Was listening to Andy S commenting on the exacta payout after the 3rd race at the Spa today. The 3-1 shot second choice finished over the Even money favorite, with the $2 Ex returning $19.20. Andy felt it should have paid around $12 and proclaimed that exactas with the 2nd or 3rd choice finishing over the favorite continues to offer the best value in racing.

My curiosity piqued, I did a quick check using discounted Harville to see if I could confirm his assertion. Turns out the expected payout was $18.90, pretty much right in line with the $19.20 actual payout. Even if you use straight conditional probability, which is inaccurate because it neglects to account for the more uniform probability distribution in the lower slots (ie. faves are less likely to finish 2nd or 3rd than their win odds indicate, while longshots are more likely to do so), the expected payout is still $15.65.

Can’t really draw any conclusions from one race, but this result doesn’t seem to support the assertion. Would be interesting to hear if anyone’s ever done any rigorous study on this, I haven't and I'm not aware of any, perhaps Barry Meadow in his latest book(?).