Rankings

[iainwp]

(note I'm new to nopesport forums, so this could have been discussed in great detail)

I seem to have jumped 150 places in the ranking tables without doing any ranking events recently . Now this is obviously because people above me have had some of their six ranking events expire and therefore have fewer points in total than last week. This time, last year was the middle of the Scottish 6, so I might expect more places next week, but by then the Lakes 5 results will be kicking in, so I might not.

My hypothesis, and that of others I've talked to informally, is that ranking points are more plentiful at bigger events, at least from a statistical perspective. (Everyone will have a single event counter example, such as when one of the best runners ran off the map, but still completed and came last.)

What do people think?

(being an an M45, but relatively novice, orienteer means that I use ranking points as a measure of my ability to complete a course, so they're quite significant in deciding whether I had a good run)

[Homer]

What do people think?

(being an an M45, but relatively novice, orienteer means that I use ranking points as a measure of my ability to complete a course, so they're quite significant in deciding whether I had a good run)

I suggest finding a better gauge of whether you've had a good run or not.

[greywolf]

My hypothesis, and that of others I've talked to informally, is that ranking points are more plentiful at bigger events, at least from a statistical perspective.

In the trivial sense that the aggregate of ranking points scored per event is directly related to the number of finishers yes, otherwise no

[Gnitworp]

Is it fair to say that the smaller the field the more likely you are to get an anomalously advantageous or disadvantageous resultant score because of the smaller statistical sample?

[greywolf]

Is it fair to say that the smaller the field the more likely you are to get an anomalously advantageous or disadvantageous resultant score because of the smaller statistical sample?

Because the system uses previous rankings scores as the basis for the total points available per course and the distribution of these points across the various finishing times, anomalous scores can occur when competitors do much better or worse than expected on the basis of previous scores.

So yes, the effect of a single competitor doing much better/worse than expected will be greater in a smaller field.

The most significant anomalies are the result of either a) clubs uploading incorrect data - wrong BOF numbers or including n/c runners or b) novices making very significant improvements in their first few events.

NB a v good orienteer having an off day or taking it easy going round 10% slower has relatively little impact, and a average orienteer who has a nightmare and takes 3 hours will almost certainly be trimmed as an outlier and not affect the scores at all - but a novice who finishes an hour behind everyone else at their first ranking event and then comes back a month later after a bit of coaching and finishes mid-field can have an very significant impact - this is a flaw in the algorithm for which a fix has been proposed but has yet to be implemented.

[canol]

I have tended to find that the physically tougher an area the better ranking score I seem to get as it slows the fast boys and girls down while I can maintain my slow plod. SWOC event at Llantrisant Forest is an example, proper Welsh forest which spat me out after hours in the fight then later let me have my best ranking score of the year. Other factors might have come into play also such as just enough ranked competitors to generate a score.

On a related matter I have run the longest course at some events such as Mens Open in urban events and then not generated a ranking as there were insufficient competitors on that course. Seems a pity that some way can't exist to combine courses when this happens.

[pete.owens]

The most anomalous ranking scores tend to be where the results are tightly clustered leading to a large number of points per minute. This leads to very high scores at the top end and very low scores at the bottom.

This is an example of a course with 31 ranked runners:
http://www.britishorienteering.org.uk/i ... &course=3&

[graeme]

Each week, all the points get multiplied by a number slightly larger than 1. So the longer you've had your points, the more they're worth, until they expire after a year. So if you aren't counting last summer's 6-day scores, expect to leap up.

[DJM]

The most anomalous ranking scores tend to be where the results are tightly clustered leading to a large number of points per minute. This leads to very high scores at the top end and very low scores at the bottom.

This is an example of a course with 31 ranked runners:
http://www.britishorienteering.org.uk/i ... &course=3&

This anomaly is actually nothing to do with results being tightly clustered but is the best known example of greywolf's:

a novice who finishes an hour behind everyone else at their first ranking event and then comes back a month later after a bit of coaching and finishes mid-field can have an very significant impact - this is a flaw in the algorithm for which a fix has been proposed but has yet to be implemented

[greywolf]

The most anomalous ranking scores tend to be where the results are tightly clustered leading to a large number of points per minute. This leads to very high scores at the top end and very low scores at the bottom.
This is an example of a course with 31 ranked runners:
http://www.britishorienteering.org.uk/i ... &course=3&

That course has been noted as a significant anomaly, but as DJM says the explanation for it is a combination of 1 unexpected performance and club upload error, rather than "the results are highly clustered".

Paul Simpson (MDOC M45) finished in 22nd running 89.46 for 8.2km scoring 1046 points - looks like a perfectly reasonable result. However his only previous ranking score was from an EPOC event a couple of weeks before, where he was dead last on Green taking 84 minutes for 3.6k and scoring 550 points
http://www.britishorienteering.org.uk/i ... &course=3&

The algorithm calculates the mean and standarad deviation of ranked runners' previous mean scores (after trimming off the last few as outliers) and uses these as the basis for the distribution of points based on finishing times. In this case there were 26 runners with previous mean scores in the range 1080 - 1300-ish and one completely anomalous score of 550 which slightly reduces the mean and massively increases the standard deviation - so giving excessively high and low points for those at the front and back of the field respectively.

If you look at the official results for the Storthes event on the EPOC website http://www.eastpennineoc.org.uk/2013eve ... green.html it says:
56th Paul Simpson +Fray Harker MDOC M40 84:00
which suggests that he went round that event in company and therefore should have been marked as uncompetitive when EPOC uploaded the results to the BOF site. If this had been done he wouldn't have received any ranking points for Storthes and so he wouldn't have had a previous score to be included in the calculations for Delamere, where the winner would have scored 1310 ish points...

[greywolf]

Each week, all the points get multiplied by a number slightly larger than 1. So the longer you've had your points, the more they're worth, until they expire after a year. So if you aren't counting last summer's 6-day scores, expect to leap up.

The overall trend is inflationary, but every now and then the multiplier is less than 1. And maybe the bigger contribution to the upward movement isn't the relative value of recent and old points, but that removing the S6D means there's loads of folk who now don't have 6 ranking events at all and plummet: Matt Speake, Rhys Findlay Robinson, Alan Cherry, Matt Halliday & Ian Embury have all disappeared from the top 50 (and are now down around 2500 with only 4 ranking scores)

[iainwp]

I suggest finding a better gauge of whether you've had a good run or not.

suggestions wrlcome ...

[Homer]

I suggest finding a better gauge of whether you've had a good run or not.

suggestions wrlcome ...

Analyse your split times for each leg. Look at your time compared with those running at about the same speed. You should be able to work out roughly how much time you lost. Total time lost is a good measure of how good your run was.
This will also help you address your weaknesses and give you ideas on what to do to improve. Good luck!

[canol]

I still recall years ago catching someone on my course and then for the remaining controls he always seemed to arrive at the same time at the remaining controls as me despite me running between controls and him walking. After some thought I learnt that the headless chicken approach is less successful than the orienteering approach he was taking and from then on I always made sure I knew where I am going before I set off.

[graeme]

.. compared with those running at about the same speed. ...

I've been practising going slower so I don't have to compare with the likes of you. Its good to know I'm getting better despite what the results say