Racing Trend Analysis gone wrong

Just recently I received an email from someone attempting to analyze draw biases for Goodwood.

On the surface his racing trend analysis seemed sound and no doubt he thought so.

He cheerfully announced “from 2005 to date there have been 27 races with 11 or more runners contested over the 5 furlong straight and it’s clear that the higher a horse is drawn in a big-field 5 furlong race the more advantaged it is”.

The problem with his analysis is that it assumes that all horses are equal. He made no attempt to correlate the chances of a particular horse winning with their draw position.

For example here is one of his analysis:

5f Wins %
Low 3rd – 6 wins – 22.2%
Mid 3rd – 9 wins – 33.3%
High 3rd – 12 wins – 44.4%
Total 27 races

Would the analysis change if the horses in the top 3rd draws, as a group, on average, had a 70% chance of winning.

One could conclude in such circumstances that there was a distinct disadvantage in a high draw.

In the end his analysis may prove correct without considering the chances of a particular horse winning the race, however his conclusions are based on faulty analysis.