Power Data Estimates for the climbing stages

[Alex Simmons/RST]

I think you are still missing the point.

Despite not knowing the wind conditions, the estimates, compared to the actually recorded data, is within +/- 2.7%. I consider the confounding factors accuracy mostly irrelevant when your overall accuracy is that close.

I notice too, despite the bragging about segment by segment analysis that you do not provide even a summary of your accuracy.

Telling.

I make no brag. Indeed I make the point that there will be errors in what I do because there are "known unknowns". The more detailed approach I outline provides for a means to better account for them if and when more data on those unknowns becomes available.


And again, the point I am making is that even if the data were 100% spot on all the time, will that make a difference?

 

[Alex Simmons/RST]

For reference, I updated a chart I did some time back on the average speed of 5 fastest riders each year the TdF went up AdH since 1982:

Comments, data and source acknowledgements in this short blog item.

 

[Dear Wiggo]

Good to see speeds are down from the EPO fueled Lance era.

 

[ralphbert]

For reference, I updated a chart I did some time back on the average speed of 5 fastest riders each year the TdF went up AdH since 1982:

Comments, data and source acknowledgements in this short blog item.

Why the limited number for the 80's assents? I presume the times would be significantly lower if you incl 5 as per the other years?

 

[pmcg76]

Why the limited number for the 80's assents? I presume the times would be significantly lower if you incl 5 as per the other years?

Yes, it's based on the Top 200 ascent times for Alpe d'Huez so there were insufficent times from that period in the Top 200.

But then if you stretched it out to Top 10, the drop-off in the Bio-Passport era would be more pronounced.

I listed the amount of times in the Top 50/100/100 from the early 00s v Bio-Passport era on the previous stage.

 

[Alex Simmons/RST]

Why the limited number for the 80's assents? I presume the times would be significantly lower if you incl 5 as per the other years?

The original data set only listed those few riders as presumably the others in those years fell outside of the top 200.

Hence why I noted that limitation, and yes if we did have the data for the other riders in those years then you would expect the average speed to be lower but by how much I don't know.

The bikes back then were heavier, so that would drop speeds by roughly 0.5km/h compared with today.

 

[Merckx index]

My argument is that we already know who the top professional riders are, and whether or not they have power meters, release their power data or estimates are made, it won't make a sod of difference to the anti-doping effort.

So you think if Froome released power data for pre- and post-2011 Vuelta, and there was, say, a 15% difference, this wouldn’t have any effect at all on what people think about his performances? It’s not all about catching dopers, you know, or so people like JV argue. It’s also about creating a climate in which omerta is discouraged, and an important part of this is identifying suspicious performances, even if the evidence isn’t good enough for a sanction.

How much reduction in doping do you think is accomplished via positive tests, and how much via creating an environment in which it’s no longer worth while taking the risk, either of getting caught or being the source of heavy, widespread suspicion? Even if a rider doesn't get caught, if he becomes suspicious enough, teams may worry that he eventually will get caught, and be reluctant to hire him. Look at Horner, for heavens sake.

But let's say you knew a rider's GME. Then what?

With that and V02 max, we would have a pretty good idea of his maximum possible FTP. If we knew everyone’s ME and V02max, we could indeed draw a fairly clear line in the sand.

[the] post you are referring to says nothing about where the lines in the sand are, rather it simply demonstrates the nature of the relationship of GME, VO2max, fractional utilisation of VO2max, and just how blurry those lines in fact are.

See above. If we know all these numbers, of course we can use them to draw lines.

[

I can't and don't speak for Coggan, Walsh or Sky.

But what you say on this subject pretty much echoes what they say.

i. estimates of power from climbing times are subject to errors, some factors are not measured and are unknown and hence an error analysis should be provided with such estimates. Some unknowns are not overly significant, but some are. I consider that the estimates that are regularly published imply a false level of precision. This is a separate issue to how one then goes on to interpret such data.

All of which, as I said before, simply begs for more transparency on power tap numbers.

Now we really don't need power estimates when comparing the same climb over long periods, as the climb times alone are sufficient for the purpose (and come with a better level of precision). Then all you do is look at the trends and note variations from year to year of the group averages (which might be to do with race context, environmental conditions, level of doping impact etc) as well as relative performance of individuals in any given year.

Obviously, both times and power data have their advantages and limitations, at the least we can argue that having both is more informative than having only one or the other.

