in: Orienteering; General

Mar 10, 2003 8:55 PM
#

I went through my results from last year looking for some trends in the data. With 34 events total, it seemed like a reasonable sample size to draw some conclusions. I initially calculated my adjusted speed using the standard 1m climb = 10m distance (the same formula computed by Attackpoint.) When I ranked the events, however, I noticed that all the hilly events sorted to the top and some flat events where I felt I had done quite well were far down the list. This made me think that 10:1 is perhaps not a very good rule of thumb. I ran a regression on my results and got a slope of 5! There are several explanations as to why this would be so different from conventional wisdom:

1) I'm a really, really good hill runner. I think we can safely discard this one without too much discussion.

2) My ROGAINE and adventure race results are skewing the data. I tossed them - no change.

3) I make more mistakes on flat courses. This sounds reasonable enough, especially considering how hilly my training areas are. I reran the regression after subtracting out an estimate of my total error for each course. This estimate was recorded at the time of the event and, while obviously subjective, I think it's a reasonable value. The resulting slope was still around 5.

4) Flatter areas tend to have thicker vegetation. I looked at the maps and gave them qualitative ratings of 1-5 for vegetation density. I then ran the regression with 2 independent variables (still using the error-free result as dependent) and again came out with a slope of around 5. Incidentally, the resulting R-squared was .88, which is a very good fit for biological data.

5) 10:1 is total bogus developed by people who don't like to run hills. Having no other explanations, I'm left with this one. Anybody see something I missed or should we ask Kenny to revise the formula?

1) I'm a really, really good hill runner. I think we can safely discard this one without too much discussion.

2) My ROGAINE and adventure race results are skewing the data. I tossed them - no change.

3) I make more mistakes on flat courses. This sounds reasonable enough, especially considering how hilly my training areas are. I reran the regression after subtracting out an estimate of my total error for each course. This estimate was recorded at the time of the event and, while obviously subjective, I think it's a reasonable value. The resulting slope was still around 5.

4) Flatter areas tend to have thicker vegetation. I looked at the maps and gave them qualitative ratings of 1-5 for vegetation density. I then ran the regression with 2 independent variables (still using the error-free result as dependent) and again came out with a slope of around 5. Incidentally, the resulting R-squared was .88, which is a very good fit for biological data.

5) 10:1 is total bogus developed by people who don't like to run hills. Having no other explanations, I'm left with this one. Anybody see something I missed or should we ask Kenny to revise the formula?

Mar 10, 2003 9:39 PM
#

I would tend to believe Eric's analysis. I haven't done a similar analysis in the past. Nothing more to add.

Mar 11, 2003 1:05 AM
#

ken:

I think I've read that a few different places. here's one source. don't know why I chose 10 (maybe it was round...or because matthias was using 10)

Mar 11, 2003 1:14 AM
#

Looking at the source Kenny quotes (well, the abstract—I don't have that particular issue), an explanation immediately comes to mind for the discrepancy. If fell (mountain) race organizers consistently underestimate the climb (which would have been the case until altimeters with good accuracy became available a few years ago), then one would need to muptiply their climb numbers by a larger factor to get a good fit.

Mar 11, 2003 1:34 AM
#

ken:

[more discussion] right...so I should at least change it to 8 now, or make it an option. perhaps choices of 5-8-10?

Mar 11, 2003 3:49 AM
#

This is testing my memory, but the number that I think I used back around WOC 93 was 4. This was arrived at by looking at the data, pulling a number out of the air, and seeing if it fit. The analysis that you guys are doing is far superior, and I'm glad to see that somebody is debunking this long standing(?) 10: 1 rule, which I always thought was way off.

Mar 11, 2003 4:20 AM
#

I think that you need to look at your results on a split-by-split basis. By using just the climb number for an entire course, you are averaging in a bunch of downhill running with your climbing. Try comparing legs where it was all uphill vs flat legs vs all downhill legs.

