PRICE ROI
4-6 -1.9%
8-11 -2.3%
4-5 -9.4%
5-6 -14.1%
10-11 -2.3%
Evens -3.4%
11-10 -6.5%
6-5 -6.5%
5-4 +3.9%
11-8 -6.9%
6-4 -6.5%
------
snip
------
9-2 -12.1%
5-1 -11.5%
6-1 -14.2%
7-1 -19.6%
8-1 -19.2%
------
snip
------
14-1 -33.8%
16-1 -37.7%
20-1 -37.3%
25-1 -47.3%

This is the so called 'favourite longshot bias' market inefficiency in all its glory. This is how you would do blindly backing all horses at these starting prices over the last 10 years. Ok, it's a UK example but it wont be hugely different for you. After all, the f/l bias was originally found in US pari-mutuel markets over 30 years ago. Currently, sane and serious rebaters channel their action in one direction only, on to favourites.

But as you tinker or hopefully root and branch prune your system, change will occur that will alter the dynamics of the market.
It is entirely predictible what will happen to the market if you do X or Y, only the extent of the change is debatable.

For instance if a 20% take actually did result in a loss on favs of -7% and on longshots of -33%, would a 10% take result in a loss on favs of -3.5% and on longshots -16.5% ie a linear change, or would it be something else. I'd say 'else' and in a specific direction, what I can't say is by how much. Either way, 10% or lower would would be decent step in the right direction. But from here, it looks like tinkering is what the 'horsemen' have in mind, if anything.