Over a sample of 1,000 Saturday Metro meetings I would expect 493 EVEN money favourites to win based on past results.

So you would need a price of at least $2.03 (roughly) just to break even.

And I know you'll say that you only bet when you can obtain BETTER than evens - that's fine in principle, but do you have stats on horses that top fluctuation or likewise is better than evens?
Could possibly be that their stike rate is worse.
Even betting top fluctuation, only in a small amount of cases would you get a reasonable return, most cases you would probably get just evens.
Which makes it a bad bet whichever way you look at it.
The number of times you could get skinny overs on an even money chance would be outnumbered by the number of times you get evens.

In this example there would hypothetically be one even money chance in the race, but say at 4/1 6/1 etc there can be multiple runners which distorts the actual win ratio.

So you can have for example:

438 winners out of 1910 races which gives a percentage of 22.69%

BUT

438 winners out of 6318 runners which gives a win percentage of 6.93%

(Just a hypothetical example)

So you can see that the results are misleading.

I see where your coming from here, but if you use statistics it has to be broken down completely to give a correct indication.