Climb comparison for TdF 2004 and 2005

[Robert Chung]

Here are the climbs for the 2004 and 2005 TdFs by category. Note that
total amount of climbing (the dotted lines) doesn't seem to explain the
categorization equally well in the two years.

http://anonymous.coward.free.fr/rbr/cols04-tdf.png
http://anonymous.coward.free.fr/rbr/cols05-tdf.png

 

[DepartFictif]

Of course not. Climb cats aren't a standard, they categorize them
according to steepness, length, distance from finish, amount of KM
covered before the climb, amount of climbs in that stage etc... You
can't look at a climb and because it is 17k long at 9percent say: "it's
a HC". The same climb might be a cat 4 in one race, a cat 3 in
another, and coupld even be uncategorized in another...

 

[gym.gravity]

Robert Chung wrote:
> Here are the climbs for the 2004 and 2005 TdFs by category. Note that
> total amount of climbing (the dotted lines) doesn't seem to explain the
> categorization equally well in the two years.
>
> http://anonymous.coward.free.fr/rbr/cols04-tdf.png
> http://anonymous.coward.free.fr/rbr/cols05-tdf.png

Did they change where some of the climbs started? Approach them from
other more shallow sides? The Madelene is at 20km of 7.9% in 2004 and
25km of 6% in 2005 and the Portet d'Aspet be 5km at 9% in 2004 and 6km
at 7% in 2005.

 

[Robert Chung]

gym.gravity wrote:
> Robert Chung wrote:
>> Here are the climbs for the 2004 and 2005 TdFs by category. Note that
>> total amount of climbing (the dotted lines) doesn't seem to explain the
>> categorization equally well in the two years.
>>
>> http://anonymous.coward.free.fr/rbr/cols04-tdf.png
>> http://anonymous.coward.free.fr/rbr/cols05-tdf.png
>
> Did they change where some of the climbs started? Approach them from
> other more shallow sides? The Madelene is at 20km of 7.9% in 2004 and
> 25km of 6% in 2005 and the Portet d'Aspet be 5km at 9% in 2004 and 6km
> at 7% in 2005.

In 2004 the Madeleine was approached from the south side and the Portet
d'Aspet from the west. In 2005, they came from the opposite directions.

 

[Robert Chung]

BTW, the Rettenbachferner in yesterday's Deutschland Tour averaged 10.7
percent over 12 km.

 

[gym.gravity]

Robert Chung wrote:
> BTW, the Rettenbachferner in yesterday's Deutschland Tour averaged 10.7
> percent over 12 km.

Wow, Jan did well on such steep grades. He's coming to form just in
time.

 

[[email protected]]

BTW, what are you trying to say?

 

[Robert Chung]

[email protected] wrote:
> BTW, what are you trying to say?

Hmmm. It's hard?

 

[[email protected]]

Robert Chung wrote:
> [email protected] wrote:
> > BTW, what are you trying to say?
>
> Hmmm. It's hard?

About this I meant:

> Note that total amount of climbing (the dotted lines) doesn't seem to
> explain the categorization equally well in the two years.

 

[Robert Chung]

[email protected] wrote:
> About this I meant:
>
>> Note that total amount of climbing (the dotted lines) doesn't seem to
>> explain the categorization equally well in the two years.

Oh, that. It's related to a discussion from about a year ago where it was
suggested that total climbing was a moderately good predictor of category.
It turns out that in 2004 total climbing *was* a moderately good
predictor--but it's not as good for 2005. Ben will no doubt seek a
re-expression that will do better.

 

[Tom Kunich]

What about previous Tours back at least to say 1990? I suspect that the
2005 race was off the charts for reasons other than the way the
climbing was situated in the race.

 

[Robert Chung]

Tom Kunich wrote:
> What about previous Tours back at least to say 1990? I suspect that the
> 2005 race was off the charts for reasons other than the way the
> climbing was situated in the race.

Get the data and I'll do the graph.

 

[[email protected]]

Robert Chung wrote:
> [email protected] wrote:
> > About this I meant:
> >
> >> Note that total amount of climbing (the dotted lines) doesn't seem to
> >> explain the categorization equally well in the two years.
>
> Oh, that. It's related to a discussion from about a year ago where it was
> suggested that total climbing was a moderately good predictor of category.
> It turns out that in 2004 total climbing *was* a moderately good
> predictor--but it's not as good for 2005. Ben will no doubt seek a
> re-expression that will do better.

I could do that - 2005 seems to rate steeper but shorter climbs higher
than 2004 - but tweaking models to fit every change in the data is a
theorist's job.

My thought was that 2005 was supposed to have less climbing than 2004
and that ASO had kept the numbers of HC, 1, 2 climbs constant - a sort
of grade inflation. However, your plot suggests that 2005 may have
had roughly as much climbing as 2004. 2005 actually had more climbs
rated HC and 1, so maybe there is still some inflation going on.
One difference between the two years is that 2005 had fewer summit
finishes.