TdF analysis - true effects of doping?

[ScienceIsCool]

Hi everyone.

I've done a bit of analysis that I thought I'd share. Recently I became curious, wondering why the Tour de France speeds slowly crept up through the 80's and have levelled off in recent times. The glib answer is doping, but if so then how did the Tour change as a result?

I began by taking to representative years; 1980 (pre oxygen vector doping) and 2000 (middle of oxygen vector doping period).

I went to "memoire du cyclisme" and looked up each stage's distance, amount of climbing, and winner's time. Firstly, I discarded prologues and all other time trials.

I then went to work calculating the speeds for all stages with multiple categorized climbs (HC, 1, and 2). The total amount of climbing and stage distance were considered as confounding variables. The shocking result is that between 1980 and 2000 the average speed on a climbing stage went from ~31.5 km/hr to ~33 km/hr with variations as expected due to length and amount of climbing.

I found that a bit shocking! What this indicated is that the change in eras resulted in less than 5% difference! Within statistical variation, this would be unlikely to completely disrupt the sport. And certainly doesn't account for the overall average change in TdF speed.

That was when I went to work analyzing the flat stages. Each Tour (1980 and 2000) had 10 stages characterized as flat. The average length of flat stage was 206.6 km and the average speed was 36.8 km/hr. The speed was a fairly good linear fit to stage distance as you'd expect.

In 2000, the average length of flat stage was 204.5 km (i.e., virtually identical), but the average speed rose to 42.9 km/hr! That's a 6.1 km/hr (16.5%) difference in speed. Remarkable. When equated to average power output, assume that wind resistance dominates and that it rises by speed^3. This means that the average Tour rider in 2000 was putting out nearly 50% more power during a flat stage than a similar rider in 1980!

I believe that this is the true effect caused by oxygen vector doping, and why the average rider would need to dope in order to survive.

Regards,
John Swanson

 

[Cobblestones]

Hi everyone.

I've done a bit of analysis that I thought I'd share. Recently I became curious, wondering why the Tour de France speeds slowly crept up through the 80's and have levelled off in recent times. The glib answer is doping, but if so then how did the Tour change as a result?

I began by taking to representative years; 1980 (pre oxygen vector doping) and 2000 (middle of oxygen vector doping period).

I went to "memoire du cyclisme" and looked up each stage's distance, amount of climbing, and winner's time. Firstly, I discarded prologues and all other time trials.

I then went to work calculating the speeds for all stages with multiple categorized climbs (HC, 1, and 2). The total amount of climbing and stage distance were considered as confounding variables. The shocking result is that between 1980 and 2000 the average speed on a climbing stage went from ~31.5 km/hr to ~33 km/hr with variations as expected due to length and amount of climbing.

I found that a bit shocking! What this indicated is that the change in eras resulted in less than 5% difference! Within statistical variation, this would be unlikely to completely disrupt the sport. And certainly doesn't account for the overall average change in TdF speed.

That was when I went to work analyzing the flat stages. Each Tour (1980 and 2000) had 10 stages characterized as flat. The average length of flat stage was 206.6 km and the average speed was 36.8 km/hr. The speed was a fairly good linear fit to stage distance as you'd expect.

In 2000, the average length of flat stage was 204.5 km (i.e., virtually identical), but the average speed rose to 42.9 km/hr! That's a 6.1 km/hr (16.5%) difference in speed. Remarkable. When equated to average power output, assume that wind resistance dominates and that it rises by speed^3. This means that the average Tour rider in 2000 was putting out nearly 50% more power during a flat stage than a similar rider in 1980!

I believe that this is the true effect caused by oxygen vector doping, and why the average rider would need to dope in order to survive.

Regards,
John Swanson

Why not simply analyze the flat ITTs? Shouldn't they show about the same improvements?

 

[goober]

Hi everyone.

I've done a bit of analysis that I thought I'd share. Recently I became curious, wondering why the Tour de France speeds slowly crept up through the 80's and have levelled off in recent times. The glib answer is doping, but if so then how did the Tour change as a result?

I began by taking to representative years; 1980 (pre oxygen vector doping) and 2000 (middle of oxygen vector doping period).

I went to "memoire du cyclisme" and looked up each stage's distance, amount of climbing, and winner's time. Firstly, I discarded prologues and all other time trials.

I then went to work calculating the speeds for all stages with multiple categorized climbs (HC, 1, and 2). The total amount of climbing and stage distance were considered as confounding variables. The shocking result is that between 1980 and 2000 the average speed on a climbing stage went from ~31.5 km/hr to ~33 km/hr with variations as expected due to length and amount of climbing.

