C1 Fracture Posted February 21, 2017 Posted February 21, 2017 Hello hubbers So I started entering road events and EVERYONE said, "Join the PPA and get better seeding..."So now that I'm in it, I'd like to understand the rules that governs said seeding. In my previous post, we have established the Group Eligibility categories, see hyperlink. Beta values... How is it possible, that for the same race, over the same distance, one year apart the difference in beta = 0.52? Bouckaert-Soenen 2016 had a beta = 1.43 and in 2017 beta = 0.91? 99er Cycle Tour 2016 beta = 1.30 and 2017 beta = 1.05? I would assume the "algorithm" accounts for the followingevent minimum distance = 100 km, elevation gained = 1000m, beta = 1 if the race is windy; i.e. head wind (>50% of the route distance) = 5m/s, add 0.05 for every 1m/s of average wind speed if race has a tail wind for more than 50% of the route distance, subtract a 0.05. If the event starts with a big climb, like Stellenbosch Cycle Tour account for it If it has rolling hills with a big climb within the last 10km's of the race add a decimal.. I hope you get my point... But for the life of me, how is it that the same distance, same route, winner's time within a 1% gets "down-rated" by so much, i.e. lowered beta value... I think, PPA / RaceTec should standardise by distance and elevation gained. A baseline beta, that will not reduce a 124 km race to less than 1. Happy commenting
Skubarra Posted February 21, 2017 Posted February 21, 2017 Hi C1 For one you also have to look at the adjusted winners time and how that differs from year to year relative to the actual winner. The combination of adjusted winner & beta is what you should compare from year to year The quality of the field that year also plays a part not only the weather conditions. That said, as others have pointed out it is a bit of a mystery how they come up with the betas & adjusted winner times. This has been discussed a lot on thehub and nobody really knows. I'm certain there is some thumbsucking involved in determining some of the factors as opposed to actual maths, as results are definitely not always consistent. I have also picked up some clear mistakes in past seeding calcs In my view, at the end of the day the PPA seedings are a bit of a hit & miss. If you do 1 or 2 races a year the odds are small that your seeding will reflect your actual ability. If you do 5 or 6 races a year the chances are good that PPA would have stumbled on at least 1 result that reflects (or exceeds) your actual ability. Feel free to indulge yourself in analysis paralysis, but if you want to get the seeding that reflect your ability just do more events.
Long Wheel Base Posted February 21, 2017 Posted February 21, 2017 You lost me at the algorithm in your heading.Just ride lots and it will work itself out.You have a "beta" chance of understanding woman.
Pure Savage Posted February 21, 2017 Posted February 21, 2017 Hello hubbers So I started entering road events and EVERYONE said, "Join the PPA and get better seeding..."So now that I'm in it, I'd like to understand the rules that governs said seeding. In my previous post, we have established the Group Eligibility categories, see hyperlink. Beta values... How is it possible, that for the same race, over the same distance, one year apart the difference in beta = 0.52? Bouckaert-Soenen 2016 had a beta = 1.43 and in 2017 beta = 0.91? 99er Cycle Tour 2016 beta = 1.30 and 2017 beta = 1.05? I would assume the "algorithm" accounts for the followingevent minimum distance = 100 km, elevation gained = 1000m, beta = 1 if the race is windy; i.e. head wind (>50% of the route distance) = 5m/s, add 0.05 for every 1m/s of average wind speed if race has a tail wind for more than 50% of the route distance, subtract a 0.05. If the event starts with a big climb, like Stellenbosch Cycle Tour account for it If it has rolling hills with a big climb within the last 10km's of the race add a decimal.. I hope you get my point... But for the life of me, how is it that the same distance, same route, winner's time within a 1% gets "down-rated" by so much, i.e. lowered beta value... I think, PPA / RaceTec should standardise by distance and elevation gained. A baseline beta, that will not reduce a 124 km race to less than 1. Happy commenting They use a pretty hectic algorithim to work it out. Our maths Phd in the team says its the bee's knees. Basically by adjusting both winning time and the Beta you get a line of best fit of the days results to the Cycle Tour. So if its really windy, and the groups were slow, this will reflect in the Beta and winning time. Best way to seed all them peeps. Find the best way to improve seeding is to suffer harder.
Skubarra Posted February 21, 2017 Posted February 21, 2017 They use a pretty hectic algorithim to work it out. Our maths Phd in the team says its the bee's knees. The model/algorithm might be awesome, but then there is still the issue of garbage in, garbage out I have seen way to many odd seeding results to have faith in that system - still think it is better than the one used by SASeeding though.
carbon29er Posted February 21, 2017 Posted February 21, 2017 As Savage said. And others have said often. Cycling is a bugger of a sport when you are a newbie. Start at the back in the worst conditions and hope to improve year on year. In the old days without in season reseeding it would take about 5 years of racing weekly in season to move from the Z newbie numbers to C or D. And it was hard work time trialling every funride with maybe 1 or 2 helpers. By comparison racing in the seeded bunches up front was a quarter of the effort but boy, was that quarter hard. My advice to OP: Forget about the seeding index calculations, it is the fairest system in existence and be grateful that if you are strong you will move into the correct batch quite quickly. And be careful what you wish for. Being an E seeder riding in A is a whole lot more pain than you ever imagined possible on a bike.
carbon29er Posted February 21, 2017 Posted February 21, 2017 The model/algorithm might be awesome, but then there is still the issue of garbage in, garbage out I have seen way to many odd seeding results to have faith in that system - still think it is better than the one used by SASeeding though.What garbage can get in? I'm genuinely interested in why you think this as there is ALWAYS a valid explanation.
