I don't have enough computational skills to answer this question myself (i.e. I can't even figure out how to download the simulator on my ancient mac).
When playing chancellor/stash, my pseudo-algorithm is usually something like: avoid silvers, buy chancellors, buy stashes till you have 4, then prefer chancellors (unless it's duchy time), buy province whenever you have 8.
I'm not sure if this is ideal. With enough chancellors, chancellor/stash has a reasonable expectation of getting a province every other turn, and with luck, you draw a chancellor as the fifth card in your stash hand, thus getting two provinces in a row. (i've found the lucky draw happens about an average of once per game using the above buying rules, but this is just my ten or so IRL games, not simulation)
Suppose we only buy three stashes. On the one hand, some of our stash turns won't get a province, which is bad. On the other, the chance of having a chancellor in hand on the stash turn doubles, which is good.
So,
1. Simple question - For simplicity, ignore duchies. Which gets to five provinces faster? Four stashes (buy rules: province, stash w/ <8 unless there's 4 in deck already, chancellor) ? Or three stashes? Note that I expect four stashes to get to 8 provinces faster, but that this is the wrong question for practical applications.
2. Hard question - What's the best win % against BMU you can make buying only chancellors, stashes, and green cards?