I found that a bit shocking! What this indicated is that the change in eras resulted in less than 5% difference! Within statistical variation, this would be unlikely to completely disrupt the sport. And certainly doesn't account for the overall average change in TdF speed.

That was when I went to work analyzing the flat stages. Each Tour (1980 and 2000) had 10 stages characterized as flat. The average length of flat stage was 206.6 km and the average speed was 36.8 km/hr. The speed was a fairly good linear fit to stage distance as you'd expect.

In 2000, the average length of flat stage was 204.5 km (i.e., virtually identical), but the average speed rose to 42.9 km/hr! That's a 6.1 km/hr (16.5%) difference in speed. Remarkable. When equated to average power output, assume that wind resistance dominates and that it rises by speed^3. This means that the average Tour rider in 2000 was putting out nearly 50% more power during a flat stage than a similar rider in 1980!

I believe that this is the true effect caused by oxygen vector doping, and why the average rider would need to dope in order to survive.

Regards,
John Swanson

And... you wasted your time (not sure why you would even start this and then try to say science is cool when you obviously are ignorant to the variables that a true science weenie would have looked at). Races are rarely run at full potential speed. Tactics far beyond speed determine the overall outcome even though the final time, derived by speed, is the result. Sorry...

 

[ScienceIsCool]

ITT's would be massively affected by the confounding effects of aero technology. Aero technology improved at about the same time that oxygen vector doping was introduced en masse. It would be very difficult to separate the two mechanisms from any changes in ITT speed.

Aero technology should not affect flat road stage speeds, however. Considering that aero wheels only save a watt or two at 40 km/hr and rider positions haven't changed radically, this would not account for a 50% change in overall rider power output.

John Swanson

 

[goober]

Why not simply analyze the flat ITTs? Shouldn't they show about the same improvements?

A much better idea - still with lots of variables; but, much much much better.

 

[goober]

ITT's would be massively affected by the confounding effects of aero technology. Aero technology improved at about the same time that oxygen vector doping was introduced en masse. It would be very difficult to separate the two mechanisms from any changes in ITT speed.

Aero technology should not affect flat road stage speeds, however. Considering that aero wheels only save a watt or two at 40 km/hr and rider positions haven't changed radically, this would not account for a 50% change in overall rider power output.

John Swanson

ITT would be much more stable an analysis then yours. Really think about what you are saying - have you raced bikes before? That might be a good place to start to better understand the dynamics of a race which are only partially driven by speed and time - again even though the result is speed and time.

 

[ScienceIsCool]

And... you wasted your time (not sure why you would even start this and then try to say science is cool when you obviously are ignorant to the variables that a true science weenie would have looked at). Races are rarely run at full potential speed. Tactics far beyond speed determine the overall outcome even though the final time, derived by speed, is the result. Sorry...

Okay, given that this is true, why would the speeds have gone up then? Is it because tactics have changed radically? Are there external causes (increased TV coverage?, race radios?) that are plausible?

The fact is that on a flat stage, the average power increased by 50%. I find that interesting. Note that the same effect was not seen in climbing stages.

My overall point is that whatever changes between 1980 and 2000, the flat stages were affected much more than climbing stages. Highly non-intuitive.

John Swanson

 

[D-Queued]

ITT's would be massively affected by the confounding effects of aero technology. Aero technology improved at about the same time that oxygen vector doping was introduced en masse. It would be very difficult to separate the two mechanisms from any changes in ITT speed.

Aero technology should not affect flat road stage speeds, however. Considering that aero wheels only save a watt or two at 40 km/hr and rider positions haven't changed radically, this would not account for a 50% change in overall rider power output.

John Swanson

Nike Swift Suit technology is at least as effective, likely moreso, than aero wheels in road stages.

...doping isn't completely the be all and end all of speed, but there are more factors involved.

...then again, some riders (hello Eki) didn't even know how to hold themselves in a decent aero position on a typical road stage... true story.

Arguably, they didn't need to worry about it when operating on jet fuel.

Dave.

 

[ScienceIsCool]

ITT would be much more stable an analysis then yours. Really think about what you are saying - have you raced bikes before? That might be a good place to start to better understand the dynamics of a race which are only partially driven by speed and time - again even though the result is speed and time.

Please explain how you would analyze ITT's. I'm curious how you'd create a model that could exclude the confounding affects of aero technology.

Please explain how/why the race dynamics have changed to introduce a 50% change in power output. I've raced off and on since the late 80's. I don't remember a huge shift in tactics that would account for this.

John Swanson

 

[Polish]

Hi everyone.

