Winner's avg. speed trends in the TdF

[Bavarianrider]

Don't forget that you have to look at total KM and average km per day too. Also how many mountains were in the Tour. Just looking at average speed is missleading. Surfaces have become a lot better too.

 

[BigBoat]

the chart itself is wrong: 2005 is the fastest TDF ever ridden at 41.68 kph....

Yeah the chart is way off. Shortening of stages & improved bikes no doubt helped increase the speed..... Going from steel frames and clips, to the much lighter weight bikes later on.....

The Tour used to be a brute, 1980 for example had a 282 km stage!! Der Kaiser's future team boss won it, Rudy.

 

[ElChingon]

Aren't there a few more factors to consider?

1. The years when they would actually take a lunch break ~30 minutes.
2. Stages that had protests or public blocking the stage.
3. Years where they didn't have two or even one rest day.
4. Years where stages were canceled or neutralized.

I'm not sure those factors would of been included in the overall picture.

 

[ChrisE]

Another point to remember is that aerodynamic drag increases with the square of the increase in velocity. Meaning going 1% faster doesn't take 1% more energy, it takes 2% more (1.01^2). So the same gain in kph is substantially harder to come by from a 39 kph average vice a 38 kph average. Which somewhat accounts for the curve being steeper before 1956 than after 1976.




But back to the OP, what else was going on that might have caused the curve to fall off so suddenly in the mid-60s?

Good catch; the aerodynamic resistance is not constant. Interesting topic, though.

 

[personal]

I've been throwing some numbers in Excel, looking for trends in average speeds of the GC at the Tour de France. To moderate the impact of changes in route and other temporary factors, I used a rolling average of 10 consecutive years. My stats begin with the first season after WWII, 1947, so the first season listed is 1956.
...

When they started riding mostly on paved roads?

 

[la.margna]

total height meters per edition would also be helpful and total paved or unpaved roads etc...

here average speed charted along average stage length... not much more enlightening...

 

[**Uru**]

I would be curious how it breaks down in terms of:
1) Average speed on flat stages each year.
2) Average speed on Cat 4 climbs per year.
3) Average speed on Cat 3 climbs per year.
4) Average speed on Cat 2 climbs per year.
5) Average speed on Cat 1 climbs per year.
6) Average speed on HC climbs per year.

Perhaps breaking out specific climbs or whole stages that have been done in multiple tours may help us see the picture better and determine what may have contributed to higher or lower average speeds.

 

[Square-pedaller]

National vs trade teams

Upthread people remarked a drop in speeds during the sixties. Looking at the graph showing the speeds of individual years, 1962 was the last high year, followed by erratic speeds. A steady upward trend seems to have started again in about 1973.

From 1930, the TdF teams were national teams. In 1962 there was a return to trade teams, but there was a brief return to national teams in 1967 and 1968. It could be conicidence, but these changes could be responsible for the slower speeds.

Return of trade teams

 

[ksmith]

There is an intresting article on the Tours average speed in Bicycle Quarterly. (Summer 2010)

 

[StyrbjornSterki]

I kept this as simple as possible because there are a great many more variables than even the most obtuse of you have pointed out, and I couldn't possibly cover all of them unless I had Powertap data on every tour winner.

To reduce the influence of the evolution of the bicycle, I excluded all the years before HD first allowed derailleurs -- and even the first three of them -- and began with the first post-WWII race.

Relative scale between the two graphs in entirely coincidental, chosen purely
to allow as close an overlap as possible while still showing the extremes.

I'm still using the moving average function because it helps make trends more apparent, only now I have it overlayed with the raw data. And thanks to la.margna for giving me the idea to use average stage length rather than tour length.

The length of the average stage has been noticeably lower and fairly consistent since about 1980. Comparing adjacent year(s) from 1947 to 1979, there was a fairly significant correlation between the direction of change in the average length of a stage and the change in that year's GC's average speed, 73% (excuse me if my tabulations are a bit off, I've looked at this until I'm bleary-eyed). From 1980 on, with stages averaging around 175 km, there's only a 52% correlation. That's little enough to write off to pure chance. Attribute whatever significance to that you will.