The reason power estimates are made is to see if comparisons can be made between different climbs. That's when the issue of precision of estimates becomes a little more problematic because you are now estimating power with some unknowns as inputs. The level of precision should be conveyed, that way we can then see how power output estimates are really ranges rather than a point value.

I don’t have a problem with this. I have a problem with not conveying any data at all.

It's very important not to place too much reliance on an individual data point, rather use of such data (e.g. climbing times) should be to examine the overall trends. Ross Tucker has gone to great pains to make this point, I think he uses the term "pixellation" but I might have the exact phrase wrong. Yet time and again people fall into the trap of being focussed on single data points/estimates. I'm not saying you do this, but rather I see people place more reliance on individual data points than they should.

But observers have long done this, and sometimes it’s helpful in focusing on a problem. LA’s climb up Sestriere was a single data point, yet it was one that led many sports insiders to believe for the first time that he wasn’t clean. It opened the door to closer examination. There have been a number of analyses in which multiple climbs by a single rider have been used, I posted one here earlier this year.

Even if you had the rider's (or riders') data, it's still not going to make a difference to the anti-doping effort. Tell me, let's say a rider released all of his power meter data (or that we had precise estimates). Now what?

Let me throw the question back at you. Suppose we had complete transparency, and as a result, we were able to show that no rider ever was capable of putting out, say, more than 6.4 watts/kg for 30-40 min. The numbers simply didn't support anything greater. Then a rider comes along who does 6.7-6.8 on some long climb. Not an outlier, with a combination of ME and V02max never before seen, but someone who did something that the numbers say is impossible.

No, we can’t sanction him without a positive test, but do you really think this would be useless knowledge? I don’t. See also my first response, above.

I'm most definitely not fine with the Impey case, so let me dispel the notion that you infer I am. A read of my posts on the Impey threads should make that pretty clear, but then I am also not actually making the points you suggest I am, so I can't say whether of not there is a correlation between those that do and how they view cases such as Impey's.

I didn’t imply that you were fine with it, I just pointed it out as an example of the problems with relying on anti-doping tests.

 

[Alex Simmons/RST]

Good to see speeds are down from the EPO fueled Lance era.

Well if you consider the bike weight penalty of the 1980s versus in the last decade or so, then the times of the top riders for the 1980s are pretty much on a par with the fastest 5 average speed from 1999, 2003, 2008, 2011 and 2013.

Make of it what you will.

I'd be happy to do similar charts for other climbs if I had enough historical data.

I use average of the fastest 5 to look at a group trend as I thought it would represent a reasonable balance between minimising a lone rider skewing the dataset versus not including those who are not making a maximal effort for the duration.

As for impact of wind differences, yes the relative wind vector constantly changes up the climb, all the boundary layer effects and so on.

Still there can be differences year to year in the amount of wind assistance.

In case of ADH, a difference in average wind assistance of only 0.5m/s (move your hand along at 0.5m/s to get a sense of little wind that is) would account to a difference of ~0.25km/h in rider ascent speeds (at same power).

 

[Dear Wiggo]

That's some mythical following wind that curves around each hair pin bend, avoiding all the spectators, Alex.

Is it generated by unicorns, by any chance?

 

[Alex Simmons/RST]

So you think if Froome released power data for pre- and post-2011 Vuelta, and there was, say, a 15% difference, this wouldn’t have any effect at all on what people think about his performances?

I suspect people will probably simply overlay their existing bias / pre-conceived idea on it and not learn much.

If I showed you my data which showed a 15% difference in power, what would you make of it?

I would only comment if I understood more detail about the context of the data.

For example, during a season, a variance in threshold power of 10% for a bike racer is pretty typical, and 15% is certainly not completely unusual depending on how their season plays out. Now as I don't know the specifics wrt to Froome, so I can't really say.

It’s not all about catching dopers, you know, or so people like JV argue. It’s also about creating a climate in which omerta is discouraged, and an important part of this is identifying suspicious performances, even if the evidence isn’t good enough for a sanction.

Fair point.

Yet when you look at those ADH times in the 1980s and allow for the change in bike weights, they are similar to some of the times during Armstrong's reign. So what sense are we to make of such comparisons? What's a sound baseline to use?

How much reduction in doping do you think is accomplished via positive tests, and how much via creating an environment in which it’s no longer worth while taking the risk, either of getting caught or being the source of heavy, widespread suspicion? Even if a rider doesn't get caught, if he becomes suspicious enough, teams may worry that he eventually will get caught, and be reluctant to hire him. Look at Horner, for heavens sake.

