A better way to rank skiers

[Horton]

@skispray I am working on a system something like this...

 

1. Each skier has a Basis Value based on their placement the previous year

2. The #1 skier from last year has a value of 10, the #2 skier has a value of 9.99, the #3 skier has a value of 9.98 and so on. ( the values are actually calculated / not linear )

3. The winning skier at an event would accrue the point value of every skier he beat

4. The second place skier at an event would accrue the point value of every skier he beat

5. The following skiers all accrue the point value of every skier they beat

 

With this system backyard events are worthless. The more skiers and higher quality of the skiers at any event the more points the winner can accrue. As long as everyone at a given event gets the same ride and ropes it is far. If the ropes are long or short it does not matter as long as it is the same for everyone. If the driving is tight or loose that is fine as long as it the same for everyone.

 

This system would mean actual competition.

 

You can look at Version Beta 1.0 here https://docs.google.com/spreadsheets/d/1PgX5ilF85QcwMgPceAPipADXa3hQa0ZYucyc7SRz5wk/edit?usp=sharing

[JeffSurdej]

@Horton doc is not viewable unless requested to be shared, just FYI

[Than_Bogan]

If I understand correctly, this favors quantity of events entered more than I would like to.

 

It's a great start, though. Thanks for sharing!

[aupatking]

My daughter ranked Mr Matteo ( @Luzz ) the best skier in the world and daddy 2nd. I’m gonna stick with her ranking system

[MattP]

@Horton is this more of a variation of the BOS Ranking we once did but removes the IWWF scores? The idea with this list was to find a way to combine Backyard & Pro Events equally to find the best skier of the year who competed against the best skiers. It removed money from the equation which the elite list is based on and accounted for the fact that not every skier attends every event. Details on this list are on this thread > https://www.ballofspray.com/forum#/discussion/comment/227196

 

[Than_Bogan]

@aupatking Apologies to @Horton for careening off-topic at full speed, but one of the funniest parts of my waterski career was the 2-3 years that my older daughter scored @MikeT and me whenever she joined us. She was probably around 6-8 y.o.. Her "system" was totally incomprehensible to us, but in the first year that she did it, I always won. We starting calling her The Russian Judge. :)

[Horton]

@Than_Bogan I think the final ranking would only include a skier's 5 or 10 best tournaments based on points earned. The bias is toward out performing the most skiers per event.

 

@MattP I started down a number of paths before I realised how simple this one is. If Fred Winter runs 47 off at a back yard event but no other top skiers are there he gets nothing. If Fred runs 22 off and that is enough to out perform the top 20 skiers in the world he gets a huge haul of points.

 

@JeffSurdej sheet is now public.

[JeffSurdej]

@Horton I'm dying to change our rankings in some fashion as this, the problem is the #'s, most events have 30-40 skiers total let alone 1-2 in their division, you almost have to use handicapping along with this so at least at each local event you have 30-40 competitors? Or is this suggestion just for pros?

[MattP]

@Horton I can get behind this idea. Are you using the dynamic or static list to create the skier value?

[Horton]

@JeffSurdej The idea was hatched around pros. It would mean a pivot to actual competition on the weekend instead of everyone just skiing for scores.

 

I can see problems with this but I am not sure the negatives outweigh the positives.

[Horton]

@MattP this is still a VERY early idea.

 

I used last years static list for the example. I think in the future every year you would use a static version of last year. That way Skier X is always worth the same amount of points when you beat him.

[klindy]

https://en.wikipedia.org/wiki/Elo_rating_system

[AdamCord]

@Horton great idea.

 

I'd argue that the skier point values should change from week to week based on their standing. That way when someone has a breakout year their point value is reflective of their current skiing instead of their skiing last year. Or if someone was injured the previous year and misses a season his value should not be zero for the entire next season.

[MattP]

@AdamCord so lock in the dynamic list for the day the "Pro" event hits the list.

[Than_Bogan]

@Horton NICE. That "top N" idea works.

 

Make it so!

[Horton]

I am in no way claiming this idea is fully cooked but I very much believe in the intended end result. I am not sure how to do with amateurs without making it super complex.

[JeffSurdej]

@Horton its exactly the next step in our sport, time to stop having tourneys for score only and add competition, but without using handicapping I just don't see a way at local events

[bigtex2011]

Do you have the ski surface area tho

[klindy]

@Horton so how is this different from the ELO scoring method I posted? I haven’t studied your spreadsheet so I really don’t know the answer.

[Horton]

@klindy I think it is basically the same.

[skispray]

@JeffSurdej I like the idea of the pros competing with no handicap. For amateurs one idea that comes to mind would be to have two division, a handicapped division and a non-handicapped division. If you sign up for a local tournament you pick which of the two you want to compete in that day, and everyone in the handicap group competes amongst themselves, while everyone in the non-handicapped division competes against themselves. Factors that would be relevant to your selection are how well you’re skiing at that moment relative to your baseline score, who else is at the same tournament, and what division they’re skiing in.

