Ranking points

[HarryO]

I have always known that I really didn't understand the calculation behind ranking system - but I blindly trusted it. Now I'm not so sure.

Thirteen runners all get the same points - despite being separated across a 5 minutes range from 1:06 to 1:11 minutes. See the Blue course at Lynford last week. Plus several other duplicates. Ditto Green course.

http://www.britishorienteering.org.uk/i ... course=10&

What's going on here?

[greywolf]

before slagging the rankings system, why not look more carefully at those "results"?
in particular, look at the Blue course competitors who supposedly finished 33rd - 40th in times ranging from 52 hours to 59 hours 57 seconds
and then compare with the results on the NOR website which shows that those competitors were actually 1st - 8th with times from 52 minutes to 59 minutes 57 seconds...nothing wrong with the rankings, but if you give it silly times it'll return silly points, just needs the correct results uploaded, that's all

[HarryO]

That's interesting. But even allowing for the rogue errors at the bottom of the list - its puzzles me that runners with different times got identical ranking points. Sure I read in another thread that there is a linear relationship between time behind winner and points. So something has gone more awry than simply misreading minutes for hours.

I really wasn't "slagging the rankings system" - I genuinely seek enlightment

[greywolf]

well yes, but if the system is trying to spread 450 points over 58 hours then each point covers about 8 minutes.

it's hardly a few rogue results: the system uses the mean and standard deviation of ranked runners scores: chucking in a few 50 hour plus result makes a massive difference to both, and to the range over which the points must be distributed

[andy]

excel time format wins the day?

[graeme]

What is really remarkable is that given this "crazy" result, the list is still remarkably robust. Of the people with 6 counting scores, the effect is no more than a few dozen places in a list of 5000+ people. The list does not claim any strong meaning to individual scores, that's why only the average* is used.

* Average over 6 best for publication/incentive, over all runs to drive the stats.

[roadrunner]

Perhaps adding a trap for incorrect times to the system might avoid this sort of mistake in the future. I can't see any ranked event having a winning time below 5 minutes, nor any runner taking over 5 hours, so how about a check when the results are uploaded that they're all between those limits?

[Angry Haggis]

Perhaps adding a trap for incorrect times to the system might avoid this sort of mistake in the future. I can't see any ranked event having a winning time below 5 minutes, nor any runner taking over 5 hours, so how about a check when the results are uploaded that they're all between those limits?

I can. We just haven't invented those formats yet, or haven't run them in the UK yet.

A lower bound of 60 seconds and an upper bound of 24 hours would catch the formatting issues like the classic Excel trying-to-help problem that's caught this race out here.

[Big Jon]

Excel is always trying to screw up Deeside Night Cup times when I enter them - its a pain in the proverbial!

[MIE]

I always make sure that the "time" field is specified as text (instead of standard)
when I import/check the data in Excel. That way Excel doesn't try to "help" by
converting the minutes (under 1 hour) to hours, which is what has happened to
the Norwich event.

Hopefully someone from NOR (or EAOA) will advise the results "uploader" of the
problem before the expiry of the correction time so that the correct points will be
allocated to those competitors who think it is important.