Friday, July 27, 2007

Some numbers from the 2006 tour

There was some discussion in yesterday's roundup whether or not Landis' Stage 17 was out-of-line with his other stages, and thereby itself suggesting something untoward. A columnist said it was so, we said it wasn't and gave some numbers.
A comment here said we were calling "black" "white", and we have some more numbers now.

[MORE]


Yesterday, we challenged a Sydney Morning Herald columnist who wrote

It's funny how we viewers can now pick the drug cheats - just look at the ones that suddenly produce an unbelievable stage effort totally out of sync with their previous form: hello Floyd Landis.

We said then,

Funny, we seem to recall Landis in Yellow not two days before, having dragged the first group up the Alpe d'Huez. I don't remember Evans being at the head of any final climb in a mountain stage this year. And like Landis, Evans was second in the first time trial. Out of sync with previous form? More misinformation. Over at the CycleOps site, Allen Lim put up a table of Landis' climb data in 2006, which shows what he did on S17 consistent with earlier climbs, particularly Alpe d'Huez.


Make no mistake, S17 was a big day, but not the biggest KJ of the tour -- Stages 11 and 16 were more work. What is tactically notable is how the big climb was the first one. I also find S15 and S17 have normalized w/kgs of 5.47 and 5.69, which are solid, but not out-of-this-world.

Out of sync? The data says the columnist is making it up.

A comment said,

The stage 17 thing seems to be one of those TBV arguments that White is Black.

Amusingly the normalized W/kg was apparently computed but not given.
Maybe because it does make appear that stage 17 was out of the ordinary after all. Here is the actual data:
* S10: 5.15
* S11: 5.12
* S14: 4.90
* S15: 5.44
* S16: 4.75
* S17: 5.69
The two best stages are S15 and S17, by are margin. So it is a little more than the peloton was slow in S17, and Landis performed as usual. The averages obvious the dramatic let down in S16, maybe because S15 took a toll, or other reason.
But then, the next day, Landis was born again, since he had its best performance of the data set. Not only that but if you look at the full table, S17 is the only stage where he consistently performed above 5.5 (S15 comes close though).
And he did it with a 125km solo breakaway (not in the comfort of the peloton or small group), catching back the racers of the previous initial break, who then dropped like flies as time went.

So yes, his performance in stage 17 was extraordinary. The media emphazied it rightly so, if you are not convinced, you are deep in denial, please google the press articles of the time.

Well, I didn't want to bore people with spreadsheet data in the roundup, but the point the commenter missed was the total energy output of the stage vs. the effort placed into climbs. I've collected and collated the relevant data three ways, to show what points one can reasonably argue, as follows.

Here's the raw data:

sorted by



stage kj w/kg kj/w/kj
10 1848 5.15 358.79
11 3531 5.04 701.06
14 1840 5.32 345.94
15 2351 5.47 429.46
16 4638 4.75 976.54
17 2989 5.69 525.22

The expended kilo-joules on a route is about the same for all riders of the same weight. When sorted by total energy output, we see S17 was the third hardest day of the tour. Stages 11 and 16 were harder, and S16 was by far the most work. It's clear from all this data that S16 was the "killer" stage of the 2006 tour, which probably wasted the riders who rode it all out to the end. This had its effects the next day on S17.


Sorted by


stage kj w/kg kj/w/kg
15 2351 5.47 429.46
14 1840 5.32 345.94
10 1848 5.15 358.79
17 2989 5.69 525.22
11 3531 5.04 701.06
16 4638 4.75 976.54

Now, sorted by weighted average w/kg total, we have the numbers mentioned yesterday. These are the worst-case looking numbers for Landis, which is why I used them. They are the rate of energy output per stage by weight; this is where S17 appears to be the "out of line" value.



Sorted by

stage kj w/kg kj/w/kg
16 4638 4.75 976.54
11 3531 5.04 701.06
10 1848 5.15 358.79
14 1840 5.32 345.94
15 2351 5.47 429.46
17 2989 5.69 525.22

Note that 5.49 to 5.69 is a 3.96% difference. Whether that difference is an indication of doping is pretty debatable.

What is more interesting is when you factor out the intensity of the total stage, and look at the last column, total work / rate of work, which is kj / (w/kg):




Sorted by

stage kj wkj kj/w/kg % diff
14 1840 5.32 345.94
10 1848 5.15 358.79 3.72%
15 2351 5.47 429.46 19.70%
17 2989 5.69 525.22 22.30%
11 3531 5.04 701.06 33.48%
16 4638 4.75 976.54 39.29%


Considering the stage difficulty, Stage 17's rate of output was mid-pack. Stage 16 was by far the hardest, followed by Stage 11 -- consistent with the amount of work required for each stage. This also show how brutal S16 was.