I think if there were adequate testing and many more positives as a result, then that would be a far more powerful disincentive.

As it is, it would appear wide spread suspicion hasn't changed much.

All we can really say is that the physiological impact, overall on average, has been reduced from the outrageous period in the 1990s. That however doesn't tell us whether or not the prevalence of doping is more/less/same.

But I agree, if the people responsible for choosing/guiding/directing riders are more attuned to such things and are ethical, then it will help.

Problem is I'm not convinced we have enough ethical people involved to provide that critical mass. More than half the DS's at last TdF had a known doping past, so one would expect they'd already know what to look for and don't need power estimation help.

With that and V02 max, we would have a pretty good idea of his maximum possible FTP. If we knew everyone’s ME and V02max, we could indeed draw a fairly clear line in the sand.

See above. If we know all these numbers, of course we can use them to draw lines.

Given the state of human knowledge on the subject, those lines will still be pretty thick and blurry.

But what you say on this subject pretty much echoes what they say.

I'll have to take your word for it as I really don't know/recall what Walsh and Sky say, and they certainly haven't have influenced my comments. I am reasonably familiar with what Andy Coggan has written.

All of which, as I said before, simply begs for more transparency on power tap numbers.

Obviously, both times and power data have their advantages and limitations, at the least we can argue that having both is more informative than having only one or the other.

I don’t have a problem with this. I have a problem with not conveying any data at all.

Well I've no problems with transparency of power data either. I just don't think it'll add much to we already know wrt doping. There are of course many practical considerations with doing so, e.g. Sky uses Stages power meters, data doping etc.

But observers have long done this, and sometimes it’s helpful in focusing on a problem. LA’s climb up Sestriere was a single data point, yet it was one that led many sports insiders to believe for the first time that he wasn’t clean. It opened the door to closer examination.

I'll have to take your word for it that this one ride was the tipping point wrt observers of Armstrong. I don't know.

There have been a number of analyses in which multiple climbs by a single rider have been used, I posted one here earlier this year.

Then you are beginning to inspect trends rather than single data points, which is good. How good depends on how many data points and their context.

Let me throw the question back at you. Suppose we had complete transparency, and as a result, we were able to show that no rider ever was capable of putting out, say, more than 6.4 watts/kg for 30-40 min. The numbers simply didn't support anything greater. Then a rider comes along who does 6.7-6.8 on some long climb. Not an outlier, with a combination of ME and V02max never before seen, but someone who did something that the numbers say is impossible.

Not sure I understand the premise of the question. 6.4W/kg has been sustained for an hour, but I am assuming you mean what if that were the best W/kg we had observed for 30-min (or whatever arbitrary time frame is chosen).

We might well observe maximal performance from historical data, but that doesn't mean better performance isn't physiologically possible.

If we see a given W/kg (or just W) for a rider, then there must by inference have been a combination of VO2max, GME and fractional utilisation of VO2max (and metabolic energy yield from O2) that led to that performance.

It's not like GME, VO2max and fractional utilisation of VO2max are static things, they change during the course of a season and from season to season. Some more than others, but they change. That's why a rider's threshold power varies.

Now if the W/kg is much higher than previously observed, then of course one will wonder how.

Just what are the limits?
GME?
24% is not outrageous for a dominantly slowtwitch beast (there are reports of higher)

VO2max?
Lemond was reported at >92ml/kg/min (others have reported higher)

Fractional VO2max for 30-min?
I'd imagine in excess of 90% is feasible for 30-min for a rider in peak aerobic condition after years of quality training.

Energy yield per litre of O2?
I used a pretty standard 21.1 kJ per litre of O2

That combination would net 6.92W/kg.
If fractional utilisation was higher, we get > 7W/kg.

But are all of those known measured examples feasible in the same person? I don't know, and I don't think anyone else does either.

Is it probable? Well for most, of course not, but we're talking about outliers.

The question then remains, is doping part of the picture when such a power number emerges? Quite possibly, indeed quite probably. But it's not an absolutely certainty based on what we know of human physiology.

No, we can’t sanction him without a positive test, but do you really think this would be useless knowledge? I don’t. See also my first response, above.

The knowledge isn't useless. But I still don't see it really aiding anti-doping all that much. But we can agree to disagree on that.

