I don't see how this is mathematically profitable to the average investor. 

For example, Primedice (according to dicesites.com) has 2,772 BTC wagered through it's site since the end of May.  Extrapolated for a months time would be about ~12,000 BTC or approximately $34 million a month. 

With a house edge of 1% you would have an expected profit value of $340,000 a month, which is a significant amount of money, if you own all of the tokens and get 100%. 

However, $5000 investment of ETH (~19 ETH as of today) would only yield 3800 tokens assuming you got it at the best price.  (200 tokens / 1 ETH first hour). 

Lets assume all 5 million tokens are sold.
3800 tokens is only 0.076% of the total 5 million available tokens (about $6.5 million at first hour price).

So if it was as popular as Primedice, 0.076% [of $340,000] would only net you ~$258 a month. 

That is a full 19 months before you even make back your original investment of $5000!

Don't forget this is all under the assumption that idice would even become as popular as Primedice (been around 4 years) and would also do so instantly.