 

You could have both handicap and non-handicap ranking lists.

 

FWIW, I still think there should be boys/girls divisions separate from adult divisions. So to me ideally there would be divisions something like the following:

 

Men’s No Handicap (18 yo+)

Women’s No Handicap (18 yo+)

Adult Handicap (18 yo+ men & women)

 

Then either leave the boys/girls age divisions alone, making them either exclusively handicapped, or having both the handicap/no-handicap options there too.

[Horton]

@skispray We can not fracture events like that. That is way to messy and makes the standings even more murky.

[JeffSurdej]

@Horton we really just need to run it side by side, keep current rankings for qualifications and start up a new one with some handicapped competition formula, this is sort of a big it they will come approach, just do the rankings and LOC will start running events as such, The hard part is getting someone to program a rankings to do so with WSTIMS imports, @MattP you busy, sounds like its right up your alley :)

[skispray]

@Horton fracture events? There are currently like 80 divisions for adults and I proposed there be 3.

 

I’m okay with strictly doing handicapped competition for amateur tournaments, but still keep the kids separate. When people are advancing against their PB’s very quickly it’s not really a competition to pit them against people with scores based on a long history. The “best” skiers at every tournament would be kids beating their PB’s by 3 passes as they reach their max speed.

[JackQ]

I think we need to continue rankings (for non pros) for all the reason stated above. However, I think taking you 3 best scores is not the best method. It incentives a few good scores. Year ago when dinosaurs roamed the was a formula for seeding at Regionals and Nationals which was something akin to 3 times your best score plus 2 times your median plus you average score divided by 6. I rewarded consistency as much as best scores and I think a better analogy for skiers ranking.

[BCM]

@Horton - I think you are on the right path here. I would love to see our rankings system get mixed up a bit, I have always thought something similar to the point system in US junior snow skiing would be applicable (See page 28), likely not a direct translation but might be good for inspiration.

[teammalibu]

@bigtex2011 you take LxW round it up and then eat some Pie!

[cruznski]

How often are the USAWS rankings updated?

[MattP]

@cruznski everyday.

Scores are uploaded to the Rankings website, immediately upon receipt from the tournament scorer. Any newly reported or updated scores will show up immediately in Score Detail displays. However, the Ranking Scores are recalculated only once each day, during the overnight hours. As a result, any changes or additions to the score database on any given day, will not be reflected in the rankings until the following morning.

@JeffSurdej sure lets put that on my plate too.. :)

[klindy]

@MattP As you mentioned, that's one of the challenges with the current ranking list. The scores are "recalculated" meaning each raw score is put into the database based on what was uploaded over the last day. The calculations that take place include applying a one of three different penalty (or not based on the number of scores and class of tournament), the ranking percentile, and determining what level the resulting average would be in. The length of time it takes to get all that done limits the number of scenarios that can be effectively included.

 

For example, for handicapping purposes, it would be far better to include some quantity of raw scores and not apply any kind of penalty to those scores. To apply a penalty is an advantage to those who only have a score or two.

 

Several years ago I requested a feature enhancement to include a few drop down boxes so I can get the data I wanted for see instead of just what's presented currently. As an example, let's say you wanted the top 5 scores without a penalty plus a column that indicated the total number of scores in the average. That kind of flexibility was impossible because of the amount of time it would take to 'recalculate' the database in real time.

 

The potential solution was to create 'a few' (2-5 likely) different handicap scenarios and hard code them into the nightly update sequence. Then you'd just 'click' on the output format you wanted and it would display accordingly.

 

The solution could potentially be used for any kind of ranking methodology. Since the raw score for each event will be the same, the only difference would be how they are ranked. So what @Horton is suggesting, the current ranking list, and some kind of 'raw score' output to be used for handicapping purposes would all coexist in the same database.

[Luzz]

So I got here because I thought someone caught my interview with Candido Moz about the ELO rankings IWWF experimented… and instead I am here because a very wise girl deemed I am 1st in her ranking!! :D

 

More on the subject matter, I fully agree with where @Horton is coming from, and incidentally so does Moz. The issue is how to do it. When I asked him why they didn't continue with this experimental ranking, he didn't provide me with a straightforward answer.

 

I believe a who-beats-who ranking shouldn't be about pros only for several reasons. However, as someone wisely told me in an interview I haven't yet released, "Water skiers are about performances almost as much as they are about placements. The first thing you do after checking the placements is the final scores, followed by the scores in the prelims, just to see…". She's not too far off.

 

Lots to think about. In the meantime, here is the experimental ELO ranking IWWF tested for a while.

