Hi
It seems two issues are coming up  repeatedly :

1, Should outlier times be removed before calculating the points score per event - i don\'t think anyone disagrees with this in principle, the question is the best /most practical way to do so.

2, Should the notional score of 1000 points per event, go to the winner or to the \"middle\" competitor, however middle is defined, as mean median etc. There\'s more debate about this but i would give it to the middle finisher as currently, as this avoids them getting less points in the case of a large field, I think this is fairer, it also allows strong performances to be recognized.

Paul, my apologies, I\'d misinterpreted your point. I see what you mean now. However, if we look at the first female Blue competitor at Castlecomer and Ballyward:

1) The first female at Castlecomer was 16 minutes behind the course winner and scored 1,141 points;
2) The first female at Ballyward was 5.5 minutes behind the course winner and scored 1,212 points.

I would argue that the winner at Ballyward was rewarded for an exceptionally fast run at Ballyward and that she\'s perfectly entitled to carry this across the league. I\'m just focusing on first female to follow your example but the same logic applies to all the course winners.

Rory, you\'ve hit the nail on the head!

Stuart

Hah, hah! Nice try Stuart, but that argument can\'t be made to work.

I can play that game as well. Here\'s one.

The winner at Ballyward was 6 minutes behind the course winner and got 1212 points but the winner at Trooperstown was only 2 minutes behind the course winner and only got 1174 points. In addition the min\\km was faster at Trooperstown (9.6 vs. 10.9).

Here\'s another. The winner at Emo court was 27 minutes behind the course winner and received 1122 points, a mere 30 points less than the winner at the Curragh who was only 8 minutes behind the course winner.

And another. The winners at Rossmore and the Pine Forest both came first on their course as well as being first female yet one got 1312 while the other only got 1250.

Within a single events\' results for a single course, the metric of minutes behind the course winner is a perfectly meaningful comparison. There is no way to make that consistent across multiple events with varying fields etc. Unless you can provide a statistical model that can balance or control for these factors? Do you have such a model that we can critique? By the same token it cannot be used to explain away the variation in points awarded that we are discussing.

Also, and I keep coming back to this - Why should the ranking of a female competitor depend on how the male runners on her course did? In particular why should it depend on the male runners who did better than her? I\'ve yet to hear any rationale for that viewpoint.

Rory\'s suggestion that

Should the notional score of 1000 points per event, go to the winner or to the \"middle\" competitor, however middle is defined, as mean median etc. There\'s more debate about this but i would give it to the middle finisher as currently, as this avoids them getting less points in the case of a large field...

Since the purpose of the league and it\'s scoring system is to find the best competitors in each category across the league, this suggestion makes no sense to me. How is this a problem? It sounds like you are talking about the magnitude of the points as if that matters, whereas what matters are the relative values. If you are talking about the magnitudes then that\'s a matter of perception. Is the issue that people in the middle of the field are used to getting a particular range of values and will be put out by a change in that? Something else? I\'m really not clear what the issue is.

Points for a single event are linearly distributed. Suppose that that the points are calculated as

800, 950, 1000, 1150, 1215

and we rescale them proportionally to a maximum of 1000, by dividing by 1215 and multiplying by 1000, then we get

658, 782, 823, 947, 1000

Now the person who would previously have scored 1000 points will now score 823, but since everyone else has had their points scaled, within this event everyone is in essentially the same relative position as they were before. But now we can consistently combine results across events, without having a 200 point range of values at the top end.

The NIOA League ( https://www.niorienteering.org.uk/wp-content/uploads/2019/11/NISeries2019Final.pdf) uses the following system:

Points: are calculated only for 2019 NIOA members and separately for males and females. The computation is:- P = (Standard Time / Your Time) x 100 where Standard Time is that of the 1st NIOA competitor of your gender on the respective course.

Notice that their system recognises the need to:

* Harmonise the scores across different events (i.e. same maximum score)
* Rank men and women separately on the same course

It still suffers from some problems. The first is that it is time based and in this case I suspect it may lead to some unjustified variation in points across the league for runners behind the winners. I have not analysed this at this point so it\'s still an open question. This weakness also applies to the scaling I used in the example above.

Another issue is the skewing effect of the way that OPC points are awarded. Since they award copies of the runners maximum score this can result in someone winning because of their OPC points rather than their competitive results. If we accept that any given runners results will follow a normal distribution then doubling their best result is a definite skew. OPC points were introduced when we had two shorter leagues and therefore there was a greater chance that an OPC might be prevented from completing a league because they only had say 6 instead of 7 events available to them. Now that we have 13 events the reduction in the number of events available to an OPC is meaningless. The other concern was that without a mechanism like this there would be a chilling effect on people\'s willingness to OPC. The group of people taking on OPC roles hasn\'t changed much since that time so I honestly don\'t think this worry is justified. Hands up anyone who has declined to OPC because they are concerned about not finishing the league as a result? We can better achieve the required results by allowing OPCs to count their average points for up to 2 events in a given league. This allows them to complete the league without skewing their results.

Another aspect of the NIOA league that\'s relevant is that the number of events is much less (best 4 out of 6) and therefore the impact of variability in points awarded across different events will be less than in our system where it\'s 6 out of 13.

Paul

Paul - yes i think a consistent mid field runner should get 1000 (or so) points, not because they\'ll feel put out if they don\'t, but because it makes scores more comparable across events. If my score fluctuates not because I had a particularly good or bad run, but because an elite competitor, a very slow finisher or a large crowd turn up, I see that as a flaw in the system.

Sure we could revert to a simpler system of 100 points for a win and so on down to 1 point for 100th place but I don\'t think it would be particularly meaningful for most except the leaders.

Quote from psmythirl on January 4, 2020, 13:19

Within a single events\' results for a single course, the metric of minutes behind the course winner is a perfectly meaningful comparison. There is no way to make that consistent across multiple events with varying fields etc. Unless you can provide a statistical model that can balance or control for these factors? Do you have such a model that we can critique? By the same token it cannot be used to explain away the variation in points awarded that we are discussing.

The statistical model currently used is simply the \'mean\'! By definition, it should be a consistent comparator across all events - provided there are enough numbers taking part. As Rory says, we keep coming back to this. The reason it works is because:

1) It\'s impossible to plan each event to have the exact same winning time for the exact same competitor;
2) Competitors can have good days and bad days;
3) Basing points purely on the \'winner\' means outliers have an even greater effect.

Using the mean attempts to balance these three factors. I fully agree it\'s not foolproof but at least it is fairly fair. It\'s not easy to possible the two models (i.e. points -v- winning order) as by definition they are completely different - and thanks for pulling me up on it 🙂

Quote from psmythirl on January 4, 2020, 13:19

Also, and I keep coming back to this - Why should the ranking of a female competitor depend on how the male runners on her course did? In particular why should it depend on the male runners who did better than her? I\'ve yet to hear any rationale for that viewpoint.

Rankings in the current system are based on the \'average\' competitor, not the winner. Thus, the ranking of a female competitor is based on ALL competitors of all genders across the field, not just the winner.

Considering that only a few of us are having this discussion, is there really any appetite to change the system? Are there specific examples of people that have been \'hard done by\' by the current system?

OPC points is a whole new can of worms and best left for a separate forum thread I think! Whatever approach is taken is highly subjective.

Stuart