Multi day scoring systems

[EddieH]

Because I don't want to detract from the excellent Croeso this is a separate topic.

The scoring system used is supposed to keep things open and reward the quality of your run - a worthy ideal. However this and other systems based on standard deviation appear in practice to be flawed.

The first time one was used in the Scottish 6 days I didn't run one day but comprehensively beat the winning Swede on the other days. Problem was that he received such a shedload of points for his win that he was unassailable (the second placed runner, some 8 minutes behind scored more than I did for any day). I was told that it was a freak.

This year I missed a control on day 3. However it turned out that this was fatal as there were mega points going on this day (Without my aberation I could have got them too despite it not being a particularly great run).

On day 6 in a 31 minute race I won by over 2 minutes and thought that just maybe the points would be good. In fact I got my lowest score of the week with around 40 points fewer thn the norm. In order to make verage winning points I'd have needed to win by 3 1/2 minutes and to get the points needed to make up for day 3's cock-up I'd have needed to do about 27 1/2 minutes with the World's no 4 M60 sprinter doing 33 . Clearly inconceivable and therefore the ide that the scoring system leaves it all open just isn't true.

Every day is a separate race and therefore I like the Scottish 6 day system - really simple to understand and closely contested classes remain open to the end. Also you look at the results on day 6 and you know. No-one has a clue until they spot a list just before prize giving, or they find out there.

My examples are far from unique - many people talked of dropping their winning run scores.

[andypat]

The scoring system used is supposed to keep things open and reward the quality of your run - a worthy ideal. However this and other systems based on standard deviation appear in practice to be flawed.

Every day is a separate race and therefore I like the Scottish 6 day system - really simple to understand and closely contested classes remain open to the end. Also you look at the results on day 6 and you know. No-one has a clue until they spot a list just before prize giving, or they find out there.

Hmmm. I take the view that you enter a race knowing what the scoring system is and run with it. There were also plenty of people who had a wee moan about the scoring system after Oban 2011.

Not sure I agree that the Scottish 6 day system makes close classes open till the end. You could lose the first 4 races by 1 second and be out of it in the Scottish - at Croeso you would be very much in it. I quite like the fact that different events use different scoring criteria. One thing I would have liked at Croeso would have been to see the score for the day alongside the time on the daily results. Only having the score on the cumulative scores did make it more difficult to visualise the score to time relationship.

[EddieH]

I've clearly made the wrong impression. I enter the event regardless of the system and go with it. However I dispute that this system achieves what it aims to.

[andypat]

Where do you read that keeping the scoring open till the end was an aim of the scoring system? It doesnt say that here.

http://www.croesomultiday.org.uk/general-info/scoring-system

[EddieH]

Not there. I just remember that that has been a major argument for using standard deviation - one criticism of other systems being that results can be wrapped up after 4 days.

[Sloop]

The issue is more subtle than "statistically-based systems are bad". If you look at the winning scores on the 4 open-fell days, they were remarkably consistent, and if all 6 days were of that nature, the system would have been perfectly fit-for-purpose.

However, it fell down for this event because two of the days were completely different and the stats weren't comparing like with like.

Day 6, being urban, was obviously different. I suspect a couple of effects - the slower runners run relatively faster in urban events, and mistakes are typically smaller - so the time spread is smaller and points harder to come by. Whatever, it ain't the same.

Day 3, being Middle and Forest was also obviously different (and IMHO completely unfair as a one-off competition). Super-fast times posted by starters in block 6; decent runs ending up miles adrift for starters in early blocks - could it possibly have tracked up in giveaway proportions through all the brambles and other floor coverings ? To get a fair system on this area when used by so many people for such short distances would have needed an algorithm that factored in start time.

So my view would be that the simple positional system used in the last Scottish 6-Days would be more appropriate for multi-day events comrpising races of different types. I would still prefer a statistically-based system where days are sufficiently similar.

Incidentally, take a wild guess which two days were picked for the UK Masters Cup.

[AndyC]

It is similar in some ways to the Tour de France where they have two jerseys Yellow based on time and Green based on position (roughly similar to performance relative to others) - maybe a two prize solution if funds allow?

[andypat]

NIce idea - maybe 10 prizes and I might get one....

[King Penguin]

If it were to be based on total cumulative time across all 6 days you would disenfranchise all Planners and Controllers and may have problems obtaining such volunteers.
I doubt there is any scoring mechanism which will satisfy all the people all the time.

[Another180]

The scoring system used is supposed to keep things open and reward the quality of your run - a worthy ideal. However this and other systems based on standard deviation appear in practice to be flawed.

The system used at Croeso is not based on standard deviation. It is based on the mean of the first 50% of times.

This year I missed a control on day 3. However it turned out that this was fatal as there were mega points going on this day (Without my aberation I could have got them too despite it not being a particularly great run).

