National Series Points

Boardcore

Hi, I've a few questions on the NS points system,

Who is responsible for updating points? TI or Chip Timing company?
Why the delay in posting results? (6 NS races awaiting scoring)
How come Lanesborough can have the system updated in less than 24 hrs?
Is the 30% calculated from full TI members, or Inclusive of one day race licence holders?

Apologies if this has been covered before.

Boardcore

mirrormatrix

My understanding is that TI are responsible for updating points.

There is a fair amount of work in calculating points from the times and, as I understand it, a lot of this is done by hand, hence the delays. I can't explain why Lanesborough can have it updated so quickly.

30 % is calculated from full TI members only.

Is it that big a deal if the points take 5 or 6 weeks to go up? Doesn't affect the end result.

Boardcore

Thanks Mirrormatrix,

If it is TI, you would assume MS Excel does the calculations for them, after clashing the TI member list against the results. 5-6 weeks of a wait however may affect the end result, as you may or may not have to race again for points if chasing individual or team places.

Also, as a bit of a stats head, I like to see the disparity in results from different races - look at Tri-Athy vs HOTW.

tunney

Boardcore wrote: »

Thanks Mirrormatrix,

If it is TI, you would assume MS Excel does the calculations for them, after clashing the TI member list against the results. 5-6 weeks of a wait however may affect the end result, as you may or may not have to race again for points if chasing individual or team places.

Also, as a bit of a stats head, I like to see the disparity in results from different races - look at Tri-Athy vs HOTW.

Okay I am not, at present, a big TI fan.

However I was race secratary about 7/8 years. Much less races, much less participants. I was responsible for doing the scoring.

My undergrad and postgrad were in computer Computer Science, at that stage I also had 5 years or so commerical experience as a developer.

Excel won't cut it on this one.

I wrote custom software to calculate scores and points. It still took me time even when the process was largely automated.

Its not that simple to do and lets be honest. Its an AG competition, TI is an elite focussed organisation (international competitors), thats what its meant to be.

Cut them some slack, be patient, and if you really are such an Excel wizard and stats head then I am sure they would love for you to volunteer your time and expertise.

BTH

From seeing Scott fielding complaints on twitter and FB over the last couple of years it seems that the problem is not actually with TI but with the race organiser/timing company. TI need finalised race results with corresponding TI reg numbers for each entrant. Results posted on the TI website are always only provisional results. Once the race is over, most RDs are going to be more concerned in catching up on real life and have had so much race related headaches that they probably don't want to think about it anymore!!

Boardcore

No worries, I'm at the regional meeting in Sept, I'll do a comparative study between what you've done and what excel can output if it speeds up the process.

AGs deserve a level of service also from TI, as it's the 9000+ members that are facilitating an international focused organisation for the elites! that's why I would like to see at max a weekly point turnaround.

EC1000

BTH wrote: »

Once the race is over, most RDs are going to be more concerned in catching up on real life and have had so much race related headaches that they probably don't want to think about it anymore!!

This. Consolidating results into the format required is not an insignificant task considering that there are a number of sources of data that must be compiled - scanned barcodes, timing company, etc. It takes time and there is SOOOOO much else to do also.

This combined with the fact that most of the committee will not have been off the hamster wheel for a few weeks doing prep.

tunney

Boardcore wrote: »

No worries, I'm at the regional meeting in Sept, I'll do a comparative study between what you've done and what excel can output if it speeds up the process.

I am long gone and suspect things have changed since.

tunney

As the lads have said, the data from the RDs and timing companies is junk hence the complexity

(ironically I now work in data management and data cleansing systems)

Boardcore

I'll ask at the meeting if TI could request all timing companies to provide the same data as per some pre-defined template. All these systems are configurable, so it might be an idea to have a NS standard in place.

peter kern

It might be better if you give Ti some time to finish the overhaul of the nat series to give it a meaning in the first place.
the truth is there is 9000 plus members and some 400 participate in the full series.

tunney

Boardcore wrote: »

I'll ask at the meeting if TI could request all timing companies to provide the same data as per some pre-defined template. All these systems are configurable, so it might be an idea to have a NS standard in place.

You do that.

EC1000

Boardcore wrote: »

I'll ask at the meeting if TI could request all timing companies to provide the same data as per some pre-defined template. All these systems are configurable, so it might be an idea to have a NS standard in place.

