Q? for strong racers
 

I have read, generally accepted and consistently for years, that in the upper reaches of cycling performance, and even merely strong racers, the differences get tight. Every minor edge is worth pursuing because one competitor is close to the next. I have generally accepted that, as practically self-evident. And, people who race confirm that so I tend not to question it.

But I've started to wonder if that might only be true for a handful of world-class elite, and maybe even the opposite really pertains. It there a table of performance vs percentile anywhere, that supports the consensus above? I look at the famous Coggan chart for example, and it looks like the difference in performance increases for each percentile tranche, not decreases.
https://cimg9.ibsrv.net/gimg/bikefor...7328ff1a02.png

Cyclinganalytics.com

The percentile tranches are from reported values by CyclingAnalytics site members. I arranged the "FT" column into a scatter plot:
https://cimg2.ibsrv.net/gimg/bikefor...bcb894d09b.jpg
And it looks to me like the high end shows a drastic increase in performance for a given percentile difference. If the difference between top competitors really get smaller, that line should be flatter not steeper.

I saw a similar situation from running statistics, which is what prompted this. The X-axis is reversed from the above chart: Percentage of "faster than runner" rather than percentage of "runner is faster than":

https://cimg9.ibsrv.net/gimg/bikefor...967d1c6c0c.jpg

My suspicion is that this reflects the general situation. Is there some data that sheds more light on this? Are there some special factors or confounding variables that invalidate it, or is our general assumption in fact wrong? What do you think?

I've read a few places that the Coggan chart above is gross misrepresentatively skewed high for the average peloton. It might be perfect for Europe or SOCAL where there's either riding all year round or just a higher level.

Also, I wonder if you're maybe thinking about it wrong. Percentage of riders with a higher than 6 W/kg is going to be pretty small, but they exist. So unless you're heavily rounding or only looking at 95th percentile, you're still going to have that final spike in the graph, even if it really only represents 0.25% of the sample.

I think the plot you listed in misleading. Try it as a bell curve. You'll see that after 5 W/kg or so, the line is essentially flat and only 30% of the sample is over the 4 W/kg mark. According to the chart, 4 W/kg is what's needed to be an entry Cat 2.
So ignoring my comment about the chart being skewed high, you can conclude that adding 0.25 W/kg to your FT jumps over less and less riders as you reach 100%.

Sidenote: I tend to think of W/kg as a pseudo logarithmic scale in terms of difficulty to achieve. 3 is relatively easy to achieve and maintain with casual training/riding. 4 much less so. 5 puts you in pro ranks, 6 puts your in WT Pro ranks.

There are a couple of issues with the w/kg scale, namely most amateur races in the US (Crits) aren't reliant on w/kg. The other factor being weight loss and most people just aren't willing to go as deep as they can. Last year went down to 7% and that was thin. Gaimon told me he raced at 4%. Those extra pounds imo account for the big jump at the end of the Cycling Analytics scale; most amateurs simply aren't either willing to go that deep or don't have the need to (if they races on the flats).

I also don't think in the high end of the sport margins are tight. Aerodynamics really help mitigate those gaps but even in the ProTour there are levels. In addition to that, very few of them are good across the Coggan Scale. For their role, they need to focus on 1 or 2 of those, and others suffer.

The w/kg is one measure not an end-all - selected here because there's at least some data. I'm not linking to the cyclinganalytics page because there were some unsupportable statements (IMO) there that I wouldn't want to get sidetracked on, and I do take the self-reported numbers with a grain of salt. But is there any reason why population tranches of other performance measures won't look similar?

If you're proposing that the top of the world class elite concentrate on their own specific physiological "edge", and those different advantages tend to balance each other out in competitions, I had that same thought. But it's not clear how closely those different "edges" would actually balance, and if so it also doesn't seem reasonable to apply it anywhere other than the tail end of the curve.

That's another way of expressing what I'm saying, and what the scatter plot shows. It takes MORE added power to jump x% of the riders as you get closer to 100%, for a given x. That's the same as saying the same power jumps fewer riders. The common knowledge suggests the reverse of that (that riders at the highest level get closer in ability, ie, takes LESS added power to jump x% of the riders).

Power =/= ability.