Pure Savage Posted February 21, 2017 Posted February 21, 2017 The model/algorithm might be awesome, but then there is still the issue of garbage in, garbage out I have seen way to many odd seeding results to have faith in that system - still think it is better than the one used by SASeeding though. How else would one compare actual results on the day to previous events? Its working pretty well judging by the spread of people across all the start groups at PPA funrides.
Pure Savage Posted February 21, 2017 Posted February 21, 2017 As Savage said. And others have said often. Cycling is a bugger of a sport when you are a newbie. Start at the back in the worst conditions and hope to improve year on year. In the old days without in season reseeding it would take about 5 years of racing weekly in season to move from the Z newbie numbers to C or D. And it was hard work time trialling every funride with maybe 1 or 2 helpers. By comparison racing in the seeded bunches up front was a quarter of the effort but boy, was that quarter hard. My advice to OP: Forget about the seeding index calculations, it is the fairest system in existence and be grateful that if you are strong you will move into the correct batch quite quickly. And be careful what you wish for. Being an E seeder riding in A is a whole lot more pain than you ever imagined possible on a bike. They have made it easier with the hill climbs as well. My only gripe is PPA does not use the DC as a seeding event
Pure Savage Posted February 21, 2017 Posted February 21, 2017 "Beta and adjusted winner's time are the slope and intercept calculated by linear regression, comparing results in that event to the cycle tour or 94.7" P Savage
Skubarra Posted February 21, 2017 Posted February 21, 2017 Its working pretty well judging by the spread of people across all the start groups at PPA funrides. My advice to OP: Forget about the seeding index calculations, it is the fairest system in existence and be grateful that if you are strong you will move into the correct batch quite quickly. Ok, just to be clear - I agree with what you said above, the system works ok, especially if you do a lot of races or in the middle & back groups where the variances & margins for errors are bigger anyway. But I'm not convinced that the way the account for weather, route & field in the beta & adjusted winner time is always consistent given some odd results I have seen, especially for mtb events. And it is more noticeable in the front groups where the margins between groups are small. Examples of garbage in / garbage out - Bouckaert Soenen mtb a few years back where a short distance rider's time was used as the winner time when calculating the seeding for the long distance mtb event. I also recall a few years back a big fat middle aged oke who have only done one short distance mtb race started in $ bunch in the Burger Road race (with his 26er and camelback), due to a seeding mixup on the one race he have done up until that stage. I have seen my seeding bumped up by ridiculously high betas (above 2) on short dirtroadie mtb events that I fail to understand why they are rated as extremely difficult...
Skubarra Posted February 21, 2017 Posted February 21, 2017 "Beta and adjusted winner's time are the slope and intercept calculated by linear regression, comparing results in that event to the cycle tour or 94.7" P Savage That is great, but how do you objectively compare the "difficulty" of race x with the cycle tour or 94.7. How you account for heat & wind & route & road conditions etc in your beta will always have an element of subjectivity as it is not something that can be exactly measured.
Organic Posted February 21, 2017 Posted February 21, 2017 I think this covers it pretty well (see In more detail, point 3): http://www.pedalpower.org.za/how-seeding-is-calculated/ Whether that's correctly applied or not, and whether it's a fair (or optimal) method, is a different matter. But it has the advantage of being simple and non-subjective.
Pure Savage Posted February 21, 2017 Posted February 21, 2017 That is great, but how do you objectively compare the "difficulty" of race x with the cycle tour or 94.7. How you account for heat & wind & route & road conditions etc in your beta will always have an element of subjectivity as it is not something that can be exactly measured. Thats what the linear regression does. The problem is when your sample size is not big enough I think. I do agree that the MTB does not work well for seeding at all as you are trying to perform the linear regression against a road event, where the gaps between first and last are much smaller.
carbon29er Posted February 21, 2017 Posted February 21, 2017 Snip especially for mtb events. And it is more noticeable in the front groups where the margins between groups are small. Examples of garbage in / garbage out - Bouckaert Soenen mtb a few years back where a short distance rider's time was used as the winner time when calculating the seeding for the long distance mtb event. snip Say no more. I cannot imagine how it is statistically valid to use a field that small to extrapolate anything, never mind seeding. I think the algorithm has a bonus for mtb riders to reward them for turning up.
Patchelicious Posted February 21, 2017 Posted February 21, 2017 Do PPA events get preferential betas?
Recommended Posts
Archived
This topic is now archived and is closed to further replies.