The template has been in place for some time now. It is not lack of a template that is the problem

Kurt Godel

I did some reverse engineering, that might help explain the NS points formula. Lets use the results from Hook or By Crook as an example.

404 people are listed in the results (both from the timing company, and on the final TI results). I don't have the data, but can work out that a figure of 377 finishers was used, so I presume 27 athletes were ODL or relay or non-TI, and weren't used to determine the NS points. So for NS points purposes there are 377 finishers (needs clarification).

The "base" points figure is 100, which is based on the 30th percentile. Why 30th? Well thats arbitrary, but it might represent an "average" club athlete, so its a decent starting point. But there's nothing magic about it, it might well be 50% or 25% or whatever. But its 30%.

To get the 30th percentile athlete (377/100)*30=113.1, so we round down to see who is in 113th place (and this is 113th position from the full 404 finishers, so someone in the know might clarify this point). David G was 113th so he gets assigned 100 points. His finish time was 1:28:25 (5305 seconds), call this TIME100. To work out anyone else's points, you use their finishing time in seconds (TIMEX), and use this formula:

NS Points = 100 * (TIME100/TIMEX)

For example, Trevor W came first in a time of 1:11:15=4275seconds, so his NS points are as follows:

NS Points = 100 * (5305/4275) = 124.09

Helen F came 152nd in a time of 1:31:40= 5500seconds, so her NS points are as follows:

NS Points = 100 * (5305/5500) = 96.454545.. (rounded up to 96.455 on TI site)

It should be very easy for Excel to strip out non TI members, and automate a formula based on the above. Although I'm a little unclear as to why non-TI members are stripped out to determine the total number of finishers, yet included when deciding who came in the 30th percentile position?

*Feel free to point out an flaws where they exist*

tunney

Kurt Godel wrote: »

I did some reverse engineering, that might help explain the NS points formula. Lets use the results from Hook or By Crook as an example.

404 people are listed in the results (both from the timing company, and on the final TI results). I don't have the data, but can work out that a figure of 377 finishers was used, so I presume 27 athletes were ODL or relay or non-TI, and weren't used to determine the NS points. So for NS points purposes there are 377 finishers (needs clarification).

The "base" points figure is 100, which is based on the 30th percentile. Why 30th? Well thats arbitrary, but it might represent an "average" club athlete, so its a decent starting point. But there's nothing magic about it, it might well be 50% or 25% or whatever. But its 30%.

To get the 30th percentile athlete (377/100)*30=113.1, so we round down to see who is in 113th place (and this is 113th position from the full 404 finishers, so someone in the know might clarify this point). David G was 113th so he gets assigned 100 points. His finish time was 1:28:25 (5305 seconds), call this TIME100. To work out anyone else's points, you use their finishing time in seconds (TIMEX), and use this formula:

NS Points = 100 * (TIME100/TIMEX)

For example, Trevor W came first in a time of 1:11:15=4275seconds, so his NS points are as follows:

NS Points = 100 * (5305/4275) = 124.09

Helen F came 152nd in a time of 1:31:40= 5500seconds, so her NS points are as follows:

NS Points = 100 * (5305/5500) = 96.454545.. (rounded up to 96.455 on TI site)

It should be very easy for Excel to strip out non TI members, and automate a formula based on the above. Although I'm a little unclear as to why non-TI members are stripped out to determine the total number of finishers, yet included when deciding who came in the 30th percentile position?

*Feel free to point out an flaws where they exist*

TIOO76O98
versus
TIO076O98

Podge83

Kurt Godel wrote: »

I did some reverse engineering, that might help explain the NS points formula. Lets use the results from Hook or By Crook as an example.

404 people are listed in the results (both from the timing company, and on the final TI results). I don't have the data, but can work out that a figure of 377 finishers was used, so I presume 27 athletes were ODL or relay or non-TI, and weren't used to determine the NS points. So for NS points purposes there are 377 finishers (needs clarification).

The "base" points figure is 100, which is based on the 30th percentile. Why 30th? Well thats arbitrary, but it might represent an "average" club athlete, so its a decent starting point. But there's nothing magic about it, it might well be 50% or 25% or whatever. But its 30%.

