Bill Meylan (TullyRunners) -- Updated Speed Ratings to track times

I wish he’d gone with his original methodology of a simple linear regression for 1600m and 3200m times. The numbers are fine for anyone running in the 180s-200s, but the scaling is totally off at a higher or lower level than that. The 1600 times change at roughly 0.87 seconds per point. His original numbers changed at 0.7625 per point, and that resulted in a lot better scaling. The constant in the equation was the only thing that needed scaling, probably because the depth in track times has gotten a lot faster in the last 15 years. The top guys are pretty much the same but now we’re seeing sub-9:00 runners ranked outside the top 50, nationally, whereas they would’ve been top 10 and likely top 5 in 2007.

I did the same thing with the previous methodology recently and my numbers were almost exactly the same for people in the 180-200 range but my numbers lined up much better everywhere else, especially at the top level (this chart would call the 1500m world record a 227 while the 5k is a 230, I just don’t think there’s a discrepancy of 2.4s in a 1500 or 9s in a 5k between the two.

So None of the Newbury Park runners can break 14:00 or 9:42 (Sahlman's PR from last year)?

Doesn't even pass the sniff test.

corrected:

So None of the Newbury Park runners can break 14:00 or 8:42 (Sahlman's PR from last year)?

Doesn't even pass the sniff test.

These are fun to use. Kessler's 1500m, 3:34, is worth about 216 pts. Nico Young's 7:56.97 3000m is worth about 208 pts. Webb's 3:53 mile is between those two.

If I read that correctly, the 5000m time listed is a xc time. So, it has Sahlman's 14:00 in xc equivalent to the 8:42 3200m he ran in track. However, I'd assume that he was quite a bit better in xc than track, maturing, another summer at altitude training, great team comp. I'd predict 13:40s on the track this year and 8:35.

You read wrong. You all did actually. Sahlman ran a 203 speed rating at RunningLane. According to Meylan, this means he can run 13:57 or 8:40 on the track, if he chose to run one of those events instead of that RunningLane race. It's also important to note two things:

1. Meylan specifically says these are not exact numbers. In his original analysis, he found the standard deviation for 3200m times to be 10 seconds. It's highly likely a runner could run 203 on the XC course then run 8:30 for 3200m, or the opposite and run 8:50. It would only be a statistical anomaly if he ran in the 8:20s or above 8:50 (and even then, statistical anomalies do exist, they don't invalidate the other data).

2. Nico Young, trained under the same system, ran a 205 on the XC course and then a 3k race worth a 208 rating. Granted, he ran that 205 alone while the 3k was in a competitive race (the NP boys this year always had great competition within their own group). However, this suggests Sahlman can run a 206 equivalent on the track this year, which would be a 13:48 or 8:35, if he can maintain the same improvement (exactly what you said).

Personally, I expect a small improvement from the NP runners in track, and I think their whole top 3 will go under 14:00 and 8:40. I don't think any of them will break a national record this year, though (except maybe outdoor 3k, it's a bit weaker). Next year, however, Lex and Leo may take good shots at any of the records, including possible Galen's 13:37 (rated at 210) or Lukas' 8:29 (rated at 211).

my two pennies wrote:

You read wrong. You all did actually. Sahlman ran a 203 speed rating at RunningLane. According to Meylan, this means he can run 13:57 or 8:40 on the track, if he chose to run one of those events instead of that RunningLane race. ...

That's correct ... The updated chart I posted correlated XC speed ratings with 5000 meter track times ... and for the most part, the data points were XC speed ratings and 5000 meter track times from the the same season (or close to it).

I then used the IAAF scoring tables + Runnerspace Performance Predictor to show approximate track equivalence of 5000 meters to the 3200, 3000, 1600 and 1500 meters.

The original chart worked OK for some States, but poorly for others because it was based solely on NY runner track times ... and I later found out I was missing many track PRs (especially 3200m) because they were never posted or available.

A concern of regressing actual track times with actual XC speed ratings:

(1) Which track times do you use ... winter or spring ... going into XC season or post XC season in the following spring? ... Minimal testing shows a wide variation of results depending on mixing of averages or PRs ... Using selective results do work better, but that will take me a lot more time.

With respect to the Colin Sahlman, his 203 speed rating at RunningLane correlates to a 13:57.0 5K on the track ... As I have noted before, top elite runners may not achieve their best XC speed ratings at very fast venues such as RunningLane or Woodbridge, so it would not surprise me if Colin Sahlman was currently capable of a 206 or 207 speed rating on more challenging course which could separate him more from other runners ... and that would put him at 13:42 to 13:45 on my updated chart ... and a number of additional months of training could make it faster.

RunningLane example ... Jenna Hutchins ran a 166.53 speed rating at RunningLane, but entered the race with a 171 speed rating earlier in the season which correlates to a 15:36 on the girls' chart which is close to what she ran on the track shortly after ..... This year, Jenna Mulhern finished 4th at RunningLane running a 16:29.51 after running a 5K track time of 16:25.22 in October.

Galen Rupp had XC speed ratings of 207 (Foot Locker), 201 and 197 his senior year, but he achieved a 209+ equivalence with his HS 5K track record ... Some runners are just better (or sharper) during XC versus track.