[JeffSurdej]

@Horton whatever we do has to be simply and programmable into scoring and rankings so why dont we try something like this. We handicap it all, so if there are 40 slalom skiers at the tournament we essentially have 40 competitors, HOWEVER we only count skiers who have an average on rankings with no penalty, so if 5 of the 35 skiers have a penalty they do not count in this "competition", we then handicap all 35 remaining skiers on a simply plus/minus system. first place gets 35 points, 2nd 34 down to 1, dave allen does all this in WSTIMS for us, shows it on live scoring and the new method gets uploaded to rankings for a new point system rankings, the more people you beat the higher you rank, so yes Nationals will be worth 600 points maybe? We can figure out the details of how many tournaments to use or how to use regionals and nationals b/c some regions just simply ski 2-3 times as many events as others. IDK but this is simple, it adds competition regardless of divisions and gender and each tournament could even and should be seeded in this manner so you truly feel like all 35 skiers are skiing against each other?

[MattP]

@Luzz the ELO List is interesting I will need to read more into the documentation.

http://www.iwsftournament.com/homologation/ELO%20Ranking%20System%20Description.pdf

 

Have you or @Horton looked at this placement points based list? This looks to be more of how your new list is based but weights the tournaments differently and takes into account the number of competitors as well as and how they were ranked in the previous year. This seems to be somewhat of a combination of what our old list was doing but using all events and not just "pro" events.

http://www.iwsftournament.com/homologation/DisplayRanking.php

http://www.iwsftournament.com/homologation/IWWFPlacementRanking.pdf

 

[Luzz]

@MattP I remember seeing this in the past. However, I think the weight on the type of tournament is not necessary and confusing. If I am a 2200 ELO and I lose against a 1800 ELO, I should lose the same points in relation to that person, no matter if it is the Worlds or some backyard tournament. Of course, water skiing is not a h2h sport like chess or tennis, so the ELO calculations have to take into account everyone attending whom one competes against.

 

E.g., Nate winning a backyard tournament with no good skiers there will not award him anything, because he is expected to win (given the lower ELO scores of the attendance). Now, you throw a few good skiers in there, and that changes.

 

Naturally, a consequence of that is that good skiers are less likely to attend backyard events, which (rolling drums) resolves a lot of the issues we have been discussing for months.

[Horton]

@JeffSurdej My idea is really directed for the Level 9 and above. As for the rest of the list IDK. The current system is the simplest.

[Moffattra]

I'll admit I don't really understand the current ranking system, or exactly how ELO or something like it would change it. But, it sounds like you're trying to come up with a way to rank skiers based on how they ski against each other on a given day, as opposed to how they ski any given course.

I know it's been brought up before, but this is the way skiers are ranked for alpine ski racing. While the system on the whole is "pretty good" at accurately ranking skiers from all over the globe, it has a couple dangerous side effects:

1. To improve your ranking you need to go to events with better (ideally much better) skiers. This often gives an unfair advantage to athletes who can afford to "chase" events for points. Or even the lucky ones who happen to live where more of those events take place.

2. To improve your ranking you need to ski closer to or better than a higher ranked skier. This leaves the door open (especially at "backyard" events) for high ranking skiers to ski below their ability, artificially elevating the scores of lesser skiers.

 

Bottom line; there is probably no system that is perfect and people will always try to bend/leverage the rules to their advantage, and while it sucks when people get ahead of you unfairly, hopefully the system separates the skiers who can get a good ranking from the ones who can ski a good ranking.

my $.02

[Horton]

In the last few day I have tweaked on this idea a little more.

• This is not to the advantage of elite skiers who do not get to ski in events against many other elite skiers

• This list ranks competition not scores

• This list sees back yard events the same as Pro Events

• Points are purely based on who beat who

• The currently imported scores are only a sample data set

 

What I is did is as follows:

• Imported last years world standings as a starting place for point values - this is imperfect but is the logical starting place. Each skier's value is the number of skiers below him on the list divided by the total number on the list. Future years list would use the previous version of this list. See tab named 2019 List

• Imported the placements for each event and accumulate points for each skiers placement. See tab named Events

• Add all points earned for each skier. See tab named Result Summer

• Display skiers ranked by points earned. See tab named RankingList

 

here is the link again https://docs.google.com/spreadsheets/d/1PgX5ilF85QcwMgPceAPipADXa3hQa0ZYucyc7SRz5wk/edit?usp=sharing

 

 

 

[Horton]

Interesting .... Moomba was the richest event in terms of points because it has the biggest and strongest field.

 

Also notice that TGAS is #3 but only has 4 events as of lunch time today.

 

[Kelvin]

Why didn't the Stillwater Pro Team Challenge make the list?

[Horton]

@Kelvin just a sample data set

[Horton]

@Luzz My eyes glaze over when I look at ELO rankings because I do not follow the math but as time goes on I am more and more inclined to think it is the right method. I think it would solve a lot of issues in the sport. I wish you could get an answer about why IWWF dropped the effort.