I don't see why not: you take the average times of the Olympic mile dash of say 1948 and figure out each participants' time relative to that, then take that same overall average and compare it to the average time for the same event throughout Olympic history (modern, since we don't have good records of the ancient Greek games), and if you want a 1948 competitor to run against a 2012 competitor, do the same averaging for the 2012 times and the relative position of that 2012 competitor relative to the pack and then normalize both years to the overall historical average and you will get a sense of what the probability outcomes are.
Very thin example: 1948 competitor Bob was 3 seconds better than the average for the 1948 mile run. The overall 1948 average was 5 seconds slower than the overall average since 1800 whatever when we started having Olympics. 2012 competitor Bill was 3 seconds slower than the 2012 overall average speed but 2012 was 5 seconds faster than the historical average. So 2012's average is 10 seconds faster than that of 1948 and Bob is 2 seconds slower than the average historically which is 5 seconds slower than 2012's average but only 2 seconds slower than Bill. So Bill wins by 2 seconds as the most likely outcome. Or something like that.