Results - Scottish 6 Day

[Bin Man Pete]

Don't pretend to understand the old or new system in detail but did I see some classes have joint winners/2nds/3rds? This doesn't feel right over 4 (counting) days of competition. I appreciate this could happen with any scoring system but does this one lend itself to this more at the top end in classes where there are fewer cometitors? Is there a simple way to produce a cummulative time for best 4 days as a tie-breaker?

[ba-ba]

rock-paper-scissors as Tie-breaker, on the podium. Nuf Said.

[graeme]

- if somebody wins 4 days, it doesn't matter how much they get beaten by on other days, that's the best performance.

Like the way Cavendish won the Tour de France ?

Is there a simple way to produce a cummulative time for best 4 days as a tie-breaker?

There are plenty of ways to extend the tiebreak which would look sensible. But since half of them would put one person ahead, and half the other, you might as well use the more spectator-friendly rock-scissors-paper challenge.

The new system has more anomalies and is easier to calculate than the old one (though with computerized results that's hardly relevant): I guess 95% of the results would be the same.

[IanD]

the only thing to improve would be giving people no points for days that they don't finish/start.

Not quite, with low points being best, people should be given high points for days they don't finish (or start).

Think of it like golf, where the lowest total for a round wins. If someone doesn't play a hole at all, would you record a zero for them, beating even a hole in one?

[andy]

the only thing to improve would be giving people no points for days that they don't finish/start.

Not quite, with low points being best, people should be given high points for days they don't finish (or start).

he means no points as in 'n/a' not '0', subtle difference

[mharky]

Like the way Cavendish won the Tour de France ?

Not quite the same, is it?

[The Malcy]

I did not like the new system (maybe I just don't like things being changed...). For me the 6 Days (as for any multi day event) is about cumulative "time trial" performance over the designated number of days. This has to mean using time or some designated metric of such. Using positions IMHO changes the dynamic of the competition but without giving the opportunity to race shoulder to shoulder. In the new method there is no greater reward for winning the day by 5 minutes as opposed to 5 seconds which I believe there should be. Also, in a tight finish I may end up having orienteered well that day but getting significantly worse points than on a day where I may not have performed so well but placed higher. The old method rewarded personal performance rather than race placings.

Another factor I did like about the old method was that it made it possible to do cross class comparisons and run small club and family sob competions using the scores as all were referenced to the average for each age class.

I can see that the different methods make little difference to the results at the top end (the best orienteers should win out under any system) but I am sure it does make a difference further down the lists.

Having said all this it was a great 6 Days which everyone in our party thoroughly enjoyed.

[Zokko!]

I agree that the previous scoring system allowed apparent anomalies such as someone could win 4 days but still not win overall. This happened to one of our family a few 6 days back. Awarding a fixed number of points for 1st place obviously prevents this happening.

What about the system in use prior to the last one, when winners got 1000 points and everyone else's score was calculated by comparing their time relative to the winner? Thus a very slow runner who still completes the course still gets some points. It is impossible to score zero, unless you mispunch or don't run.

Reverting to that older system would surely overcome most of the problems raised in this thread. Ties would be unlikely and every second lost would damage your score on that race, unless you are in first place. Badge awards can be points based using the same points ranges for every class, allowing a kind of comparison with others on different classes. It's simple to understand and 4000 is the maximum score. No-one can get extra points by the lucky chance of others running badly one day.

Our whole family enjoyed the event and it didn't matter to us what scoring system was used since we were all well out of the prize zone anyway.

[superstartradesman]

Just for info - a few additional news items have been posted to the Tay 2009 web site since the end of the competition, including:

• downloadable certificates for top-3 finishers on each day;
• PDF documents containing individual results for each day;
• answers to the Tay Today quiz;
• access to the post-event survey, the results of which will help monitor impact of the 6 Days event on the local area, and also be used for feedback on the event format (including views on the points scoring system )

[frog]

I didn't like the fact that the survey focussed on Perth and Kinross and didn't mention wider Tayside and Fife which are also within the 6 day region. Was this because Perth council were funding it?

[andy]

Reverting to that older system would surely overcome most of the problems raised in this thread.

and introduce its major flaw - that if someone wins by a loooong way, everyone else is heavily penalised.

sadly there's no 'perfect' system... or we'd be using it

[SIman]

Routegadget for the Sprint race is now available here. See how you got on racing head to head against your rivals.

[david_rosen]

I am pleased that a discussion has started about the scoring system used last week. It is good that the organisers have requested feedback on the scoring system in their survey, but the question emphasises the simple nature of this year's method which will encourage people to forget that there are other important requirements for a scoring system.

Actually I’m intending to write an article comparing the various different multi-day scoring systems but here are a few thoughts.

The variable nature of orienteering courses, orienteering terrain and the rather random nature of our own performances means that there can be no scoring system which will be perfect.