I think that's where we're differing. At the lower levels of the sport, simply adding power can drastically increase your success. At the upper levels, power is just part of the equation that adds up to ability.
To wildly make up numbers, if at say Cat 3, power is 60% of your success. At the WT Pro level, it's maybe 25%. Sure you have to have massive power to be a Pro. But every Pro has massive power. So it takes more than that to be successful.

I have the opposite opinion of that infamous chart. I've always felt that it under-estimated the power or w/kg needs of racers slower than the "pro" and "elite" ranks. Ignoring the whole "racing with a team" or "has good race smarts" part of the equation.

For example, for between a Cat 4/5.......the one hour power number on that chart at my weight would only be 207w and 5min power at 250w. That's it. That ain't much for a Cat 4/5 racer at all. Maybe if it is bone-flat and you weigh 80kg and can sprint.

I'll pose a controversial opinion:
Maybe it's either the winners of the genetic lottery or the drugs........or both when you see outliers.

Also, since it's w/kg you're getting a larger numerator and smaller denominator in the equation as you get more elite. Meaning 1w difference to a Cat 5 doesn't mean nearly as much in w/kg as to a grand tour rider.

For what it's worth that chart is not the final authority in performance, as it doesn't take into account how important fatigue resistance and repeatability (and other things) are for getting good results.

edit: like IIRC hasn't Gaimon's ftp technically gone up since he's been training for hill climb one and done smash fest efforts? But to do that he sacrificed his ability to repeat his efforts so he'd get kicked in the face if he tried to jump back into the WPT.

I think we're getting a little hung up on the W/kg and the chart. It's only one measure, and selected only because cycling analytics added percentile tranches. I think that it's very reasonable that we'd see the same sort of chart if we looked at absolute power.


What metric is more appropriate, percentage of podiums per race? VO2max? I found a "calculator" on this site Fitness Ranking (no guarantees how accurate) and entered in a bunch of numbers. The scatterplot looks similar.

https://cimg0.ibsrv.net/gimg/bikefor...2a0097d31f.jpg

At the upper reaches, it takes greater differences in V02max to climb the percentile, so those athletes are further apart not closer.

You're still missing the point.

When we're talking about the elite WT Pros, they're _ALL_ in the top 1% (maybe less, top 0.5%). All of them.
So while the jump from 96% to 97% may be large, you're still pretty far from inking a WT Pro contract.

Further, as much as I love CyclingAnalytics, I've been a subscriber for a very long time, I doubt many WT Pros use it. So what you're looking at is probably 99% amateur racers.

This graphic from 2014 for American Football/NFL is likely very true for cycling as well. We have much lower numbers, so percentages might be slightly better, but breaking into the WT peloton is probably just as hard.
https://imgur.com/zNOVaO6

I don't think that's a valid point, because 1. the range of differences appears to be still increasing in that region and 2. the point (from OP) is that the common concept is also applied even to "merely strong racers". We are not talking about only the best of the elite.

The question is one of objective data. If there is some data that shows a diminishing of differences even in that top 0.5% or even a smaller range, I'd love to see it and would change my supposition in a heartbeat. But what we have here so far against is speculative without data. Show me data for those who have inked a WT Pro contract if you posit that the chart will look different in that range.

All I know is that bike racing is (mostly) an art, not a science.

Rider smarter?

I think somewhere I've seen stats of the actual power riding in the middle of the peloton, and it is lower than one might think, despite the speed. Of course, the riders have to get out front every once in a while and take the wind at 25 to 30 MPH.

And, if say they're hitting 200W, and occasionally 300 or 400W, they're doing that for hours.

Question: Does that Coggins study date back to the era of EPO? If so, nearly all the elite bike racers had a huge edge in sustained power and recovery not available to us mortals. That would explain the curve going vertical.

Are you including track stars?

Some of them can put out impressive amounts of power, at least in short term efforts.

Steroids?

It's a valid point because I'm not just talking about the FTP of the Pros. There's going to be riders with drastically different FTPs in the Pros. But they have different abilities that make them less or more successful, otherwise the guys with higher FTPs would win every race by just riding faster than everyone else. So when you're _just_ looking at W/kg, sure the range might still be increasing, but it takes more than just power to be more successful than someone. You might close the power gap, but if they're more aero than you, or have better race craft, they'll still win. This is what people mean when they say "It so close at the top end of the sport", they're not just talking about power.