The first five years I considered, average (averaged) speed was 32.5 kph and average stage length was 219 km. The five most recent years (the post-Pharmstrong era), average (averaged) speed was 40.0 kph and average stage length was 175 km. Strictly from a standpoint of aerodynamic drag, and disregarding differences in rider stature, that represents an increase in rider output of 51% and a decrease in stage length of 25%.

Remember that drag increases with the square of velocity, so going twice as fast takes (2^2=4) four times as much energy
and going just 1.4x as fast takes twice as much (2^0.5= 1.41).

Since you guys brought them up (and I already have the raw numbers in a spreadsheet), I compared the averages for all TdF wins for Lemond, Indurain and Pharmstrong:

Rider - - - - - AvgSpeed - Avg Stage Length

Lemond- - - - 37.6 kph - - 167 km

Indurain - - - 38.9 kph- - - 185 km

Armstrong - 40.4 kph - - -173 km


Again considering aerodynamic drag alone, Indurain's numbers represent 7% more output and over 11% longer stages than Lemond's.
Armstrong's numbers best Lemond's by 16% and 3%, output and distance.
And while Armstrong's numbers show 8% more output than Indurain's, it's over 7% shorter stages.

And Comparing Pharmstrong to the post-Pharmstrong era, average speed has fallen 2% but average stage length has increased by the same (or -1.9% and +1.6%, if you want to split hairs).

These are admittedly rudimentary numbers and I'm not attributing any overwhelming significance from them. However, to my question from the OP regarding the early 60s dip, average stage length remained above 187 km from 1959 to 1969, averaging 199 km. And there appears to be more a correlation between average speed and average stage length as length approaches 200 km. So that would seem a reasonable explanation for my plateau. That and the arrival of the trade teams.

 

[Escarabajo]

I kept this as simple as possible because there are a great many more variables than even the most obtuse of you have pointed out, and I couldn't possibly cover all of them unless I had Powertap data on every tour winner.

To reduce the influence of the evolution of the bicycle, I excluded all the years before HD first allowed derailleurs -- and even the first three of them -- and began with the first post-WWII race.

Relative scale between the two graphs in entirely coincidental, chosen purely
to allow as close an overlap as possible while still showing the extremes.

I'm still using the moving average function because it helps make trends more apparent, only now I have it overlayed with the raw data. And thanks to la.margna for giving me the idea to use average stage length rather than tour length.

The length of the average stage has been noticeably lower and fairly consistent since about 1980. Comparing adjacent year(s) from 1947 to 1979, there was a fairly significant correlation between the direction of change in the average length of a stage and the change in that year's GC's average speed, 73% (excuse me if my tabulations are a bit off, I've looked at this until I'm bleary-eyed). From 1980 on, with stages averaging around 175 km, there's only a 52% correlation. That's little enough to write off to pure chance. Attribute whatever significance to that you will.

The first five years I considered, average (averaged) speed was 32.5 kph and average stage length was 219 km. The five most recent years (the post-Pharmstrong era), average (averaged) speed was 40.0 kph and average stage length was 175 km. Strictly from a standpoint of aerodynamic drag, and disregarding differences in rider stature, that represents an increase in rider output of 51% and a decrease in stage length of 25%.

Remember that drag increases with the square of velocity, so going twice as fast takes (2^2=4) four times as much energy
and going just 1.4x as fast takes twice as much (2^0.5= 1.41).