I didn’t imply that you were fine with it, I just pointed it out as an example of the problems with relying on anti-doping tests.

OK, thanks and I agree that the current testing methods and processes are woefully inadequate.

 

[Alex Simmons/RST]

That's some mythical following wind that curves around each hair pin bend, avoiding all the spectators, Alex.

Is it generated by unicorns, by any chance?

yes, they are all lined up along the course and fart mexican wave style as the riders pass.

Your knowledge of wind is clearly vastly superior, so please do share your observations of what actually occurs.

You do realise that you can't even feel a wind as slight as 0.5m/s?

Look, I don't know how much wind there is and neither do you, but I know that when I see shots like this one, view for about 20secs and see the flags, then even at rider level there will definitely be air movement. Indeed the channel of people and mountain rock face can in some instances amplify the wind velocity, and can also create stall zones.

https://www.youtube.com/watch?v=rsO8dgU4XPg#t=579

I'm sure if you watch the whole thing you'd see signs of wind in various directions relative to the riders. What that actually translates to, who knows?

 

[Kicker661]

Is it probable? Well for most, of course not, but we're talking about outliers.

The question then remains, is doping part of the picture when such a power number emerges? Quite possibly, indeed quite probably. But it's not an absolutely certainty based on what we know of human physiology.

This is the part i never understand.

Why is it that we should set aside everything we 'do' know in case an event arises that we have never seen before?..... 7 billion people on the planet yet we can't draw a line in the sand because the 7,000,000,001 person born could be 0.5% better than everyone else?

Humans don't evolve over 20yrs. During our rather short lifespans (in terms of Human race), we change our environment to allow us to do things we otherwise couldn't.

If an outlier who is doped out of their brains rider gets 6.4w/kg then that's the line in the sand. A natural bloke isn't going to beat it, not in our lifetime.

 

[Dear Wiggo]

I'm sure if you watch the whole thing you'd see signs of wind in various directions relative to the riders. What that actually translates to, who knows?

Strictly speaking, no, we can't know.

But I do find it incredibly curious that you choose the rather seemingly impossible scenario that it followed the riders all the way up, non-stop, as your "devil's advocate" example of "what-if" scenarios.

I prefer Velo's example where the wind stays pretty much consistent in terms of direction and alternates between headwind and tailwind, netting an underestimate of power. Contrary to some claims it's not because this preference supports a narrative, but because it supports what I have found in real life, and seems more reliably realistic.

 

[Alex Simmons/RST]

This is the part i never understand.

Why is it that we should set aside everything we 'do' know in case an event arises that we have never seen before?..... 7 billion people on the planet yet we can't draw a line in the sand because the 7,000,000,001 person born could be 0.5% better than everyone else?

Humans don't evolve over 20yrs. During our rather short lifespans (in terms of Human race), we change our environment to allow us to do things we otherwise couldn't.

If an outlier who is doped out of their brains rider gets 6.4w/kg then that's the line in the sand. A natural bloke isn't going to beat it, not in our lifetime.

What about the doper who rides at 5.8W/kg so they can comfortably ride in the grupetto and stay fresh for the flatter stages?

The line isn't all that helpful now, is it?

Indeed the doping plausibility line is focussed only on the very top handful of performers, and well if we are not already assessing their doping status effectively, then what's the point?

 

[Alex Simmons/RST]

Strictly speaking, no, we can't know.

But I do find it incredibly curious that you choose the rather seemingly impossible scenario that it followed the riders all the way up, non-stop, as your "devil's advocate" example of "what-if" scenarios.

No, I don't think I've ever suggested that's the case. Rather I am talking about the errors for instantaneous power calculations, or individual segments with same conditions if you like.

If you look at the chart I reposted a couple of pages back, you'll see it shows the impact of wind on the power / speed relationship. It makes no suggestion about the total climb time.

I've always said the relative wind vector will vary, and it's an unknown.

I prefer Velo's example where the wind stays pretty much consistent in terms of direction and alternates between headwind and tailwind, netting an underestimate of power. Contrary to some claims it's not because this preference supports a narrative, but because it supports what I have found in real life, and seems more reliably realistic.

Yeah, that's reasonable for calm to very light wind days although it seems from the study data I've seen (and no doubt I've not seen all of it) that the estimates show both over- and under- estimations compared with power meter measurements.