I've done a bit of analysis that I thought I'd share. Recently I became curious, wondering why the Tour de France speeds slowly crept up through the 80's and have levelled off in recent times. The glib answer is doping, but if so then how did the Tour change as a result?

I began by taking to representative years; 1980 (pre oxygen vector doping) and 2000 (middle of oxygen vector doping period).

I went to "memoire du cyclisme" and looked up each stage's distance, amount of climbing, and winner's time. Firstly, I discarded prologues and all other time trials.

I then went to work calculating the speeds for all stages with multiple categorized climbs (HC, 1, and 2). The total amount of climbing and stage distance were considered as confounding variables. The shocking result is that between 1980 and 2000 the average speed on a climbing stage went from ~31.5 km/hr to ~33 km/hr with variations as expected due to length and amount of climbing.

I found that a bit shocking! What this indicated is that the change in eras resulted in less than 5% difference! Within statistical variation, this would be unlikely to completely disrupt the sport. And certainly doesn't account for the overall average change in TdF speed.

That was when I went to work analyzing the flat stages. Each Tour (1980 and 2000) had 10 stages characterized as flat. The average length of flat stage was 206.6 km and the average speed was 36.8 km/hr. The speed was a fairly good linear fit to stage distance as you'd expect.

In 2000, the average length of flat stage was 204.5 km (i.e., virtually identical), but the average speed rose to 42.9 km/hr! That's a 6.1 km/hr (16.5%) difference in speed. Remarkable. When equated to average power output, assume that wind resistance dominates and that it rises by speed^3. This means that the average Tour rider in 2000 was putting out nearly 50% more power during a flat stage than a similar rider in 1980!

I believe that this is the true effect caused by oxygen vector doping, and why the average rider would need to dope in order to survive.

Regards,
John Swanson

John, why only look at 1980 and 2000?

How about 2010 and 2011?

How do you explain Tommy V and Tommy D both riding faster than Big Mig and most of the other dopers of the 90's?

The fact that they are riding clean tells me there are some more important attributes besides doping that are leading to increasing speeds.

shhhhhhh

 

[ScienceIsCool]

I chose only 1980 and 2000 for the following reasons:

- my time is limited, or I would have analyzed 30-40 years worth of data
- between these two time periods, road conditions and stage lengths remained stable
- they represent clearly differentiated periods, pre and post introduction of widely available oxygen vector doping products

John Swanson

 

[Hugh Januss]

John, why only look at 1980 and 2000?

How about 2010 and 2011?

How do you explain Tommy V and Tommy D both riding faster than Big Mig and most of the other dopers of the 90's?

The fact that they are riding clean tells me there are some more important attributes besides doping that are leading to increasing speeds.

shhhhhhh

The whole point of your argument becomes null and void when you start with a faulty premise.

 

[MacRoadie]

John, why only look at 1980 and 2000?

How about 2010 and 2011?

How do you explain Tommy V and Tommy D both riding faster than Big Mig and most of the other dopers of the 90's?

The fact that they are riding clean tells me there are some more important attributes besides doping that are leading to increasing speeds.

shhhhhhh

We get it. For the 100th time, we get it.

Lance raced clean. Clean riders can ride faster than doped riders. EPO makes no difference. Doped or not, Lance was faster than everyone else.

Trained harder.
Laser like focus.
Weighed food.
Reconnoitered stages.
Best coaches.
Wind tunnels.
P!ssing excellence.

We get it.

 

[MacRoadie]

I chose only 1980 and 2000 for the following reasons:

- my time is limited, or I would have analyzed 30-40 years worth of data
- between these two time periods, road conditions and stage lengths remained stable
- they represent clearly differentiated periods, pre and post introduction of widely available oxygen vector doping products

John Swanson

More importantly, you'll notice you weren't asked why you didn't include 2001-2005...

 

[poupou]

Thanks for your work.

Of course juice is a part of the difference. An other is that in the old days, live TV were just at the end of the stages.

 

[Polish]

More importantly, you'll notice you weren't asked why you didn't include 2001-2005...

Those would be good years to analyze too.
Post EPO Test and the declining usage of EPO by TdF DangerMen.
Can you even notice that EPO use is declining?

Buts lets face it.
2001-2005 are special years.
Can not compare them fairly to any other years.
Lance was a level above.
Heck, two or three or seven levels above lol.

 

[Damiano Machiavelli]

And another potentially interesting thread is tnrown into the crapper by Polish's trolling...

 

[ludwig]

Neat stuff.

