Seeding: Who determines BETA?

[daniemare]

Following my PB at the CTCT, my PPA seeding shot up to my best ever E.  But what was interesting was that the Beta for the CTCT (1.07) made it theoretically more difficult than the Tour de PAA (1.05)

 

Now every year I do these 2 races, which are 2 weeks apart. And every year I am +-30 minutes quicker in the CTCT than the Tour de PPA.  Mostly same fitness, weight etc and from my wheather notes, also coincidently, very similar for the years I did both

 

Now for me, the Tour de PPA is vastly more difficult than the CTCT. But from the winning times being simmilar, that appears not the case for the sharp end of the field. And that got me thinking, who determines BETA and with what rider in mind?

 

For me:

- The CTCT last 15km is a coast after Suikerbossie, whilst the Tour de PPA finishes with Adderly Street's mind numbing kilometers of uphill drag. The last 15-20% of the race is normally where the weekend warrior falls flat, thus an easier finish is always more welcome

- The shear numbers of the CTCT means I can always find the next group when dropped, whilst the Tour de PPA often leaves you on your lonesome or with 2 or 3 other stragglers as you make your way to the finish.

 

So for me it look very similar to golf's course ratings, that predominately considers the Pro's abilities, forgetting that crossing water 15 times a round is daunting for the hacker, even if the course is short.

 

Long and short, I think the CTCT BETA is way overestimated, and I know I will be punished and very quickly on my own if I ride the next PPA fun ride from E.

 

[Skubarra]

I think the BETA is a random number generated by the racetec computer - that is the only answer that would make sense of the betas spewed out by the seeding computer.

 

Given that the Cycle Tour is the base event for the BETA it makes no sense that this year with perfect conditions would have a beta of 1.07

 

That said one also has to look at the beta together with the adjusted winners time and not on its own then sometimes the numbers make slightly more sense.

 

In reality though I have long given up on trying to understand this and just accepted the numbers are random 

[Jehosefat]

Beta is determined using a statistical method (linear regression) where every competitor that has done both that race as well as the "base" race (usually 94.7 or CTCT) has their relative time between the two races compared.

 

Relative time in this case being their time/winner's time.

 

The beta of 1.07 means that, on average, people were 7% slower relative to the winner than they were in the base race.

 

As far as I know there is no subjectivity in the calculation of the beta. Any subjectivity is applied to the "adjusted winners time" and they don't disclose anything about how they come to that adjustment.

 

IF you want more detail on how the regression works, let me know.

[Skubarra]

 

The beta of 1.07 means that, on average, people were 7% slower relative to the winner than they were in the base race.

 

This does not make sense to me, all else being equal people finish closer to the winner when conditions are good.

 

Just an example of the inconsistency

 

2014 Cycle Tour winning time was 2:39:30 and there were less than 500 sub-3's and had a beta of 1

 

this year winning time is 2 minutes faster but there is more than 1000 sub-3's and the beta is 1.07

 

Given that conditions & times were generally better this year I would not expect a higher beta

 

So whatever they use as "average" is garbage

[Michieldt]

• ((((Time / Winner) – 1) / Beta) x 100) + Penalty

[Time] is your time for the event,

[Winner] is the time of the winner (sometimes adjusted)

[beta] is a number that indicates the difficulty of the conditions where more than 1 means windy or hilly and less means flat or calm

[Penalty] is a number that gets added to the index usually for the time that has passed since the event (more than 5 months old)

From PPA website. Seems beta is subject to conditions.

 

[Quagga]

Following my PB at the CTCT, my PPA seeding shot up to my best ever E.  But what was interesting was that the Beta for the CTCT (1.07) made it theoretically more difficult than the Tour de PAA (1.05)

 

Now every year I do these 2 races, which are 2 weeks apart. And every year I am +-30 minutes quicker in the CTCT than the Tour de PPA.  Mostly same fitness, weight etc and from my wheather notes, also coincidently, very similar for the years I did both

 

Now for me, the Tour de PPA is vastly more difficult than the CTCT. But from the winning times being simmilar, that appears not the case for the sharp end of the field. And that got me thinking, who determines BETA and with what rider in mind?

 

For me:

- The CTCT last 15km is a coast after Suikerbossie, whilst the Tour de PPA finishes with Adderly Street's mind numbing kilometers of uphill drag. The last 15-20% of the race is normally where the weekend warrior falls flat, thus an easier finish is always more welcome

- The shear numbers of the CTCT means I can always find the next group when dropped, whilst the Tour de PPA often leaves you on your lonesome or with 2 or 3 other stragglers as you make your way to the finish.

 

So for me it look very similar to golf's course ratings, that predominately considers the Pro's abilities, forgetting that crossing water 15 times a round is daunting for the hacker, even if the course is short.

 

Long and short, I think the CTCT BETA is way overestimated, and I know I will be punished and very quickly on my own if I ride the next PPA fun ride from E.

Where did you get this info, my CTCT does not show on PPA website

[Shebeen]

http://www.pedalpower.org.za/how-seeding-is-calculated/

 

 

Some races have a better quality of field than others. For instance, some events may have visiting pro riders participate and some events (especially shorter routes) have no top riders present. So we adjust the winner’s time to take this into account.