On day 6 in a 31 minute race I won by over 2 minutes and thought that just maybe the points would be good. In fact I got my lowest score of the week with around 40 points fewer thn the norm. In order to make verage winning points I'd have needed to win by 3 1/2 minutes and to get the points needed to make up for day 3's cock-up I'd have needed to do about 27 1/2 minutes with the World's no 4 M60 sprinter doing 33 . Clearly inconceivable and therefore the ide that the scoring system leaves it all open just isn't true.

The reason that day 3 was a mega points day was because the difference between the winning time and the par time was large, and it was a short day. The par time on day 3 was about 42 minutes and the par time on day 6 was about 36 1/2 minutes. The scoring system quite rightly reflects the fact that a 28 minute run on day 3 is far more impressive than a 31 minute run on day 6.

A system based on standard deviation would have helped you slightly in this example, with day 3's winner receiving less points than they originally did because the times were more spread out; day 6's winner would have received more points than originally to reflect the fact that the times were more concentrated. Day 3's winner would still receive more points than the winner of day 6 because the winning margin is so large (12 min to 10th place, compared to a 4 minute gap)

Every day is a separate race and therefore I like the Scottish 6 day system - really simple to understand and closely contested classes remain open to the end.

I just remember that that has been a major argument for using standard deviation - one criticism of other systems being that results can be wrapped up after 4 days.

Under the Scottish system the racing for the first 2 places in M60L would have been over after the first 5 days. Under the Croeso system it was still possible to win overall and therefore the race was still open till the end.

The Scottish system is clearly flawed because it does not reflect the quality of wins or keep the competition open. However, it is simple to understand which is useful. The Croeso system is obviously better because it does reward exceptional runs and keeps the competition open, but it is more complicated. Another flaw of the Croeso system is that it can not differentiate between days where the spreading of the runners is different, for example forest days and urban days. A system using standard deviation would succeed in solving this problem, however it would add yet more complexity.

Example:

Runners Points = 1000 + 250 * (Runner's Time - Average of Middle 50% of times)/(Average of Top 50% of times - Average of Middle 50% of times)

This system works well at rewarding the runs of winners (especially when they win by a lot on a day when the spread of the times is small), however, it heavily penalises runners taking a long time with the prospect of receiving 0 points for a particular day being far greater than under the Croeso system. The only way to remove this problem is to add in a log which adds far more complexity than it is worth and makes the system utterly incomprehensible for anybody other than the results team.

Therefore the only fair and simple system is to use overall time and count all days (but I can't imagine that being very popular in Britain, Sweden on the other hand seems to manage perfectly well

[ricardito]

... Therefore the only fair and simple system is to use overall time and count all days

It's certainly simple, but it's difficult to argue that it's "fair" in the situation where different days are of widely different styles and (crucially) expected winning and average times (for exactly the reasons you've pointed out elsewhere in your post).

I doubt there is a system that is both "fair" and "simple". Suggestions on a postcard...

[Big Jon]

The Swedish 5-Days works having no one from local clubs running (except juniors) because firstly the O-Ringen only comes round to a region once every 10-15 years so the loss of a set of races is not desperate and secondly it is standard for Swedes to not run when their club puts on a large race.

[Sloop]

You seem to have missed the point that the "exceptional times" on day3 were very likely not due to exceptional performances, but exceptional tracking up. They should NOT have been worth extra points.

[EddieH]

"Exceptional" runs How you can say that one run is definitely way better than another I cannot fathom. I can only assess performance by judging how well I was running both physicaly and technically. On day 3 Jean Dermine's run was excellent but not out of sight, and certainly not 70 points better than the normal days.

The fact remains that the result was not open for change on day 6. Could I have equalled Jean's 1442 points on day 3? - most definitely. Could I have run 1770 points on day 6? In my dreams. Could I have approached 1400 points? Absolitely and emphatically not.

The reasons that I like the S6D system are:

It acknowledges that each day is a race in its own right,
It is easy to see how you are doing instantly,
You know rhe final result early on day 6.

The reasons I would like the Welsh system if it really delivered would be:

It keeps competition open until the end,
It is intriguing.

Sadly I don't think it does.

Fairness says the only way is total time over all days. However even that is not perfect when one massively hard day will decide the competition (c.f. Grindelwald in Swiss 5 day around 1979 won by Oyvind Thon 8 mins ahead of brother in turn 7 minutes ahead of the rest).

Whatever system is used is fine by me - I enter the competition regardless. I simply query those that believe that Croeso has come up with the perfect fair answer. It is one answer and will always suffer from impenetrability and slowness in results appearing.

[Homer]

IMHO the exceptional results on Day 3 were due to the fact that most people were making more mistakes than usual. This was more down to the dodgy map than anything else.
The leaders on M45L were all agreed that someone should have run around 30-32 mins. In fact the winning time was 38 mins and the score for that was in line with winning scores on other days. If someone had done 30 mins they would have scored 1500+ points and that would probably have sealed overall victory.

Whilst I sympathise with Eddie, I don't think there is any completely satisfactory scoring system which allows you to drop one or more day's result.