Just a random observation. When you compare flat stage speeds, there may be another variable in play. I don't know what the Tour culture was in 1980, but in the Giro (even today) the peloton sometimes collectively decides to take it easy. If this was happening in the Tour in 1980 it impacts your analysis.

 

[slcbiker]

Most of the numbers I have seen for performance improvement on EPO have been 10-15%. Which is pretty significant at a pro level.
It's pretty hard for me to believe 50%.

 

[Escarabajo]

Haven't we talked about this before?

Here. At least related.


http://forum.cyclingnews.com/showthread.php?t=14288

 

[ScienceIsCool]

Most of the numbers I have seen for performance improvement on EPO have been 10-15%. Which is pretty significant at a pro level.
It's pretty hard for me to believe 50%.

See, this is what I also found interesting. One thing to consider is that EPO will enhance FTP by 10-15% (is this correct? I don't have any references handy), while average effort went up by 50%. This means that previously, riders must have been way, way, way below their sustainable max output during a long and flat stage. I think this is reasonable.

Perhaps, because of lack of live TV coverage, riders were using these stages to recover a bit. I've heard words to that effect back in the late 80's when I started racing (though I can't source you any references - treat it as anecdotal only). Or perhaps the riders were simply trashed from hard racing every day and had no choice but to maintain a slower speed. I don't believe it's because they didn't want to race (i.e., down to tactics) - lets face it, it's the Tour and some hungry young rider(s) would have loved to jump out of a slow group and take a stage!

I'd like to suggest that just maybe the biggest advancements due to PEDs isn't raw performance (i.e., climbing stages didn't change much in average speed and top-end sprinting speeds are the same), but the ability to recover from and maintain sustained efforts.

John Swanson

 

[DirtyWorks]

Sean Kelly

Sean Kelly discussed this at length one day on the Tour coverage this year while filling time nearing the end of a stage.

To summarize:
-There's no patron of the peloton essentially controlling the race. In his day and prior (I guess) there was such a thing.
-Racing was pretty slow until 60KM or thereabouts. Now, it's hard from the start.
-There were truly easy days according to Kelly. Now, none.
-Cut off times were much longer.
-Race's format has changed. In other words, ASO has tweaked the incentives to go hard such that it inspires competition from the start. ex. 2011's mid-race bonus sprint.

Also, don't discount tarmac. It seems most new routes (with exception to routes using ASO's Spring events) have new tarmac laid. Editorial: I don't care for this at all. Nothing like random events to make the race interesting.

 

[Damiano Machiavelli]

There was that study a while back that showed that time to exhaustion went up by a whopping 50% with EPO.

 

[Cobblestones]

Sean Kelly discussed this at length one day on the Tour coverage this year while filling time nearing the end of a stage.

To summarize:
-There's no patron of the peloton essentially controlling the race. In his day and prior (I guess) there was such a thing.
-Racing was pretty slow until 60KM or thereabouts. Now, it's hard from the start.
-There were truly easy days according to Kelly. Now, none.
-Cut off times were much longer.
-Race's format has changed. In other words, ASO has tweaked the incentives to go hard such that it inspires competition from the start. ex. 2011's mid-race bonus sprint.

Also, don't discount tarmac. It seems most new routes (with exception to routes using ASO's Spring events) have new tarmac laid. Editorial: I don't care for this at all. Nothing like random events to make the race interesting.

That's what I would guess, plus what was said about the TV coverage. Case in point the Giro where TV coverage just a few years ago was only available for the end of the stage. That's why it was considered a much easier race than the Tour.

I think the analysis is interesting and I would like to encourage the original poster. On the other hand, the increase in speed on flat stages has many driving forces. I'd think that if one could compensate for improvements in aerodynamics (are there drag coefficients available between 1980s and today?) analysis of the ITT might be more suitable to get to the effect of EPO.

 

[Dr. Maserati]

That's what I would guess, plus what was said about the TV coverage. Case in point the Giro where TV coverage just a few years ago was only available for the end of the stage. That's why it was considered a much easier race than the Tour.

I think the analysis is interesting and I would like to encourage the original poster. On the other hand, the increase in speed on flat stages has many driving forces. I'd think that if one could compensate for improvements in aerodynamics (are there drag coefficients available between 1980s and today?) analysis of the ITT might be more suitable to get to the effect of EPO.

Agree - and just to add to yours and 'Dirtyworks' analysis is that teams have much more riders than back in the 80's.

In the 80's a lot of teams had approx 15 riders - nowadays with teams of a minimum of 25 a team can be selected specifically for the Tour and to peak at the Tour.
Also of note is that by Tour time in the 80's the majority of the peloton would have already ridden a Vuelta or Giro prior to the Tour.