[Event] Cape Town Cycle Tour 2019

[DieselnDust]

My thinking is this:

 

There are riders who put in a better effort this year than last, and riders who did the opposite. On the whole, they will balance each other out. Therefore, any average performance difference (across all riders who took part this year and last) can be attributed to external factors: wind, course changes, etc. The intent of the model (with all its flaws) is to give you a time that you can compare fairly with last year's time. It doesn't matter if you were fitter, faster, more focused, whatever, because that is what the model should show up.

 

I did a sub 4:02 this year . The model tells me that this is equivalent to a 3:44 under 2018 conditions. In 2018, I did 3:38. Therefore I did better last year. That sucks.

 

 

But thats exactly what it isn't illustrating in the case of some of the people I know who finished in the top 750 this year. All show higher NP and Ave W values for similar TSS score and similar Ave HR.i.e. they were fitter this year but their times are slower whilst finishing position is improved.

Yes i think the model works for an average rider riding the time anyone inputs and it will deliver an equivalent 2018 time. 

But each one of us using the model should not look at it as an a predictor of whether they rode a relatively better or worse race in 2019.

[Fat Boab]

My thinking is this:

 

There are riders who put in a better effort this year than last, and riders who did the opposite. On the whole, they will balance each other out. Hypothesis 1.

 

Therefore, any average performance difference (across all riders who took part this year and last) can be attributed to external factors: wind, course changes, etc. Hypothesis 2.

 

The intent of the model (with all its flaws) is to give you a time that you can compare fairly with last year's time. It doesn't matter if you were fitter, faster, more focused, whatever, because that is what the model should show up.

 

I did a sub 4:02 this year . The model tells me that this is equivalent to a 3:44 under 2018 conditions. In 2018, I did 3:38. Therefore I did better last year. That sucks.If

 

If you wanted to go to town, then you'd have to test hypotheses 1 and 2, rather than just assuming them. But who has time for that?!

[Rouxenator]

Cool name that came past me : http://results.racetec.co.za/results_by_event.aspx?RID=2019&EN=Sun%2010%2F03%2F2019%202019%20Cape%20Town%20Cycle%20Tour%20%28109km%29&FN=punkhead 

[MudLark]

Cool name that came past me : http://results.racetec.co.za/results_by_event.aspx?RID=2019&EN=Sun%2010%2F03%2F2019%202019%20Cape%20Town%20Cycle%20Tour%20%28109km%29&FN=punkhead 

 What? "Sorry no results"?

[HelloRyanFowler]

But thats exactly what it isn't illustrating in the case of some of the people I know who finished in the top 750 this year. All show higher NP and Ave W values for similar TSS score and similar Ave HR.i.e. they were fitter this year but their times are slower whilst finishing position is improved.

Yes i think the model works for an average rider riding the time anyone inputs and it will deliver an equivalent 2018 time. 

But each one of us using the model should not look at it as an a predictor of whether they rode a relatively better or worse race in 2019.

 

The intention of the model is very much to be a predictor of whether or not someone rode a better ride this year than last. If I've done a reasonable job with the maths, it should be a reasonable indicator - at least for most finishers. But the model is based on only three variables, and requires only one input to get a result, so there are bound to be inaccuracies somewhere along the line. And if you've got a handle on your NP, Ave W, TSS score and other such things that I had to Google, then you've probably already got a much more reliable indicator than a generic model will ever give you.

 

This might need to become an annual exercise; possibly throwing that huge and beautiful data set at an artificial neural network capable of identifying and quantifying the relevant variables. Of course, even a perfect model is not going to satisfy everyone because we probably place a disproportionately high value on our own impression of our performance. That's been the hard part for me.

 

* But if I factor in 7 minutes of stoppages to deal with a broken spoke, a buckled rim, and touching brakes (which had to be released for the second half of the race), then maybe I can consider my performance on par with last year!

Bt thats exactly what it isn't illustrating in the case of some of the people I know who finished in the top 750 this year. All show higher NP and Ave W values for similar TSS score and similar Ave HR.i.e. they were fitter this year but their times are slower whilst finishing position is improved.

