there is no better rating service than the betting public.
In 16265 races all venues all tracks,all weather,all distances and 7-14 runners the fav won 5151 times 2nd fav 3098 3rd fav 2216 4th fav 1670 and the 5th fav won 1247 times. This is 82.28% win rate. The fav win rate is 31.67%. But as more Joe Public has got it correct the win rate changes. at $1.50 it is 62%, at $2 it is 42%, at $3 29%, at $4 20.8% and down to 14% at $5. But you follow Joe Public then you just give your hard earnt away.
In a straight trifecta with the first three favorites you should get a 2% strike rate. It is then pot luck whether you get any value ie 2 x $50 divvie or a $100 divvie. If you box the first three then you increase your strike rate to about 7% at the cost of increasing your outlay 6 times. If you drop the favorite and box the next top three then you halve the strike rate but increase the divvie. but again you have increased your outlay by 6 times.
As you go wider you increase your chances and increase your outlay. Most systems I have seen on exotics show one or two real big divvie that they snared. These biggie's distort the reality and should be treated as the serendipity that they are.

Reframing my query. If a reliable rating system was used instead of market to select the top three as a straight trifecta, then it should follow that the about 2% strike rate would also be consistant. Say the likes of Don Scott's ratings which are fact based and not emotion based as the market is. These ratings do not always follow the market and therefore a consistant reliable stream of selections that are not overbet, is available. Should these selections follow a 2% strike rate as a straight trifecta? Ditto the neurals or any rating service. Short of a trial run is there a quick way to check this out?

Regards Beton