Since you guys brought them up (and I already have the raw numbers in a spreadsheet), I compared the averages for all TdF wins for Lemond, Indurain and Pharmstrong:

Rider - - - - - AvgSpeed - Avg Stage Length

Lemond- - - - 37.6 kph - - 167 km

Indurain - - - 38.9 kph- - - 185 km

Armstrong - 40.4 kph - - -173 km


Again considering aerodynamic drag alone, Indurain's numbers represent 7% more output and over 11% longer stages than Lemond's.
Armstrong's numbers best Lemond's by 16% and 3%, output and distance.
And while Armstrong's numbers show 8% more output than Indurain's, it's over 7% shorter stages.

And Comparing Pharmstrong to the post-Pharmstrong era, average speed has fallen 2% but average stage length has increased by the same (or -1.9% and +1.6%, if you want to split hairs).

These are admittedly rudimentary numbers and I'm not attributing any overwhelming significance from them. However, to my question from the OP regarding the early 60s dip, average stage length remained above 187 km from 1959 to 1969, averaging 199 km. And there appears to be more a correlation between average speed and average stage length as length approaches 200 km. So that would seem a reasonable explanation for my plateau. That and the arrival of the trade teams.

Excellent Job StyrbjornSterki (Gosh that was hard to spell).

I can see the abrupt change from the 90's and the 80's. After the 90's started you can see the gradual change implicating the potential increase in the abuse of EPO plus the perfectioning of the doping programs.

We can see also some other changes taking place from the 70's and the 80's but that is more related with technology and other factors.

Your estimated increase in percentages agrees with the speculated increase of power given by the newer drugs.

Before we had the graph performed on power/kg by the main contenders and if I remember right there has been big discussion about the first two years won by Indurain where the power increase was not so obvious. But maybe we can combine with these charts and make a better conclusion.

Here it is:

Here is the complete link if you want to read some more:

http://www.sportsscientists.com/2009/07/tour-de-france-2009-power-estimates.html[/QUOTE]

Your chart actually shows the difference between the end of the 80's and the 90's which was not clear in this other chart. This gives credit to Lemond saying that since you ride faster with EPO in every terrain if you are riding clean you could be exhausted by the time you reach the mountains.

Is it is not too much to ask, can you please put the dots on each year.

Thanks.

 

[FoxxyBrown1111]

If i read the graphs correctly, from 1966 to 1977 the Avg-Km AND Avg-Speed got down. No one really could explain it up to now. It absolutely makes no sense (that´s more mystic than all "Epo- and Blood years" from 1990-2006).

Could it be that the doping controls started in 1967? Man if that´s true, it seems doping had at least an as big effect up to 1966 as the "Hi-TechDoping and-Blood-Years".
May Dr. Maserati can help out here.

Back to the OP-Graph. I can´t see any influence from EPO (1990+) compared to the 80´s/late 70´s. It was a steady improvement in km/h from the late 70´s to the mid 2000´s. No "special hit" at around 1990. I think to use the 10-Yr-Avg is very good, coz it includes the steady process of lighter bikes, less mountains, wider talent pool, etc. It makes a smooth transition so the numbers are not poluted.

Conclusion: Either Epo is overrated or ampethamin/steroid mixes had the same effects.
The only "true hit" (compared to the 60-70 downfall) is the decreasing speed since 2006. It indeed looks like the passport works at least so much, that it prevents catastrophic over use of doping....

Anyway, great work Styrbjorn.

My recommondation is: I think you really should compute a total mountain milage vs. lenght of tours vs. improvement of technique vs. lenght of stages compared to Avg.-Speed formula. I can´t do it. Because unluckily i am not a mathematican. But i am sure you can do it, and once and for all answer the impact of doping question. That would be nobel prize worthy

 

[Escarabajo]

If i read the graphs correctly, from 1966 to 1977 the Avg-Km AND Avg-Speed got down. No one really could explain it up to now. It absolutely makes no sense (that´s more mystic than all "Epo- and Blood years" from 1990-2006).

Could it be that the doping controls started in 1967? Man if that´s true, it seems doping had at least an as big effect up to 1966 as the "Hi-TechDoping and-Blood-Years".
May Dr. Maserati can help out here.