The other argument not addressed is that 5.69 is not an earth-shattering number -- the border line of "extraterrestrial-land" seems to be over 6.0 and higher.

6 comments:

ShepFan said...

So yes, [Landis'] performance in stage 17 was extraordinary. The media emphazied it rightly so, if you are not convinced, you are deep in denial, please google the press articles of the time.

The media is the last place to look for proof of anything relating to cycling. Still, let's have a look at Google. OK, here's one, from August 28. 2006. by Dr. Allen Lim:

http://tinyurl.com/2rgllf

With respect to the data, Floyd averaged 281 watts for the entire 5 hour and 23 minute ride. In training before the Tour and even before the Tour of Georgia, Floyd would regularly perform 6-hour rides at 300-310 watt averages. As a point of reference, the overall average for the mountain days in the 2006 Tour de France was 269 watts +/- 16 watts, while the average in the 2005 Tour de France for the mountains was 274 watts +/- 20 watts. So Floyd's performance on Stage 17 this year was within the range that we would expect for the mountains in the Tour and below what he is capable of doing while training.

My emphasis.

If you don't like the data, Mr. Columnist, stay out of the conversation. Have a nice day.

Tholmies said...

Well, while I appreciate the effort, once again the argument gets less and less relevant.

First the total of the "kj" is indeed a good way to quantify the effort put on a stage, it is rather irrelevant to quantify whether a stage performance is exceptional or not. Stage 16 was the hardest of the tour, and this was known even before the race started. Stage 16 was an exceptional stage compared to... other stages, but that's not the point.
Floyd Landis' performance in that stage, was, however not exceptional at all, on the contrary: in fact he collapsed in the last climb, and lost most of his lost time there.

Second, the "kj/w/kg" is just bad physics. Since the w/kg is the same +/- 10%, it amounts basically to the same as "kj".

Ah, and yes 4% "w/kg" is a significant difference ; Pereira lost to Landis an amount of time equal to 2% of the duration of S17.

Third as for "ShepFan" reference, Dr. Allen Lim was the physiologist of Landis, is also the source of the data provided here in the table, and his intent was explicitly to disprove that Landis performance was exceptional as Landis was accused of doping one month before. The data shows actually nothing, as if the argument was correct, why didn't Floyd Landis simply win all mountain stages with solo breaks? And in addition it got mangled somewhere, here is the same, more complete data.
Look at it closely:
"attacking about a quarter of the way up the Col des Saisies [...] and left everyone in his dust after 30 minutes at an average power of 401 watts." And also: "Averaging 281 watts when moving for the whole ride". This is after 16 stages of the TdF.
Compared to the training ride: "Peak Power for 30 min: 391 watts" ; "Average Power: 247 Watts".
After 16 stages, he seems to outperform a carefully selected results of one training race. That's not even the point.

Recall that the originally disputed statement was: "the ones that suddenly produce an unbelievable stage effort totally out of sync with their previous form: hello Floyd Landis."

Recall that Floyd Landis collapsed at the end of stage 16. The next day, he does a solo break of 125 km, putting a 5.5 w/kg for every climb, something he didn't in the previous stages... Not only that but he manages to keep all his time lead with respect to Yellow Jersey, who was confortably chasing him with a 5-men group in the last two climbs, including the last, steepest climb of the tour. And then goes on to win his only stage victory in the TdF of his entire career. How does that go for exceptional?

ShepFan said...

Less relevant, indeed. Tossing around formulas, drawing conclusions from them without explanations... You lost me.

I think you've blurred the math and overlooked a key fact when you try to correlate the "4% w/kg" difference with the final time difference between Landis and Pereiro. Not only is the mathematical transform unsupported, the conclusion doesn't factor in the 30-minute gift Pereiro got from the peloton on Stage 13.

So, I read your "same, more complete data" citation. It is not the same data, it is from Floyd's training before the 2005 race. It's a year older than the data Dr. Lim examines in the article I cited. Even so, your citation concludes:

This of course proves absolutely nothing - except that Floyd did nothing super-human that day.

Your own source doesn't doesn't support the "exceptional" claim. If you're open-minded, read the Lim article--he explains in detail Floyd's relative performance, and how other factors were in his favor, like easy access to water and cooling.

Cub said...

