see huge gains over short periods of time. And we know that the average (by its nature) doesn't apply to every kid; there are extremes, outliers, and extenuating circumstances.
All kinds of things become second nature to stat junkies, but they are also measurable if you dig into the aggregate of the stats. And most of us have done these kind of studies... many many times because we're nerds and find it fun.
About the Mitchell Invitational
Something was not right about these times. Was it inadvertently measured short? Was it cut? Did people turn around too soon on this out-and-back course? I don't know... we can opine all we want, but we don't know. Was there a timing error that effected all four races about evenly? No... that didn't happen. What we do know by looking at the statistical facts is that the times are not accurate and should not be considered 5000 meter personal bests, regardless of how "fast" the course was since it was all pavement or how "perfect" the conditions were.
It is really unfortunate that some of the same kids were caught up in last week's course that we also deemed short: the West Pasco Invitational. That is a terrible coincidence and I feel badly about that. Last week I had to make the call on slightly shaky ground... I was pretty certain, but since it was a small meet there wasn't as much data as I'd like to be 100% sure. This week it was a big meet so data points are not an issue. Something was well off.
Most coaches and some runners--even those there whose PRs are in limbo here--know this to be the case. They have reached out to me. They knew it was off from the minute they saw the time. They, in some cases, have gone back and re-wheeled it and reported back to me on their results in private. Some of these coaches have already given their runners "adders" to add to their time to come up with a more accurate idea of what the time should have been.
I am sorry that this finding is not going to be popular with certain individuals and teams. Possibly they will lash out and call me horrible names. Be that as it may... stats don't lie and the integrity of our database is more important than making some enemies along the way.
Here is what I found....
I compared 37 runners from Wiregrass Ranch, Nature Coast Tech, Springstead, and Hernando that took part in both the Lecanto Invitational and the Mitchell Invitational. These are both guys and girls. I threw out any guys running over 20 minutes and any girls running over 24 minutes because their fitness level is more of a factor for week-to-week improvement.
Statistical analysis of the Lecanto High School course indicates that it is a +27 seconds slower than a "neutral" course at 20:00 pace. For comparison, McKeathan Lake is a course that is almost 100% pavement and known to be lightning fast. The same scale indicates McKeathan Lake is a -3 seconds faster than "neutral". [[ Purely for more context, Chain of Lakes Park (home of flrunners.com) is a +7 seconds ]] So... the Lecanto Invitational should run about 30 seconds slower than the Mitchell Invitational at 20:00 pace.
The average time for the Lecanto Invitational for these runners was 20:09. The average time for the Mitchell Invitational for these runners was 18:45... for a difference of 1:24. So adjusting for that 30 seconds makes the actual times run at Mitchell to be 54 seconds faster on average than one would expect.
The mean temperature on September 10th in Lecanto was 78 degrees. The mean temperature in New Port Richey on September 17th was 85 degrees. There was no rain either day. So weather wise it was arguably "more perfect" at the Lecanto Invitational. So the weather should not be considered a factor.
Fitness Improvement Week to Week
There can be expected week-to-week improvement in fitness. It is greater early in the season and less as the season goes on (already fit athletes improve less). We can measure this by comparing masses of athletes week-over-week in aggregate. I have done this in the past, but don't have the data in front of me. And while I did some analysis tonight, I didn't measure enough athletes to necessarily be conclusive and scientific. However, 15 seconds seems like a fairly rationale expectation for average week-over-week improvement this early in the season.
That brings our difference to 39 seconds. A 20:00 5K converts to roughly a 19:18 three mile time... or 42 seconds. Math seems to add up here that the course is leaning more toward the three mile side than the 5K side.
I don't pretend to know exactly why or how the times are off. I just know that they are. The statistics illustrate this very clearly.
There have been various re-measures since the meet with each person or group having their own specific motivations for doing so. And each person or group who has done so has different levels of experience on HOW to wheel a course correctly.
The simple fact that one group (the least experienced) can wheel the course just 15 meters short, another can wheel it as 3.08 miles (50 meters short), and yet another (arguably the most experience) can report to me privately that the re-wheeled it at 3.04 miles (108 meters short) tells you just how much difference technique can make. So, without doubting anyone's intentions or sincerity, you see what someone says they wheeled it at only has so much weight... stats don't lie and math doesn't vary based on how wide you take the corners or how much you weave.
Just Some Specific Stats
Lecanto to Mitchell
Athlete Team Lecanto Inv Mitchell Inv Faster Sam Hippely Wiregrass Ranch 16:46 16:11 0:35 Ermias Bireda Wiregrass Ranch 16:51 16:00 0:51 Tyler Mattera Wiregrass Ranch 17:16 16:39 0:37 David Hill Wiregrass Ranch 17:26 16:25 1:01 Ben Hall Wiregrass Ranch 17:56 16:50 1:06 Thor Alastre Wiregrass Ranch 18:15 16:41 1:34 Patrick Hill Wiregrass Ranch 18:16 17:13 1:03 Ryan Ludwig Wiregrass Ranch 18:36 17:25 1:11 Alec Lampasona Wiregrass Ranch 19:03 17:58 1:05 Andy Hippely Wiregrass Ranch 19:13 17:55 1:18 Nikita Shah Wiregrass Ranch 19:44 18:21 1:23 Elise Cedre Wiregrass Ranch 21:17 20:02 1:15 Chelsea Davis Wiregrass Ranch 22:03 20:10 1:53 Victoria Morea Wiregrass Ranch 22:28 20:52 1:36 Emily Payne Wiregrass Ranch 22:49 21:06 1:43 Berlin Watters Wiregrass Ranch 22:54 21:05 1:49 Savannah Goode Wiregrass Ranch 23:26 21:43 1:43 Tyler Moore Nature Coast 17:38 17:03 0:35 Cody VanNatter Nature Coast 17:39 16:16:00 1:23 Stephen Murphy Nature Coast 17:55 17:02:00 0:53 Kevin Ciccone Nature Coast 18:01 16:22 1:39 James Harkless Nature Coast 18:02 16:54 1:08 William Sandifer Nature Coast 18:41:00 17:48 0:53 Tatiana Lasanta Nature Coast 22:48 21:19 1:29 Alyssa Deligio Nature Coast 23:18 21:16 2:02 Brinna Horn Nature Coast 23:50:00 22:18 1:32 Parker Steinkamp Hernando 19:27 16:45 2:42 Bradley Geisler Hernando 20:46 19:22 1:24 Ty'Rhonza Harris Springstead 22:24 20:58 1:26 Jenna Barbee Springstead 24:32:00 21:53 2:39 Allysa Deligio Nature Coast Tech 23:18 21:16 2:02 Devon Rennie Springstead 23:16 21:04 2:12 Tatiana Lasanta Nature Coast Tech 22:48 21:19 1:29 Carl Zee Springstead 18:27 17:20 1:07 Zack Romblad Springstead 19:05 17:54 1:11 James Amodie Springstead 19:44 18:25 1:19 Jai Jackson Springstead 19:55 18:37 1:18 Average 20:09 18:45 1:24 If I throw in Red Mule 5K results... those times were almost exactly the same as Lecanto. Red Mule 5K to Mitchell
Red Mule was three weeks earlier so more improvement factor... but is also a lightning fast course.
Athlete Team Red Mule Mitchell Faster by