Back to the OP-Graph. I can´t see any influence from EPO (1990+) compared to the 80´s/late 70´s. It was a steady improvement in km/h from the late 70´s to the mid 2000´s. No "special hit" at around 1990. I think to use the 10-Yr-Avg is very good, coz it includes the steady process of lighter bikes, less mountains, wider talent pool, etc. It makes a smooth transition so the numbers are not poluted.

Conclusion: Either Epo is overrated or ampethamin/steroid mixes had the same effects.
The only "true hit" (compared to the 60-70 downfall) is the decreasing speed since 2006. It indeed looks like the passport works at least so much, that it prevents catastrophic over use of doping....

Anyway, great work Styrbjorn.

...

Honestly FoxxyBrown, I see a peak starting in 1990 forward. It is very clear. That's why the OP estimated an 8% increase to the first section (Indurain's Tour) and 16% to the second section. It is clear in the plot.

About the other changes I am sure it has nothing to do with doping or doping control. I am not familiar with the history of the Tour in the 70's and 60's so somebody else would have to pitch in and give a better assessment on the drop.

 

[FoxxyBrown1111]

Yes in 1990 there is a little spike in km/h, but i also see a big drop around there in km per stage. If somebody would do a correlation, i guess there will be no difference between the late 70s, 80s, and early 90s. So what i truly see is either Epo is overrated or the 80s drugs (mostly amph. and steroids) are underrated.
Later in the 90´s when the stages got longer again, the improvment of km/h doesn´t go up as much.

Much more of a spike i see from 1999.

But all those are nothing compared to the mid60-mid70 drops. Unless someone else comes up, it must have something to do with big time fear (Simpson´s death and new doping controls).

 

[Polish]

What are your thoughts?

Great chart Styrbjorn.

Four major parts:

1) About 10 years of Increasing speeds from 56-66.
2) About 10 years of Decreasing speed from 66-76
3) 30 years of steadily Increasing speeds from 76-06
4) Flat since 2006

The steadily increasing speeds from 76-06 are due to many things of course.
But I feel the focus on "hi-tech" recovery played a big part.
The "Needle Era" for lack of a better name.
Interesting the blip up in speed in 1990 when Greg won
And the blip down in speed in 2002 when Lance won.
Otherwise a very smooth chart.

Of course, what catches my eye - is the PEAK in 2005/2006.
Is that the limit of AWESOMENESS?

Will that 10 year moving average peak EVER be matched again?

Maybe not.
Would not suprise me in the least.
The limit of human cycling awesomeness.

 

[Sylvester]

Threads like these are the ones that keep me addicted to this forum. Keep going!

 

[la.margna]

thanks to la.margna for giving me the idea to use average stage length rather than tour length.

That's alright. Thanks for starting a subject where we could calculate, discuss, chart, analyse without end! Wish we had all data to take this stat-frenzy massively further!

 

[Teddy Boom]

Remember that drag increases with the square of velocity, so going twice as fast takes (2^2=4) four times as much energy and going just 1.4x as fast takes twice as much (2^0.5= 1.41).

Drag varies by the square of velocity, but the power required to maintain speed is the product of drag times velocity, so the power required to overcome wind resistance varies as the cube of velocity.

Interesting thread though

[edit: corrected my own square-cube error ]

 

[Zweistein]

Great chart Styrbjorn.

Four major parts:

1) About 10 years of Increasing speeds from 56-66.
2) About 10 years of Decreasing speed from 66-76
3) 30 years of steadily Increasing speeds from 76-06
4) Flat since 2006

The steadily increasing speeds from 76-06 are due to many things of course.
But I feel the focus on "hi-tech" recovery played a big part.
The "Needle Era" for lack of a better name.
Interesting the blip up in speed in 1990 when Greg won
And the blip down in speed in 2002 when Lance won.
Otherwise a very smooth chart.

Of course, what catches my eye - is the PEAK in 2005/2006.
Is that the limit of AWESOMENESS?

