[2008-11-11] OL: Analyzing Pack Dynamics in Orienteering

Note: female form is used / male form implicit / competitors comments on the calculated results are highly welcome!

New: WC France 2010 Men and WC France 2010 Women

Extra-information during a competition

An orienteerers performance is the result of a combination of her physical and mental skills. She permanently has to extract, combine and interpret the relevant information from map and terrain. She has to evaluate and to decide which route to choose and last but not least she has to run it.

Ideally she is supposed to do it alone.

Liisa Anttila and Kathryn Ewels at the 12th Control (No.48) of the WOC 2008 Long Final Women. Series © by Petr Kaderávek.

In reality there are more or less runners on the same course and every time two runners meet in the forest, they provide each other with a sort of extra-information (without even talking).

This extra-information can effect multi-plane benefits. A rough listing here

• Mind: objectivation of both runners performances (time won/lost until this point of the course, significance of the pack partner) can effect in changed motivation
• Orienteering: Both runner can delegate the orienteering partly, mainly or entirely to their pack partner; route choices can be made dependend to the pack partners choice; mistakes can be detected earlier (if both orienteer by themselves); pilotage in the controls area;
• Running: mutual pacemaker effect; information on microroute choice

Pack-Analysis

Due to the electronic punching systems used these days, we know it, if two runners punch a control in sight of each other and we know it if they do it for two and more consecutive controls. So we are able to estimate the amount and the kind of effect these extra-informations generate. This estimation is done by a PHP-coded application we call PackFigures.

15 Seconds

As a thumb-rule we assume that 15 seconds / approx. 60 m is the limit where mutual influence is about to break off. We got this figure from inquiries like the following, where we analyzed the intervalls between two sequent punches at all the controls at WOC 2008 Long Final Men.

Fig.1. Punching intervalls of two following punches at the controls of WOC 2008 Long Final Men

We found an agglomeration effect in the first 15 seconds after the previous punch. A whole 22.57% of all punches in this competition happened not more than 15 seconds behind a preceding punch.

The Model

For our calculations we introduce two kinds of packs: If two (or more) competitors punch at a control not more than these 15 seconds apart we call it a Controlpack CP. If the same two runners also build a CP at the following control we talk of a Legpack LP.

In our model we take the following assumptions:

• Runners in LP ran the whole leg under mutual influence
• We renounce on estimating the time runners in CP influenced each other, splits under influence of CP are ingnored
• Legs without CP nor LP influence are assumed as solo performance

The basic model compares the competitors solo performance with her performance in LP both as a percentage of the top solo performance for the according legs.

Further we relate these to each other to calculate the positive or negative effect (the boost) of LP on a runners performance.

Finally we calculate the time won/lost in LP according to the runners boost and her time in LP (the gain) and we get an estimation of a competitors packfree running time.

Analysis

Here I link all the analysis made with PackFigures.

Select a PackFigures; WOC 99 Long Final Men WOC 03 Long Final Men WOC 03 Long Final Women WOC 03 Middle Final Men WOC 03 Middle Final Women WOC 05 Long Final Men WOC 06 Middle Final Men WOC 06 Middle Final Women WOC 06 Long Final Men WOC 06 Long Final Women WC Final 06 Long Women WC Final 06 Middle Women WC Final 06 Long Men WC Final 06 Middle Men WOC 07 Middle Final Women WOC 07 Long Final Men SM 07 Long Men EOC 08 Long Final Men EOC 08 Middle Final Men WOC 08 Long Final Women WOC 08 Long Final Men NOC 09 Long Final (all classes) NOC 09 Middle Final (all classes) WOC 09 Long Final Women WOC 09 Long Final Men WOC 09 Middle Final Women WOC 09 Middle Final Men WOC 09 Long Final Women EOC 10 Long Final Men EOC 10 Long Final WomenEOC 10 Middle Final Men EOC 10 Middle Final WomenEOC 10 Sprint Final Men EOC 10 Sprint Final Women WOC 10 Long Final Men WC France 10 Men WC France 10 Women

Note: Feel free to conact me if you could provide me with more intressting data

Discussion

By relating the runners speed with the top solo splits the model takes into account effects like holding up the speed on the last part of a long distance race. Instead the model is not capable to handle any mistakes shortened or not done due to packs, where made mistakes are taken into account. So there is a certain oddity.

Ignoring the CP-legs leeds to a conservative value for the gain: potential benefits from controlpacks are ignored. Legs after a CP leg not being classified as LP leg are handled as solo performance even if there might be a certain influence from the CP-partner.

The estimations of boost and gain can be unprecise, if a runners performance on the legs not taken into account (due to a CP) differs significantly from the rest of her performance or if there are to few solo splits to calculate a usable solo performance figure.

Non the less the calculated boosts are far from arbitrary. The general tendencies apear to fit very well to reality. Generally small misfits, no benefits, small benefits and big benefits can be found as results of packs. Where no benefits are rather seldom, the incidence of misfits depends on the courses difficulty (means the probability, that mistakes are made even in packs). The general tendency is that runner profit from packing, some more and some less depending on the packs constellation. At hierarchical packs (H-Packs) the benefit is distributed uneven among the pack runners. Where the leader usually gains only few percents the inferior runner benefits up to 20% of boost. At equal packs (E-Packs) where no real hierarchy is to be found both runner benefit substantially, because both play an active part in the pack. Their benefit with about less than 10% seems not as big relative to the one in H-Packs, but E-Packs have a big influence on the top ten results (See: WOC 1999 (Valstad-Berger, Bjorseth-Davidik), WOC 2003 Long Final Men (Mamleev-Omeltschenko), WOC 2005 Long Final Men (Khramov-Lauenstein), WC Final 2006 Long Men (Gueorgiou-Novikov-Lakanen), EOC 2008 Long Final Men (Hubmann-Johannson)).