Mar 11, 2003 8:58 PM
#

I've done some of the analysis that Jeff suggests (although it has all been exploratory - looking at individually chosen legs rather than a more rigorously pre-defined sample). I've found that my downhill legs aren't significantly faster than flat legs, unless they are on trail (exceptions exist, of course). Thus, I think just looking at total climb over a course is reasonable. The 5:1 rule has held up pretty well for the uphill legs I've applied it to.

Mar 11, 2003 9:02 PM
#

A problem with using fell running data, at least with the level of technical expertise generally seen at the A-level in the US, is that few of us can run full speed on the flats/downhills and stay in contact with the map. On hills, one can generally push hard and still read the map. Therefore, the effect of hills is mitigated versus the effect it would have without navigating.

Mar 13, 2003 8:30 PM
#

feet:

A quick calculation to suggest that somewhere around 5 is correct for exceptional athletes:

Mount Washington road race: 7.6 miles (=12.228km), 4650 feet climb (=1417m), record 58:20. World record for 20km on the track: 56:55. The standard conversion suggests that is the equivalent of 20.47km in 58:20. Hence

12.228 + 1.417x = 20.47

or x around 5.8.

Since the Mt Washington race record would probably be lower if the world's real best ran the course, 5.8 is probably upward biased. That suggests around 5:1 as a conversion for real athletes (not orienteers like you and me necessarily). I suspect the conversion factor is different for less fit people and may also be nonlinear (is 100km + 3km climb = 100 * (1km + 30m climb) ?)

Mount Washington road race: 7.6 miles (=12.228km), 4650 feet climb (=1417m), record 58:20. World record for 20km on the track: 56:55. The standard conversion suggests that is the equivalent of 20.47km in 58:20. Hence

12.228 + 1.417x = 20.47

or x around 5.8.

Since the Mt Washington race record would probably be lower if the world's real best ran the course, 5.8 is probably upward biased. That suggests around 5:1 as a conversion for real athletes (not orienteers like you and me necessarily). I suspect the conversion factor is different for less fit people and may also be nonlinear (is 100km + 3km climb = 100 * (1km + 30m climb) ?)

Mar 15, 2003 1:14 AM
#

Here's what I meant by the bias in using fell running data:

Take your average Blue runner who is fit enough to run though flat, white woods at 5:00/km. Suppose for a moment that the 8:1 rule is correct. Add 100m of steady climb to the kilometer and this person now requires 9:00 to complete the K.

However, suppose that, in order to stay in contact with the map, this same runner can't go faster than 6:00/km. The uphill K will still take 9:00 (no reason to slow down just when you're already below your navigation speed), but the flat K now takes 6:00, so the climb has added only 3:00 which equals the effect of 500m of flat navigation, rather than 800.

Thus, slower navigation tends to reduce the effect of hills.

The point is that there is more than physiology limiting the speeds in orienteering. A rule derived from a form of racing without such constraints will yeild dubious results.

Take your average Blue runner who is fit enough to run though flat, white woods at 5:00/km. Suppose for a moment that the 8:1 rule is correct. Add 100m of steady climb to the kilometer and this person now requires 9:00 to complete the K.

However, suppose that, in order to stay in contact with the map, this same runner can't go faster than 6:00/km. The uphill K will still take 9:00 (no reason to slow down just when you're already below your navigation speed), but the flat K now takes 6:00, so the climb has added only 3:00 which equals the effect of 500m of flat navigation, rather than 800.

Thus, slower navigation tends to reduce the effect of hills.

The point is that there is more than physiology limiting the speeds in orienteering. A rule derived from a form of racing without such constraints will yeild dubious results.

Aug 8, 2003 8:47 PM
#

I'll add my 2 cents from recent training. I can do 8 flights > 168 steps > max energy burn (in other words I can't take another step) > 68 seconds. So I go out and sprint for 68 seconds as to achieve max energy burn simulataneously and I go 180 meters. As you can tell, I'm not a world class athlete. For that rough one minute, under controlled surface to run on, with no distraction of map reading) it is ratio of 1 to 4.89.

This discussion thread is closed.