To get the 30th percentile athlete (377/100)*30=113.1, so we round down to see who is in 113th place (and this is 113th position from the full 404 finishers, so someone in the know might clarify this point). David G was 113th so he gets assigned 100 points. His finish time was 1:28:25 (5305 seconds), call this TIME100. To work out anyone else's points, you use their finishing time in seconds (TIMEX), and use this formula:

NS Points = 100 * (TIME100/TIMEX)

For example, Trevor W came first in a time of 1:11:15=4275seconds, so his NS points are as follows:

NS Points = 100 * (5305/4275) = 124.09

Helen F came 152nd in a time of 1:31:40= 5500seconds, so her NS points are as follows:

NS Points = 100 * (5305/5500) = 96.454545.. (rounded up to 96.455 on TI site)

It should be very easy for Excel to strip out non TI members, and automate a formula based on the above. Although I'm a little unclear as to why non-TI members are stripped out to determine the total number of finishers, yet included when deciding who came in the 30th percentile position?

*Feel free to point out an flaws where they exist*

Nice going---- now I know why you use the handle "Kurt Godel" ;-]

Kurt Godel

Podge83 wrote: »

Nice going---- now I know why you use the handle "Kurt Godel" ;-]

If only I could exist in that esoteric world... tunney is right, the flaw is human error. Must starve myself and avoid bugs;)

NorthernRaider

It should be very easy for Excel to strip out non TI members, and automate a formula based on the above. Although I'm a little unclear as to why non-TI members are stripped out to determine the total number of finishers, yet included when deciding who came in the 30th percentile position?

Well explained, Kurt, but I don't think that non-TI members are included in deciding who came in the 30th percentile (after all, they're not regarded as finishers, so it follows that they also do not determine percentiles). They system is far from perfect, but broadly gets it right!

BTH

Hey KG, could have saved you half an hour.

https://www.triathlonireland-newserver.com/national-series-points.html

https://www.triathlonireland-newserver.com/national-series-points.html?race=624&season=2014

Excel or no, the problem as EC1000 has said is combining the timing chip co results.with the ti card scans. And having the time to do so.

Kurt Godel

I had a look at a couple of other races, which has me scratching my head. For instance, take the race results from Two Provinces Tri. 534 listed. Take out the DQ, the DNF's, and the relay's, and you're left with 503. To find the 30th percentile (503/100)*30=150.9. I'd argue this should be rounded up, but its rounded down to 150, and sure enough finisher 150 is assigned 100 points.

Everyone else's points are based off this 150th finishers time (4365 seconds). For example, the winner finished in 3486 seconds, and so his points are:

NS points = 100 * (4365/3486) = 125.215, exactly what he was given.

There appears to be an inconsistency between how the 30th percentile was calculated for my first example as for my second (the flaw may be mine). However, if I'm right, then the NS points for Hook or by Crook are lower than they should be, given that the 30th percentile finisher should be lower than 113th place, so everyones points should be a bit higher.

*Caveat* I may well be wrong, I don't have full data in front of me. If I'm wrong I'll hold my hands up. But it merits further analysis.

Kurt Godel

BTH wrote: »

Hey KG, could have saved you half an hour.

https://www.triathlonireland-newserver.com/national-series-points.html

https://www.triathlonireland-newserver.com/national-series-points.html?race=624&season=2014

Excel or no, the problem as EC1000 has said is combining the timing chip co results.with the ti card scans. And having the time to do so.

I don't get your point BTH? I'm querying the mathematics used (in fact querying the consistency of how its used). Its a largely volunteer driven organisation, I'm not for a second saying results should be up quicker because volunteers need to make more time for us, or anything like that.

You're good with numbers- if you can see a flaw in what I've posted please point it out!:)

Kurt Godel

NorthernRaider wrote: »

Well explained, Kurt, but I don't think that non-TI members are included in deciding who came in the 30th percentile (after all, they're not regarded as finishers, so it follows that they also do not determine percentiles). They system is far from perfect, but broadly gets it right!

I think I've got it backwards as stated- but in any case see my post above with different example which appears to calculate things differently.

Boardcore

NS events require your TI card to be scanned, this should remove the typos that Tunney refers too. The official start list comes from an online application linked to your login. It shouldn't be that difficult to align it all, if those "templates" are used correctly. This should rule out a lot of the time consuming manual manipulation and spurious Timing Company entries.

mossym

it's obviously more important to some than me, but i have to wonder is a few days delay in putting up NS series results really what TI should be focusing their time on? Loads of other issues to be solved

Kurt Godel

I've been using reverse engineering to work out how points are formulated. The data is from what is listed on the TI site. I've explained above how I'm working out figures (how 30th percentile is worked out; how points are then worked from this).