I don't know what work was done to assess the validity of data from the meters used (quite a range of considerations there) - of course that may work either way, i.e. the results might align even better, or they may not.

I do still think we need to be careful when looking at individual data points as the overall average wrt accuracy doesn't tell us the error for any one point.

 

[Dear Wiggo]

No, I don't think I've ever suggested that's the case.

It's exactly what you suggested.

 

[Alex Simmons/RST]

It's exactly what you suggested.

Where?

I checked back to look at my original blog item on the topic written over four years ago and I didn't say it then, indeed I made the point about variable wind vector and suggested use quoting ranges of power or indicator the error would be worthwhile.

 

[GoodTimes]

I do still think we need to be careful when looking at individual data points as the overall average wrt accuracy doesn't tell us the error for any one point.

I havent read all the bloggers out there, but it seems to me that the following makes sense:

1. use power estimate model to come up w estimated power for a climb
2. validate model by comparing estimated power to acheived power. do this for a large sample size. 100 climbs, on different mountains, by different riders, in different weather conditions is a good start. more samples will reduce variation found in 5.
3. use differences found in 2 to assess whether therer is any significant bias in model.
4. if so, revise model to account for bias.
5. determine variability of normalized power estimates via student T distribution.
6. determine what confidence level is required for the conversation. 95% is typical.
7. determine error in power model, at this confidence level.
8. use model to compare different climbs. use error to assess whether there is statstical significance between different climbs analyzed.

none of the above requires consideration of wind vector variabilty, and wind probability functions. yhis may be useful as part of the process of generating model, but not as part of validating it.

If the above is done, we may appy the model to a single performance, and say, with 95 % confidence, that the actual power output is y +/- x.

This seems like it would be a good engineering masters project.

 

[pmcg76]

For reference, I updated a chart I did some time back on the average speed of 5 fastest riders each year the TdF went up AdH since 1982:

Comments, data and source acknowledgements in this short blog item.

Alpe d'Huez

Ok was bored so did my own rudimentary calculation on average Top 5 times. They reflect what was in the graph, in actual figures of average time of Top 5.

1999: 41.30
2001: 39.24
2003: 40.49
2006: 39.03
2008: 41.14
2011: 41.50
2013: 40.31

For interest
1989: 43.07

Average Top 5 time for 2001-03-06 is 39.57
Average Top 5 time for 2008-11-13 is 41.08

So the average time for the Bio-Passport era is over a minute slower than the early-mid 00s.

Lets substitute 1999 for 06 and call it the Armstrong era.
Average Top 5 time for 199-01-03 is 40.49, still 20 seconds faster than the Bio-Passport era.

The other thing I analysed was the Top 10 times from the slowest early 00s ascent and fastest Bio-Passport ascent.

So 2003: Top 5: 40.49 Top 10: 41.21
2013: Top 5: 40.31 Top 10: 41.25

So even though the Top 5 in 2013 was faster than 2003, the Top 10 fell just slower in 2013.

All useless info of course but what the hell.

 

[Kicker661]

What about the doper who rides at 5.8W/kg so they can comfortably ride in the grupetto and stay fresh for the flatter stages?

The line isn't all that helpful now, is it?

Indeed the doping plausibility line is focussed only on the very top handful of performers, and well if we are not already assessing their doping status effectively, then what's the point?

Yes this thread is focused on the top handful so the line in the sand relates to their performances.

I think it's obvious they are not assessing doping status accurately when 'clean' riders are beating or matching the top known doped riders times. A top GT clean rider is not going to come close to a top GT doped rider. So therefore why can't they introduce power as an additional means of control? Isn't that the point? To improve doping controls?

 

[Alex Simmons/RST]

Yes this thread is focused on the top handful so the line in the sand relates to their performances.

I think it's obvious they are not assessing doping status accurately when 'clean' riders are beating or matching the top known doped riders times. A top GT clean rider is not going to come close to a top GT doped rider. So therefore why can't they introduce power as an additional means of control? Isn't that the point? To improve doping controls?

I don't see how it's going to improve doping controls. We already know who the top riders are that should be subject to doping controls. They are the ones who win/place in the big races. In fact they earn a salary from racing bikes.

Unless you mean using a rider's performance as a trigger for doping sanction? That just isn't going to happen.

I think the main point of power numbers/estimates is to have more pub chat fodder. The extent that it creates a better anti-doping culture within the professional peloton and its support network, well perhaps there is some small merit in that.

