Results and Rankings.

[AAH]

Duplicate post from 'Are Old Men Getting Too Quick' 'cos it belongs here really.

A while back I thought about writing a post on this thread that went like this:

The ranking scheme fits a straight line in a plot of time versus points. For every minute you are behind the leader you lose x points, depending on the race. So that means that if you worked out a typical gradient for that line you could express ranking as times in a mythical 'typical' race instead of points. For example, the person with the top score (average 1349.5 points) might be 60 minutes in this typical race and a person with 1000 points might take 120 mins. For every 5.825 points you are behind the leader, you would take an extra minute.

So the ranking list would look like this:

1. 60'00 Craney
2. 62'48 Rocky
3. 62'55 Mharky
4. 64'19 OJ
5. 64'36 Richard
5. 64'36 Spongey
7. 65'24 Kris
8. 65'28 Hector
9. 66'14 Oleg
9. 66'50 Ralph
10.66'50 Coombs

1280. 120'00 Average Orienteer

This might make an interesting display option for the ranking list that is a lot more intuitive.

However, I then had a look into what the correct gradient would be for a 'typical' race. If the top person takes one hour what does the person with 1,000 points take?

You could express this as how many points you would lose per minute if the race times were scaled so that a person with 1349.5 points took 60 minutes. (e.g. For a middle distance race in which 1349.5 points takes 30 minutes, you double all the times etc.)

The answers for some important recent races come out like this.

'Points lost per minute behind leader if 1349.5 points took 60 mins' / Race

2.41 Bigland (WOC Qual Middle)
3.21 JK Tyrella
3.60 SOL Culbin East
4.13 BOC Wharncliffe
4.14 Holme Fell (WOC Qual Long)
4.22 Slieve Croob
5.05 SOL 1 Dalruzion
5.25 Hambldeon
5.80 SOL 2 Dallschyle
6.00 CSC Longshaw
6.11 British Middle
6.39 Robin Hood Trophy
6.78 British Sprints
6.92 Thetford Thrash
7.38 Edinburgh City Race
8.63 London City Race
9.41 JK Sprint

Which pretty much lists the events in reverse difficulty order but more importantly means that you could expect a huge difference in speed ratios between orienteers at two different scores.

So someone with an average score of 100 points (per event) behind the leader would expect to take 1.18 times as long at the JK Sprint but 1.69 times a long at Bigland in the WOC Qualifier middle distance...

This goes to show a few things. It would be pretty hard to come up with a typical race to show ranking as race time. Plotting straight lines through curvy data goes a awry in places. But it's also relevant here in that it shows that course length ratios are also very terrain dependent... The course length ratios to get a given running time for an urban race and for a difficult race in technical and physical terrain have little resemblance to each other.

But one could use a calculation such as this, using numbers from a previous race, to get an idea of the sort of time someone at 1200, 1100, 1000 ranking points etc. is going to take and so customise the course length ratios to an area?

(Calculation: fit a straight line through two ranked runners times/points. Work out what time 1349.5 would have got. Work out ratio to make that 60. Multiply the gradient of that line in points per minute by the same ratio. All figures calculated for longest course. Would have put some more SOLs in but only 3 had ranking points on the black course.)

[ic]

The Swedes do (or at least did last time I looked at anything like that which was probably in 2004) express their ranking list in terms of minutes behind an idealised winning time (of I think 75 mins).

[King Penguin]

Different sport - Wild Water Racing (canoeing) converts the winners time to 20 minutes no matter what the length of the race then applies the same factor to everyone else's time.
Points = 500 minus (seconds your adjusted time is > 20 minutes), so winner gets 500 points.

HOWEVER - the big disadvantage is it is very, very dependent on who turns up, as it is all based on the winner's time. Basing calculations on an average or distribution of results smooths the anomalies and therefore gives a "better" solution.