I've worked with 2 different PhD friends of mine when they were providing expert witness testimony for cases similar to this (our product does something x% better than yous type things that someone got their balls sued off for).
If you had the accurate data (big assumption) the standard deviation from retail for the player sample and the non player sample would allow you to make a case for or against this being improper benefits.
NADA black book that the dealers use, for the month the car was purchased, far right hand column (ex clean). How much off that is each individual buyers order? All the single case good deal exceptions are taken into account by the average number.
If joe public std deviation is -$500 from retail and Joe Player's is -$650 then [censored] off, nothing to see here, go on about your bidness.
If Joe Public std dev is -$500 and Joe Player is -$1,500 then we have a problem. All the "yeah but's" would be baked into the numbers.
The other important thing I'd use to make a case is what % of the dealers business does this 50 some odd athletes represent? Is it a 100 car a month store or some big 500 plus dealership? If the players are more than say 20% of the total for the time period in question it gets harder to explain away I would think.
At least this is what seems to make sense to me for no more than I've looked at it. I haven't asked any actual expert witness/PhD types their thoughts.