I've looked at King of the Hill, Two Provinces, Tri Athlone, Hell of the West; and the figures don't add up.

KOTH: 290 non-relay, nonDQ (call them "proper") finishers: 30th percentile finisher based on 237 finishers.

Two Prov: 503 proper finishers: 30th percentile based on 500 finishers.

Tri Athlone: 651 proper finishers: 30th based on 526 finishers.

HOTW: 880 proper finishers: 30th based on 887 finishers.

I gave up after that. Any slight discrepancy in proper finishers and "based on" finishers can be accounted for by rounding error, but its clear to me that NS points are being calculated differently for different races.

EC1000

Kurt Godel wrote: »

I've been using reverse engineering to work out how points are formulated. The data is from what is listed on the TI site. I've explained above how I'm working out figures (how 30th percentile is worked out; how points are then worked from this).

I've looked at King of the Hill, Two Provinces, Tri Athlone, Hell of the West; and the figures don't add up.

KOTH: 290 non-relay, nonDQ (call them "proper") finishers: 30th percentile finisher based on 237 finishers.

Two Prov: 503 proper finishers: 30th percentile based on 500 finishers.

Tri Athlone: 651 proper finishers: 30th based on 526 finishers.

HOTW: 880 proper finishers: 30th based on 887 finishers.

I gave up after that. Any slight discrepancy in proper finishers and "based on" finishers can be accounted for by rounding error, but its clear to me that NS points are being calculated differently for different races.

Have you taken out ODLs? If so, how?

EC1000

Boardcore wrote: »

NS events require your TI card to be scanned, this should remove the typos that Tunney refers too. The official start list comes from an online application linked to your login. It shouldn't be that difficult to align it all, if those "templates" are used correctly. This should rule out a lot of the time consuming manual manipulation and spurious Timing Company entries.

If only. A large % of cards dont scan properly. To avoid long registration queues on race morning, these are normally entered manually, often with typos. Start lists are no better, when transfers, duplicates, etc are dealt with. Race morning is a pressure situation. One messed up registration process can ruin the good reputation of a race. You would be surprised how much manual intervention is involved. I just spent the evening working on it finalising a set of NS results.... it takes time

Kurt Godel

EC1000 wrote: »

Have you taken out ODLs? If so, how?

I'm working from the TI results lists, so I can't see who is ODL or not. But those results appear to contain ODL finishers too. And the problem is that sometimes the 30th percentile finisher (from whom everyone's points are based off) is calculated by stripping out the ODL's, but then applied to the full set of results. Other times the 30th percentile finisher is calculated from the full results, and applied to the full set.

I study maths&statistics which is why I'm banging on about this. I totally get the points that TI have more important things to be doing, and also that RD's have to put an awful lot of their free time into satiating NS midpackers like myself. So if anyone knows Excel and wants a hand on the maths side of things, PM me and maybe there's a way to make it easier for everyone that we can show to TI.

EC1000

Kurt Godel wrote: »

And the problem is that sometimes the 30th percentile finisher (from whom everyone's points are based off) is calculated by stripping out the ODL's, but then applied to the full set of results. Other times the 30th percentile finisher is calculated from the full results, and applied to the full set.

Just wondering how you know this when a standard published set of results would not distinguish between full members and ODL holders?

Kurt Godel

EC1000 wrote: »

Just wondering how you know this when a standard published set of results would not distinguish between full members and ODL holders?

Go to the list of NS points for each race, note who is assigned 100 points.

Then find his/her name in the results, and note their position.

Say their position was 71st. You can work out the number of entrants X as 71 is the 30th percentile: (X/100)*30=71, so X=236.667, suggesting their were 237 TI-regged finishers. In the set of results I worked this from, there were actually 290 non-relay finishers, so I'm presuming* 290-237=53 ODL holders.

*big assumption

tunney

Kurt Godel wrote: »

Go to the list of NS points for each race, note who is assigned 100 points.

Then find his/her name in the results, and note their position.

Say their position was 71st. You can work out the number of entrants X as 71 is the 30th percentile: (X/100)*30=71, so X=236.667, suggesting their were 237 TI-regged finishers. In the set of results I worked this from, there were actually 290 non-relay finishers, so I'm presuming* 290-237=53 ODL holders.

*big assumption

I know how you can get waaayyyyyy faster for 2015

Stop fvcking about with the NS stats. If you feel a stats urge coming on, go sit on your turbo