I think the focus should be on why they are not getting caught and putting resource and effort into addressing that.

It's pretty well known that the tendency to break a rule of this nature is strongly inversely correlated with the perceived risk of being caught and much less so with the actual sanction (just look at road safety campaigns as prime examples).

So a big part of the answer lies in increasing the the perceived risk of being caught. We can look at power numbers / climb times all we want, but unless that means a rider will be actually caught doping, it's not really helping much.

 

[Fearless Greg Lemond]

Alpe d'Huez

Ok was bored so did my own rudimentary calculation on average Top 5 times. They reflect what was in the graph, in actual figures of average time of Top 5.

1999: 41.30
2001: 39.24
2003: 40.49
2006: 39.03
2008: 41.14
2011: 41.50
2013: 40.31

For interest
1989: 43.07

Average Top 5 time for 2001-03-06 is 39.57
Average Top 5 time for 2008-11-13 is 41.08

So the average time for the Bio-Passport era is over a minute slower than the early-mid 00s.

Lets substitute 1999 for 06 and call it the Armstrong era.
Average Top 5 time for 199-01-03 is 40.49, still 20 seconds faster than the Bio-Passport era.

The other thing I analysed was the Top 10 times from the slowest early 00s ascent and fastest Bio-Passport ascent.

So 2003: Top 5: 40.49 Top 10: 41.21
2013: Top 5: 40.31 Top 10: 41.25

So even though the Top 5 in 2013 was faster than 2003, the Top 10 fell just slower in 2013.

All useless info of course but what the hell.

Certainly not useless but you are disregarding race tactics. Look at what a cleanish Zulle was able to do in 1999 versus his former Epo fuelled performances. Three minutes slower? Armstrong didnt have to go faster but I bet he could have if neccesary.

Thats why you should allways ask 'who is doing what where and why'.

 

[sniper]

Alpe d'Huez

Ok was bored so did my own rudimentary calculation on average Top 5 times. They reflect what was in the graph, in actual figures of average time of Top 5.

1999: 41.30
2001: 39.24
2003: 40.49
2006: 39.03
2008: 41.14
2011: 41.50
2013: 40.31

For interest
1989: 43.07

Average Top 5 time for 2001-03-06 is 39.57
Average Top 5 time for 2008-11-13 is 41.08

So the average time for the Bio-Passport era is over a minute slower than the early-mid 00s.

Lets substitute 1999 for 06 and call it the Armstrong era.
Average Top 5 time for 199-01-03 is 40.49, still 20 seconds faster than the Bio-Passport era.

The other thing I analysed was the Top 10 times from the slowest early 00s ascent and fastest Bio-Passport ascent.

So 2003: Top 5: 40.49 Top 10: 41.21
2013: Top 5: 40.31 Top 10: 41.25

So even though the Top 5 in 2013 was faster than 2003, the Top 10 fell just slower in 2013.

All useless info of course but what the hell.

interesting, good effort.

I agree with FGL's point. However, if we have a big enough data base, the influence of tactics might be averaged out (?)

 

[Ferminal]

I don't know why you are looking at one climb

 

[pmcg76]

Certainly not useless but you are disregarding race tactics. Look at what a cleanish Zulle was able to do in 1999 versus his former Epo fuelled performances. Three minutes slower? Armstrong didnt have to go faster but I bet he could have if neccesary.

Thats why you should allways ask 'who is doing what where and why'.

I don't disagree about race tactics but then why is it when someone equals a time of a doper on a climb, it is always labelled as proof of doping in the clinic. I have seen you personally do it in the past so that must be ok. Everyone goes crazy if someone equals or gets near a time of Armstrong etc whilst completely ignoring tactics etc.

Also taking a 3 year spread lessens the effects of tactics, wind etc. Yes, there is a possibility of all 3 years being affected, but a lot less likely than comparing year on year which is again done frequently in the Clinic.

As to Ferminals point about only one climb, yes that is true but I chose Alpe d'Huez because it is pretty much the most frequently used climb in the Tour, with also the most amount of historical data available. 8 times in the last 15 years which is way more than most other finishing climbs making it easier to group together.

Alpe d'Huez is also less influenced by the wind than a climb like Mont Ventoux which is very exposed. Having stood sheltering from the wind on top of Ventoux in the middle of summer, it is easy to understand how wind really has an impact there.

At the end of the day, I just presented data and facts. People can make what they want from it.