It would also be unfair to get the same seeding for being the same percentage behind the winner in a fast, flat ride as in a hilly, difficult ride.  To allow for this we calculate a difficulty or “beta” factor. This makes the gap between the winner and you count less in difficult conditions and more in easy conditions.

First, an extract is done of all the riders who did the event in question, as well as the base event. Every rider with an index better than 100 is taken for this calculation. The assumption is made that the same riders should have the same index for both events, so the winner’s time of the funride is now adjusted and the “beta” is calculated to achieve this.

In statistical terms, a linear regression is performed for the event relative to the indexes of the people in the event who also rode one of the base events. This determines how much the winner’s time should be adjusted and what the difficulty factor “beta” should be. There is no subjectivity in this process – it is an automated calculation without human intervention.

In layman’s terms, the adjusted winner’s time should be roughly the time that the winner of an event would have done if the winner of the Cycle Tour had ridden that event at the same effort. The beta factor will be 1 if the event is as difficult as (i.e. on par with) the Cycle Tour, less than 1 if it is easier (eg flat and fast), and more than 1 if it is harder (eg hilly or very windy).

[Jehosefat]

This does not make sense to me, all else being equal people finish closer to the winner when conditions are good.

 

Just an example of the inconsistency

 

2014 Cycle Tour winning time was 2:39:30 and there were less than 500 sub-3's and had a beta of 1

 

this year winning time is 2 minutes faster but there is more than 1000 sub-3's and the beta is 1.07

 

Given that conditions & times were generally better this year I would not expect a higher beta

 

So whatever they use as "average" is garbage

 

But the 1000 sub-3's are only 3-4% of the field... So you can't just look at that to determine whether the 1.07 is correct. It looks at everyone who finishes and has also done whichever race they use as the base race.

 

Also, they might have changed the methodology since 2014 which could explain the difference (I only started cycling at the end of 2016 so I don't know)

[lechatnoir]

The 109km One Tonner had a beta of 0.94 with the last +30km (felt longer) into a 35kmh (felt faster) headwind... Winners time 20min slower  than CTCT, with admittedly a lesser field, but still... this year's CTCT >10% harder than One Tonner? I don't think so

[Skubarra]

But the 1000 sub-3's are only 3-4% of the field... So you can't just look at that to determine whether the 1.07 is correct. It looks at everyone who finishes and has also done whichever race they use as the base race.

 

Also, they might have changed the methodology since 2014 which could explain the difference (I only started cycling at the end of 2016 so I don't know)

 

I get that it is an "automated non-subjective" process - still doesn't mean the results spewed out make sense.

 

Ok lets use middle of the field for 2014 vs 2018 CTCT

 

Time is 4:40 vs 4:28, again the event with the faster times got the higher beta.

 

Point I want to make to the OP is that even though in theory more difficult events should get higher betas and vica versa whatever variables they throw into their calcs often means that this is not the case.

 

For the REALLY difficult and REALLY easy races the betas usually make sense but for a whole lot of races inbetween the betas are all over the place.

[lechatnoir]

For the REALLY difficult and REALLY easy races the betas usually make sense but for a whole lot of races inbetween the betas are all over the place.

 

So.... a joke then?

[Pure Savage]

Hmmm, I think the algorithm that adjust the winning time and beta is rather accurate. 

 

Below shows my races against a similar field with me at a similar fitness, fatigue in TOGH kicked in stage 5. 

 

It makes sense, I struggle with the more hilly races, but on the flatter roller ones I do better. 

 

Only event missing was my tour de ppa result which is also a 4/5 index.

 

So whatever the thing does, the seedings are not far off.

[Skubarra]

Hmmm, I think the algorithm that adjust the winning time and beta is rather accurate. 

 

Below shows my races against a similar field with me at a similar fitness, fatigue in TOGH kicked in stage 5. 

 

It makes sense, I struggle with the more hilly races, but on the flatter roller ones I do better. 

 

Only event missing was my tour de ppa result which is also a 4/5 index.

 

So whatever the thing does, the seedings are not far off.

 

Dude, if you are in A bunch finishing a few seconds behind the winners then the betas make no difference to your seeding 

 

The lower you go down the field the impact of the beta can be massive

[Jehosefat]

Dude, if you are in A bunch finishing a few seconds behind the winners then the betas make no difference to your seeding 

 

The lower you go down the field the impact of the beta can be massive

 

Sure, but a higher beta improves your seeding so I'd rather complain when the beta is too low rather than too high...

[Skubarra]

Sure, but a higher beta improves your seeding so I'd rather complain when the beta is too low rather than too high...

 

I think I can list about 7 races I did in the last year where I finished within seconds of the front of my group but a low beta meant my seeding stayed the same - got a bit frustrated with the whole seeding system as might be obvious from this thread.

 

So yes, not complaining after the CTCT but nobody will convince me this system is not a bit broken.

[Pure Savage]

I think I can list about 7 races I did in the last year where I finished within seconds of the front of my group but a low beta meant my seeding stayed the same - got a bit frustrated with the whole seeding system as might be obvious from this thread.

 

So yes, not complaining after the CTCT but nobody will convince me this system is not a bit broken.

ha ha, next Sportif I will drop back and we can mess with the system