Will that 10 year moving average peak EVER be matched again?

Maybe not.
Would not suprise me in the least.
The limit of human cycling awesomeness.

The biggest tech improvements were areo bars and reduced spoke and deep dish wheels. People always prance about the reduced weight but bikes total weight has always been a small fraction of the total weight. Stripping 3 pounds off a bike isn't going to increase your climbing by 3%. Old bikes weren't that heavy. too. I have an old Colnago that is steel that weighs 18lbs. You don't need carbon, aluminium, or titanium to have a light bike.

Another thing to take into consideration is the fact that Hinault was a patron. If he said to ride slow, everyone rode slow. Always the course and the participates play a big role in determining the speed of the race. In the 80s, you had epic full throttle battles between fignon, hinault, lemond, roche, and delgado and they didn't save everything just for the final hill.

Someone should take these data points and cube them to give a scaled comparison of how much more power has been put out by the riders. This I believe is still poor for comparison but I think climb times like up the Alpe de Huez are much better for comparison. Riders from the 90s on have been climbing with motorcycles.

 

[Shardi]

the stagnation at 2006 and onwards indicate that after the 1995 edition, the max speed has been reached for the period 1996-2010, for a 10-year-average. Cudos Bjarne for setting the standard...

 

[Escarabajo]

Drag varies by the square of velocity, but the power required to maintain speed is the product of drag times velocity, so the power required to overcome wind resistance varies as the cube of velocity.

Interesting thread though

[edit: corrected my own square-cube error ]

+1.

I was getting ready to post the same thing yesterday. But I did not have time.

In order to estimate closer numbers to power we really need more details than this, but the chart itself without these calculations is very telling.

Drag Power: 0.5*Cd*air density*A*v^2*d/dt

Cd= drag Coefficient
A= frontal area
v= Drag speed
d/dt= distance over unit time which is also the same velocity.

Again a correction would have to be made for the mountain stages.

 

[Escarabajo]

The biggest tech improvements were areo bars and reduced spoke and deep dish wheels. People always prance about the reduced weight but bikes total weight has always been a small fraction of the total weight. Stripping 3 pounds off a bike isn't going to increase your climbing by 3%. Old bikes weren't that heavy. too. I have an old Colnago that is steel that weighs 18lbs. You don't need carbon, aluminium, or titanium to have a light bike.

....

+1.

I have done these calculations and the weight difference is not going to make for that speed difference. The competition that enters the race and how they race it is far more important.

 

[JMBeaushrimp]

With the number of variables that keep getting brought up, this is starting to look a whole lot closer to chaos theory.

Yay, Chaos!

 

[FoxxyBrown1111]

Ah JM. It´s like the Heisenberg theroy. The closer you look the more unclear things become.

We should go back to Post '1 graph and only need a answer to the speed drops of the 70s. Everything welse is smooth and clear, even the 2006+ downfall.

 

[Albatros]

Great chart Styrbjorn.

Four major parts:

1) About 10 years of Increasing speeds from 56-66.
2) About 10 years of Decreasing speed from 66-76
3) 30 years of steadily Increasing speeds from 76-06
4) Flat since 2006

The steadily increasing speeds from 76-06 are due to many things of course.
But I feel the focus on "hi-tech" recovery played a big part.
The "Needle Era" for lack of a better name.
Interesting the blip up in speed in 1990 when Greg won
And the blip down in speed in 2002 when Lance won.
Otherwise a very smooth chart.

Of course, what catches my eye - is the PEAK in 2005/2006.
Is that the limit of AWESOMENESS?

Will that 10 year moving average peak EVER be matched again?

Maybe not.
Would not suprise me in the least.
The limit of human cycling awesomeness.

Have you factored the difficulty of the Tour each year? Whether the stages were shorter or longer, the number of stages, whether there was more or less mountains etc. That could affect the speed average and explain its decline during the late 60's and early 70's.