J,
Sartin's thesis was/is: speed = energy = class. And some would say that using adjusted Total Energy (TE) alone (though I use the corollary of Perceptor/Primary Line Score) you get a Win % in the golden ratio range (~62%) from the top 3 and that this is sufficient measure of 'class' to nudge you towards breakeven or positive ROI.
Perhaps the 'cheap speed' you observed could be discounted by other means - are you able to cite an example (e.g. from the Demo Database?) Of course, single instances will always occur, and sometimes 'local knowledge' does the trick (ask me how I know to discount Ft. Erie speedballs at Woodbine
). But this is a reason I urge caution using CR (APV, consistency) as a measure of 'class'. Sartin had dropped CR/APV after the mid-1990s for much the same reasons I'm urging caution - it suppressed more obscure horses with 'actual class' represented by energy measures and disbursement relative to today's competition. I reintroduced it as much as a measure of 'back class' or a rationale to seek a line further back, as anything else.
On re-reading your original post, I think you may be asking if one risks much (or have studies been done) by eliminating from Win the bottom half (or so) of the CR/APV range. Interesting study, now that I think of it. Compound that with Total Energy or the VDC rank and you could perhaps be on to something.
Perhaps the closest I can come to advice re the keeping/eliminating horses based on CR, is when considering eliminating a 6th ranked horse relatively tied on other counts with a 5th ranked horse, I'd err on the side of CR. Or, horse has high TE, high CR but layed off 180+ days (non-Stakes): I'll move it out of the Win slot but consider keeping for ITM. Or, the high TE and high CR rating comes from a few months ago and I'm trying to decide if I should go back those 4 lines. Something like that.
Examples might help.
Ted