Yes i think the model works for an average rider riding the time anyone inputs and it will deliver an equivalent 2018 time. 

But each one of us using the model should not look at it as an a predictor of whether they rode a relatively better or worse race in 2019.

 

 

[Rouxenator]

 What? "Sorry no results"?

 

[Zub]

Seeding? The west coast express and tour de ppa. They are best to up seeding I have found.

 

 

Sent from my iPhone using Tapatalk

What about 1 Tonner 170km? Beta of 2, so all you have to do is finish within reasonable time.

[LeoKnight]

But thats exactly what it isn't illustrating in the case of some of the people I know who finished in the top 750 this year. All show higher NP and Ave W values for similar TSS score and similar Ave HR.i.e. they were fitter this year but their times are slower whilst finishing position is improved.

Yes i think the model works for an average rider riding the time anyone inputs and it will deliver an equivalent 2018 time. 

But each one of us using the model should not look at it as an a predictor of whether they rode a relatively better or worse race in 2019.

 

is it not maybe just a factor of the 80 / 20 rule.. some kind of indicator is sometimes better than none to compare how you performed the day, given there are lots of factors that could have played a role (some within and some outside of you control).  

[DieselnDust]

is it not maybe just a factor of the 80 / 20 rule.. some kind of indicator is sometimes better than none to compare how you performed the day, given there are lots of factors that could have played a role (some within and some outside of you control).  

 

 

YEs it is fine for that purpose and a good model. Obviously refinement will only make it better. I think we owe Ryna a dobt of gratitude regardless of how accurate it compares performance across the 2018 and 2019 CTCT's. I think we all better understand how individual performance was affected.

[Mlayne]

What about 1 Tonner 170km? Beta of 2, so all you have to do is finish within reasonable time.

one tonner only had index of 2 last year due to heat 35+ Celsius. Normally lower than that and less than 1.2 previously. I think Stellenbosch cycle Tour is also good for seeding

[daveno7]

The intention of the model is very much to be a predictor of whether or not someone rode a better ride this year than last. If I've done a reasonable job with the maths, it should be a reasonable indicator - at least for most finishers. But the model is based on only three variables, and requires only one input to get a result, so there are bound to be inaccuracies somewhere along the line. And if you've got a handle on your NP, Ave W, TSS score and other such things that I had to Google, then you've probably already got a much more reliable indicator than a generic model will ever give you.

 

This might need to become an annual exercise; possibly throwing that huge and beautiful data set at an artificial neural network capable of identifying and quantifying the relevant variables. Of course, even a perfect model is not going to satisfy everyone because we probably place a disproportionately high value on our own impression of our performance. That's been the hard part for me.

 

* But if I factor in 7 minutes of stoppages to deal with a broken spoke, a buckled rim, and touching brakes (which had to be released for the second half of the race), then maybe I can consider my performance on par with last year 

 

even though we ride bikes...we still get older...we lose the top end of our VO2 max...we gather weight...we save our legs for chapmans to become a hero...but we still get older...that wind...really wasn't a great issue...we were just riding a bit more smart...

[Man with no name]

But thats exactly what it isn't illustrating in the case of some of the people I know who finished in the top 750 this year. All show higher NP and Ave W values for similar TSS score and similar Ave HR.i.e. they were fitter this year but their times are slower whilst finishing position is improved.

Yes i think the model works for an average rider riding the time anyone inputs and it will deliver an equivalent 2018 time.

But each one of us using the model should not look at it as an a predictor of whether they rode a relatively better or worse race in 2019.

My tss was down 30 and my np down 35. My time was 13 minutes slower BUT my position improved from ~650 to 335.

 

I ascribe both tss and np decrease due to the bunch going generally slower in the first half and the lower wind drag on n the bunch despite the headwind.

 

 

Sent from my iPad using Tapatalk

[andydude]

Seeding? The west coast express and tour de ppa. They are best to up seeding I have found.

 

 

Sent from my iPhone using Tapatalk

Yeah that 2018 west coast's seeding is completely out of whack! But i am not complaining. Will keep my a for years without doing any rides [emoji4]

 

Sent from my SM-G950F using Tapatalk