Register | username: pw: 
Attackpoint - performance and training tools for orienteering athletes

Discussion: Point Thieves

in: Pink Socks; Pink Socks > 2016-02-05

Feb 8, 2016 7:16 PM # 
Pink Socks:
One thing that's interesting about the Winter Series scoring is that for the season standings, there can be "point thieves", which I define as people who didn't run 4 races, but did well enough in the 1-3 races that they did run to steal points away from others.

JV Girls

This one is interesting! Jessica and Corinne tied with 390 points for the season, with Jessica winning the 7th and final tiebreak, basically because Jessica ran all 7 races, while Corinne just ran 6.

However, head-to-head, Corinne beat Jessica 4 of 6 races!

Looking at the results, the key here was at Race #2, where Caitlin made her debut, won the race, and then didn't show up again until this weekend. Her 100 points bumped Corinne down to 95, Jessica to 92. If Caitlyn doesn't show up, Corinne gets 100, Jessica gets 95, and Corinne wins the season championship 395-392! Unlucky Corinne!

Women's League

Rebecca beat Marissa for the season 392-385, but Rebecca didn't beat Marissa head-to-head until Race #5! Marissa won the head-to-head battles 3-2! Looking at the results, Marissa had points "stolen" from not one, not, two, but THREE point thieves!

Marissa started 3-0 against Rebecca to start the season, but since Heather was still healthy, none of those translated into 100 points. Furthermore, at Race #2, Celia finished second, pushing Marissa another spot down. And then at the end of the season, this mysterious Nina person shows up, stealing 2nd place, and bumping Marissa another spot down. Heather only got in 3 races, Celia just 2, and Nina only 2, so no qualifications from any of them.

If none of those ladies had raced, then Marissa finishes with 395, as does Rebecca, but Marissa wins the tiebreak.

Marissa isn't as unlucky as Corinne, though. Marissa (and Rebecca) had no chance of winning if Heather doesn't get hurt. And who knows what happens if Celia runs in more races. But Corinne lost out basically because one runner showed up one day and had a fluke race!
Feb 8, 2016 10:54 PM # 
Yes, but when Marissa beat me, it was by a margin of 2-3 minutes. When I've beat Marissa, it's been by a margin of 7-13 minutes. ;)

If you added up our times at the races we were head-to-head, I think my cumulative time would be shorter. (though I'm too lazy to check)

Unfortunately, the places in Winter-O don't have much meaning, since they poorly portray a racer's actual performance. The difference between X and Y place could be 10 seconds or 10 minutes, and the Winter-O scale doesn't care.

I really wish we had the % scale that is common at A-meets. Not only would it be more fair, but it would be a ton more useful for gauging how I'm doing.
Feb 8, 2016 11:28 PM # 
Pink Socks:
I figured that you had a better cumulative time than Marissa, but I was too lazy to check, too. I think Jonathan had a better cumulative time than I did, because he had that one great race at St. Ed's that got me by 12 minutes.

With the scoring system we have, there can be point thieves that can change outcomes between completely different people, which was really the point. If your 7-13 minute wins came early in the season, and her smaller wins late, she probably would have won the series because of Heather. In the Women's League's case, the better orienteer won (you, yay!), but probably not in Corinne's case.

And yes, there are obviously better methods of ranking orienteers. But, I've long defended the Winter Series system, chiefly because it's super easy to use, explain, and implement. For example, heading into Fire Mountain, you had to win it or finish second. We didn't have to do any math like you needing to beat Marissa by 5.3% or whatever.

High school XC uses a point system (the basis for the Winter Series system, except that we count down, not up), as do pretty much all sports. In football, a 20-yard field goal is worth the same as a 50-yard field goal. A 2-foot shot is worth the same as a 20-foot shot, but a 21-foot shot is worth one more. Can you imagine a basketball scoring system that better rewards the best shooters to an exacting scale? "5 seconds left, we're down by 3.17 points! Quick Bobby, shoot the ball from 21 feet 8 inches! Bobby makes it! Oh, wait, that was 2 inches too close! We lost by 0.07 points. Nooooo!"

The Winter Series scoring system doesn't crown the best orienteer, necessarily. But it does crown the orienteer with the best win/loss results. Oftentimes, these are the same people, but sometimes they are not.

A very famous example of this is the 1960 World Series. The Yankees outscored the Pirates 55-27 (a lot!), outhit them 91-60 (a lot!), outbatted them .338 to .256 (a lot!), out-homered them 10 to 4 (a lot!).... but they lost the World Series 4 games to 3.

We do use the % scoring for Ultimate-O, so there's that. The A-Meet algorithm is altogether different. If we used the A-Meet scoring system yesterday, I think it would have broken down in Elementary Boys, since there were only two starts. (I'm not an expert on that algorithm, though)
Feb 9, 2016 2:38 AM # 
I'd argue that for simplicity's sake it's fine for WIOL to use the current system, since it's probably a tough ask to get most kids to care about/understand a system with a bunch of oddball numbers, percentages, decimals, etc.

But I see no reason for the Winter Series to use a demonstrably inaccurate and dumbed-down system, when an objectively better one exists. An example from this season: Person A at Race 1 who finishes third, 0:17 behind the winner, should not earn a lower score than Person B at Race 2 who finishes second, 6:14 behind the winner. An example from last year: Person A wins Race 1 by 0:04 over Person B; Person B wins Race 2 by 6:07 over Person B; they should not earn an equal total score from the two races. Also, there is no need for consistency between the school league and public, as they are completely separate competitive entities. Other benefits of the % system:

- Ties (while theoretically possible) would in practice be eliminated;
- There would be infinitely more permutations to each event result, creating greater intrigue and suspense in general;
- Certain mathematical symmetries of the current system would be preserved, i.e. one could still clinch the Series with four "perfect" 1000-point scores;
- One could also still drop a bad race (or three);
- It can't be that hard to calculate (presumably we have the equation somewhere), and there are only two sets of calculations.
Feb 9, 2016 2:53 AM # 
But really, IMO what you said sums up the entire problem:

The Winter Series scoring system doesn't crown the best orienteer, necessarily.

