This was what Davio communicated in a post-match discussion with me, where I got 10 golds and he got 1 and he swept the floor with me.
However, I am not sure about this. Some lines of simulating codes in R follow.
Let's take two more or less typical Big Money decks, both with 4 coppers and 2 silvers. One deck has hoarded a boatload of 8 golds, the other one has worked towards 4 golds and 1 platin.
values.au <- c(rep(1, 4), rep(2, 2), rep(3, 8))
values.pt <- c(rep(1, 4), rep(2, 2), rep(3, 4), rep(5, 1))
Lets simulate hands of five with both of these and tally the sums:
table(replicate(9999, sum(sample(values.au, 5, replace=F))))
6 7 8 9 10 11 12 13 14 15
8 68 309 755 1696 2257 2239 1664 696 307
table(replicate(9999, sum(sample(values.pt, 5, replace=F))))
6 7 8 9 10 11 12 13 14 15 16 17
51 188 666 1087 1724 1538 1687 1252 1123 467 193 23
Obviously, the gold deck hits $11 more consistently than the platinum deck, although the average values are very close.
This simulation doesn't say too much about snowballing (the platinum deck will manage to draw more platinums freshly bought) and dilution (the platinum deck will be hit by colonies faster) but the platinum > gold argument seems to need a more solid foundation than a gut feeling.