However, there are certainly some which are better than others.

To decide which is best, we first have to agree on the requirements and weight them according to their importance. Too often, people only consider one requirement at a time, pick the system which best satisfies that requirement and then later realise that there are problems because other requirements are now not satisfied. That tends to lead to switching scoring system every few years.

The ‘best’ system is not going to satisfy all requirements equally, so it is important to be a bit robust when there is criticism that one less important requirement is not fully satisfied.

My list is roughly as follows with percentage weights attached to each point:

1. Competitors can drop one or more days (15%)

2. The scoring method is very simple to understand (10%)

3. The scoring system can produce a credible chasing start if required (5%)

4. The system is good for commentators (10%)

5. The system produces a 'fair' overall result for all competitors (the also-rans as well as the top few) (60%)
i.e.
a) A runner’s counting runs should be the ones where he/she ran best. Another way of stating this is that a runner’s points for a day should depend entirely on the quality of their run that day. It should be possible to achieve equally high scores on each of the days of a multi-day event. It is frustrating for the competitor (and a waste of the efforts of the planner) if on one day, it is just not possible to score good points. Conversely, it is not fair to those who miss out a day (perhaps they are planning or controlling) if that is the one day when especially high scores are achievable.

b) The scoring should be linear with respect to time i.e. On any one day, 10 minutes near the top of the results is worth the same number of points as 10 minutes lower down.

c) The scores should as far as possible be invariant to the performance of any one competitor.

For me, the 'gold standard' for multi-days is the O-Ringen system where times are cumulative and there is a chasing start on the last day. Even that is not perfect - not least because my first requirement above is not satisfied at all! It was good that the 6-Day used a cumulative total time system for the Elite this year - at least they had a pretty fair competition.

As you can see, I rank fairness way above other requirements. People will never (in the long term) be satisfied with a very unfair system.

The most important thing is to calculate the position of runners in a class over the 6 days based on their performances on the individual days.

All the non-Elite classes at the 6-Days used a time-trial system on each day. In a time trial, the performance is the time that you achieve, not the position. In a mass start race, if you are in the lead near the finish, you can look over your shoulder and ease up at the end knowing that a win by a second is as good as a win by a minute; you can't do that in a time trial.

If a time trial is the only race in a competition, then, yes, positions in the race are all that matter in the end. But in a mult-day event, just using the positions on each day and throwing away time differences leads to all sorts of anomolies.

Actually, this phenomenon is well known by mathematicians and it is accepted that when calculating an overall final score from a number of separate performances, the actual scores/times from each performance must be combined rather than just their
ranking.

I'll be analysing the results from this year to identify where such anomolies occurred. Because the overall results sheets do not show times as well, it is quite difficult to immediately see such problems.

Andy, could you let us know when the next Scottish 6-Days meeting is so that we can send any comments to you in time?

David

[graeme]

Excellent summary David. I would add two points...

1. Competitors can drop one or more days (15%)

This is essential for the 6-day. If you don't have a system where volunteers can have a chance of competing, we'll just walk away. And you'll either have no six-days, or a six-days run by people uninterested in competition.

a 'fair' overall result

The word 'fair' has come to mean "in accordance with my prejudices". It's usually only use by people who can't be bothered to make a coherent argument. But since you define it, I'll let you off.

I completely agree with your (a) and (c)

"linear" with time is overrestrictive - total time is the only way to do that. Why linear, but only within a day? Presumably because your system has that, and others didn't . The right criterion is "monotonic" - the faster you go, the better you score.

It would be interesting to know how different the results would be under the old system. I suspect not very... e.g. Despite the different formats and special scoring the elites had, their 1-2-3 has only one difference when scored by same system as everyone else...
Matt Speake would be second - which would hardly be embarrassing.

[PG]

Is another requirement to be able to hold a prize giving within an hour of the last possible prize-winner finishing? As I understand it the system used previously relied on having all finishers back to calculate the standard deviation accurately. This year the prize giving started about 40 minutes before courses closed.

A runner’s counting runs should be the ones where he/she ran best.

I have long had a belief that the use of standard deviations in S6D and Ranking Lists has meant that a good run on an easy area where standard deviations are low is more likely to lead to high points than a good run on a highly technical area where the spread of times is wide. I started to try to prove it at one point, but didn’t have enough time. Maybe I’ll have another go…

BTW.

Having a brief scout through the M60L results I reckon it would have been very very close on the old points system.

Based on the points calculated in the upload to the ranking system (although these are weighted the weighting should have been very similar each day), Andy Hemsted would have just beaten me on the old system. But to back my point above my 3-minute win in the technical forest of Dalrulzion on day 1 scored me less ranking points than my third place at Kinnoull Hill the following day.

Peter Gorvett.