In my head, this is equivalent to saying:
"Our 100-meter dash scoring system doesn't crown the person who sprinted 100 meters the fastest, necessarily."
"Our high jump scoring system doesn't crown the person who jumped the highest, necessarily."

It should be necessary.
Feb 9, 2016 7:08 PM # 
Pink Socks:
I'll grant you a few things:

1) I'm a sports/numbers guy. I'm a subscriber to Kenpom. And I obviously know that the current Winter Series scoring system isn't the best way to determine the best orienteer through 7 races.

2) One of the arbitrary scoring system's main flaws is amplified at where you guys (Rebecca & Will) are: at the very very top, because this is where the winner gets a "bonus" 5 points over 2nd, 2nd gets a bonus 3 points, etc. So I can definitely understand why you'd both be in favor of ditching it.

There's a little more nuance to it than comparing this to 100m dash and high jump, because those are reliably the same competitions each time, but orienteering races aren't each time. Let's say that this were the Winter High Jump Series instead and we had to determine the best season-long high jumper using 4 of 7 results. Seems like the most logical way to do that would be to just add up the 4 highest measurements. (Or the 4 fastest times if it's the 100m Series).

To account for each orienteering race to being different, you have to go through some sort of extra step, which is arbitrary to a degree. And since so many other sports have arbitrary scoring systems*, I don't find the Winter Series scoring system all that egregious: it's simple, it's easy to implement, it's unbiased, and it generally gets it right.

* I played tennis in high school, which has a pretty wacky scoring system. You can score more points and win more games than your opponent and still lose, because it's point
Feb 9, 2016 7:50 PM # 
Pink Socks:
With the % system, instead of Point Thieves, there could be Point Suppressors. The only time this becomes an issue is when you have a super-elite who doesn't show up very often. Which is pretty unlikely, oh wait, never mind, Alison Crocker.

In the 2014-2015 season, Ali showed up twice and crushed everyone, natch.

In a % scoring system, Celia still wins, and Gina still finishes 2nd, but barely. Those two ran six times, and their four highest scores are from races without Ali.

After that, it gets interesting:

Current Scoring Format
3rd: Abra, 360
4th: Rebecca, 360
5th: Sue, 360
6th: Marissa, 355

Current Scoring Format, without Point Thief Ali
3rd: Abra, 366
4th: Sue, 364
5th: Rebecca, 361
6th: Marissa, 358

% Scoring Format
3rd: Rebecca, 3361
4th: Sue, 3295
5th: Marissa, 3256
6th: Abra, 3193

% Scoring Format, without Point Suppressor Ali
3rd: Sue, 3779
4th: Abra, 3663
5th: Rebecca, 3637
6th: Marissa, 3464
Feb 9, 2016 11:27 PM # 
Poor Marissa, never gets a chance on being better than 5th :(

On a more serious note, % scoring system I think would more accurately represent the reality and crown the best orienteer. IOF, US rankings and TT rankings alleviate to some extent the Point suppressor problem by averaging the top 3 times to 100 so one aberrant super-elite result will have less affect on the regular O-goers. This could be further modified to be even less sensitive to top results variability with averaging the top 5 to 100%. Patrick if you don't mind, what would the results be if you use those systems in your example above?
Feb 10, 2016 12:13 AM # 
Pink Socks:
% scoring system I think would more accurately represent the reality and crown the best orienteer.

I don't disagree. I'm just saying I like the simplicity of the current method. I'm the guy that drinks Bud and not craft beer.

Patrick if you don't mind, what would the results be if you use those systems in your example above?

It might have to be after Sprint Camp. I just quick-and-dirtied this one in 10 minutes, but the top-3 method would require a little more time than I've got right now.
Feb 10, 2016 1:43 AM # 
I think the question to pose is: what do the participants want from event results? And by participants, I mean adults in Winter-O.

Is it more valuable to participants to have results that are simple to understand, or to have results that more accurately reflect their true performance?

I would argue that the latter is more valuable, regardless of how more or less "fair" the system might be, since it provides better information about an individual's performance.

And, it is valuable to a greater amount of people. Kate may not be vying for the podium, but she would like to know how she is doing. She is often at a loss of being able to figure that out, since she feels like she doesn't have much peer competition (that being people of a similar age, fitness, and ability).
Feb 11, 2016 6:19 AM # 
Pink Socks:
Kate may not be vying for the podium, but she would like to know how she is doing. She is often at a loss of being able to figure that out.

The data is available, so she can always do this on her own. Will just did this in his log, I keep a spreadsheet with all of my results in it, etc.

The Yankees outperformed the Pirates by a good margin in 1960, but they still lost. Should baseball change its rules so that the better performing team always wins? The University of Kentucky had the best college basketball team in 20 years, according to the latest and greatest statistics, but they didn't win the 2015 national championship. Should the NCAA change it's scoring rules?

I know this isn't apples-to-apples, but no matter how you do it, there's always something arbitrary anytime you try to score a series (or season) of events.

Obviously, enough people (Rebecca, Will, Nikolay, Kate) are for a change, and I'm the only one (so far) who is content with the current system. So if this is a trend, then Will should probably bring this up to the board, or lead a larger discussion and see if we should change it, and what system to change it to, for the next season.
Feb 12, 2016 6:18 AM # 
If the Winter Series organizers wanted to change the scoring system, could they just do it without Board approval?

Please login to add a message.