As a theorist, I can muse about empirical issues generally safe from the fear that I might seriously explore them. Next summer is another Olympic year of competition and commentary. Consider the 100M dash. Victory spells fame and fortune, second place historical obscurity. Effort expended is real, and competitors can roughly see how their nearest rivals are doing in real time, although admittedly in a bit of a blur. To what extent then can we understand behavior in these races using auction theory? If we regress winning times on times of predecessors, is the second fastest time the best predictor, and does it obviate the power in the other order statistics? And is this more true in the 10,000M run, where events are less of a blur, or less true, because at some point the race is often a foregone conclusion?
Another prediction of auction theory is that the best times should be more clustered in head to head race, for instance, than if we just asked runners to race alone, not knowing their rivals’ times, and then picked the fastest time.
Cheap Talk 
2 comments
Comments feed for this article
June 22, 2011 at 2:31 pm
Isaac
Stage races in cycling actually offer a nice example of this, since there are individual time trials as well as head to head stages.
The problem is that because of benefits of drafting there are other mechanical reasons why finish times cluster more in the stage races than the time trials. Though mountain stages offer less benefit of drafting and finish times tend to be more spread out.
Also, in individual time trials the previous finish times are not hidden from the current rider, and the best riders go last…
June 22, 2011 at 4:31 pm
Jason Kerwin
Usain Bolt’s showboating aside, participants in the 100m final tend to focus more on Olympic and World records than winning the race. I would expect to pick this up in the qualifying heats, but using the 3rd-place time instead of 2nd (since typically the top 2 move on).