Objective data is probably not very easy to come by. Most pros are extremely guarded when it comes to power files. Even going as far as having it removed before posting to Strava. Hell, CyclingAnalytics even built in a 'remove power before upload to Strava' feature because people asked for it. When the power files are released, they're pretty spectacular.

Check out this one: https://www.velonews.com/2012/08/ana...allenge_235832 - Jens averaged 4.25 W/kg for the entirety of the 3:54 long race.

Yeh probably. But the percentiles graphed against W/kg are self-reported from Cycling Analytics site subscribers and are unrelated to the (in)famous graph, other than being overlaid on it. I probably should have just left the chart off and showed just my own chart with "data reported by Cycling Analytics" and avoided all the byplay about the Coggan chart, but I wanted to show at least somewhat attribute the source.

The CA data is from CA members, possibly some track racers considering the amount of power claimed.

The running data is aggregated from 20 years of available 5K race results. The chart looks very similar to the CA chart if you reverse the x-y axis.

The VO2 data, is unclear where they got it, or else calculated from some other data. I probably shouldn't list it without more investigation but it's strikingly similar to the other charts.

Can one get real stats for TDF racers?

Compare GC leaders/winners, Hill Sprint winners, Flat sprint winners, Hill Climb leaders (poka dot), etc.

Rider weight, bike weight, etc.

How many riders are logging power stats?

It is likely easier to get real data from a race than getting it from a lab. Is 100% effort in a lab the same as 100% effort during a race?

What about international track finals? TT cyclists?

You're assuming a level of analysis which wasn't present in the generation of the table. If you look back at the history of the chart, Coggan got data for for maximum and minimum performances for each of the durations pinning the top and bottom of each column. He then divided the range into 50 equally spaced steps. The purpose was for people to look at how their performance ranked across the durations to identify strengths and weaknesses. After it was first published there was demand to include performance levels along side the columns, so Andy (reluctantly) put the labels (untrained, fair, etc.) at equal intervals on the left side of the table. There was no analysis of actual Cat 4, 3, 2, etc. racers behind those labels. There was a lot of back and forth about whether those labels should even be there. They definitely should be taken with a grain of salt.

I should never have shown the table at all :( My chart has nothing to do with the definitions of the Coggan table or the Coggan data. I'm not assuming anything about it.

To be more clear, here is the chart of the Cycling Analytics respondents to some survey they had. I used the power numbers in the table, where crossed by the CA percentile tranche. The Coggan chart is only a background image for the data.

https://cimg8.ibsrv.net/gimg/bikefor...4432b06807.jpg
Compare to the running data, where I have altered the axes to conform with my power chart, for a more direct visual comparison
https://cimg1.ibsrv.net/gimg/bikefor...973b345a2c.jpg

Both charts show similar behavior. Overall performance in a footrace, and one of the measures of ability in cycling. They show an increasing distance at the higher percentiles, independently of Coggan's chart or data.

I still say that jump in the tail end is because the pro's optimize diet and run much leaner. Athletes in the 75% could probably get there as well but training hard and racing at 4-6% body fat isn't something most (even some very dedicated) to do.

In other words, Frank is naturally 3% better than me, we both optimize diet and get lean, and now he's 5% better because it amplifies the difference. It seems (to me) to be a reasonable hypothesis. That would mean (based on the 2 charts) that the effect begins around the 90-95 percentile and with increasing effect for higher percentiles so the theory needs to explain that somehow.

If true, it also supports my own conjecture that the difference between athletes gets larger, not smaller, at the elite end of the curve.

I follow Alex dowsett on strava and he posts his analytics after most races. Unless it's a REALLY hard climbing day, very few of his files look like something a strong cat3 couldn't pull off.

Hell, 90% of the people I see on zwift could do it :P

The "issue" I have with the charts is they are not about International WT pro races that should be measured after 5 or so hours, better yet - after 20 days and 5 hours.
I was a bit surprised when I saw junior results (several) and posted about them. The kids I know, mine inc, hit the WC levels in their teens, but on single day/short events.
I think the charts are rather useless / outdated and / or just have such limited input. USA 23 (and junior) amateur can top all the WT pro numbers listed above in the < 5 min range.