What surprises me is that Floyd's average power output on the climbs of stage 17 was only 4% higher than on the on any other stage. I thought there would be more of a difference since he was riding hard from the start (or at least from halfway up the first climb), while on other stages he was just riding at the peloton's pace until the last climb or two.

I sure don't see any evidence of doping in his average power numbers.

What I do see is that even when in the peloton the riders are working really hard on the early climbs of a stage. No wonder the sprinters get dropped so fast.

I think Floyd's stage 16 average power numbers also show his bonk on the last climb where his power output was way lower than on any other final climb.

TBV, I think your last column - kj/(w/kg) - is really just a measure of the time spent climbing on each stage. kj/(w/kg) boils down to mass*time, and Floyd's mass was constant, so time is the only variable there. But it's been a long time since I took physics classes, I might have made a mistake.

Tholmies said...

As for Pereiro, I just meant 4% performance is a big deal. Actually, in any sport, almost universally, in most cases, the winner cannot afford to lose 4% performance and still win.

As for the data, it was posted in July 2006, and the data was posted/repeated by Lim from June 2006. So I naturally assumed that Lim used the same data in August 2006, and gave the average power while pedaling as 300-310. Now, if you want to argue that from 2005 to 2006, at age 31, Floyd Landis went from 247 to 300-310, that opens an entire can of new questions. Also, why Lim, who was not shy of publish the data of 2005, and of most stages of the TdF, didn't give the detailed information for 2006 in his defense of his customer, is a mystery to me.

"read the Lim article--he explains in detail Floyd's relative performance, and how other factors were in his favor, like easy access to water and cooling."

Well Lim's defense of his partner, is fishy, because he first argues that the stage performance was not exceptional, then gives excuses why the stage performance of Floyd should be exceptional: i.e., Landis gained much of the time when descending not climbing, then the water excuse.
However the water excuse is a lie. Precisely, this part is a lie: "Because of the direct and immediate feedback from the power meter, Floyd came to an immediate and extraordinarily important realization during his ride -- that every time he poured ice cold water on his body, his power output went up. We had discussed staying hydrated and cool that morning."
Apparently, if you read that article, for the first time in the entire history of cycling and after 100 years of TdF, it was suddenly discovered by a extraordinary racer and by a young and brilliant physiologist that plain cold water would bring a decisive success, more than performance-enhancing drugs; a discovery who dramatically changed the face of cycling. Not. Give me a break.

It is a lie, because it was very well known before; in sept. 2005, Lim said: "For example, we've got a good idea of how much sweat is required to keep Floyd cool on a normal day of racing (1 liter per hour for every 100 watt average)". It is impossible, suddenly he (and Floyd) discovered at S17, that water was a so wonderful performance-enhancer, having scientifically studied its effects. Actually, Lim had discussed importance of water logistics for S4 already.
Lying about this fact been discovered in stage 17 is very convenient; then you do not have to explain why the hell Floyd Landis didn't do the same in the previous stages. This is what I wrote: "as if the argument was correct, why didn't Floyd Landis simply win all mountain stages with solo breaks? "
Also it is unclear to me why Pereiro, Sastre and other high profile chasers (not just Dave Zabriskie and Christian VandeVelde as in the article), who ended to be in groups of very small numbers, would not get access to water easily, and had to get deshydrated.

"Even so, your citation concludes:

This of course proves absolutely nothing - except that Floyd did nothing super-human that day.
"

But the data of the source do not back that conclusion: going from 247 W practice to 280 W after several days of race is a feat. Now if it is really 300-310 W, then ok - then an explanation for that sudden dramatic jump is in order.

The bit "This of course proves absolutely nothing" is correct: comparing a practice race to the 17th TdF stage is rather meaningless. After all, this amounts to arguing that winning the last TdF mountain stage (the day after the most grueling stage) with 5-10 chasing teams on your heels, is identical to a practice race. IMHO, this is dubious. That why I wrote: "That's not even the point." The point is that we were comparing Floyd Landis' TdF stages to other Floyd Landis' TdF stages, and the point was made very clear:

Recall that Floyd Landis collapsed at the end of stage 16. The next day, he does a solo break of 125 km, putting a 5.5 w/kg for every climb, something he didn't in the previous stages... Not only that but he manages to keep all his time lead with respect to Yellow Jersey, who was confortably chasing him with a 5-men group in the last two climbs, including the last, steepest climb of the tour. And then goes on to win his only stage victory in the TdF of his entire career. How does that go for exceptional?

jrdbutcher said...

Tholmies, you are out of your depth.