I taught myself how to beat Treasure Bot (our name for this effect, to distinguish from the wait-for-platinum strategy Prosperity introduces) by having simple rules which -- while not on their own necessarily better than treasure bot, were useful for developing the "feel" of a good deck.
(The zeroth rule: if you see a way to buy a province, do so. This is called the Goose Rule, and being able to get a province is called "having the Goose".)
The first rule: on the first two turns, you're likely to get a 3/4 split. unless your 4 is going to supply you with money, or either card will trash your junk, your 3 really should be silver. (And likely anyway; remember that a silver / chapel start is really good, even though they're both cheap.)
The second rule: if you had a terminal action (an action that doesn't give you +Actions) in hand that you did not play because you were out of actions, do not buy a terminal.
The third rule: if you had unused actions, do not buy a Village. Usually, don't buy anything that gives +multiple actions -- the other benefits are not as good as what you ought to be able to get elsewhere. (Having a bunch of action splitters and no terminals is called the Wolf Trap.)
The fourth rule: is the card you're thinking of buying better than three of the cards in your opening hand? In other words, were there three cards in your hand that, had you had this card instead, the turn would have been better?
The fifth rule: When 5 of the provinces are gone, start buying duchies at every opportunity. When 9 are gone, buy all the green you can.
The sixth rule: don't buy more than one of any given cost 2 card, 2 of any given cost 3, or 3 of any given cost 4 until you find yourself wishing you had it in a hand you didn't.
As you play, you'll learn when to ignore these rules or make exceptions; the fifth rule might get some caveats about whether or not you're winning, or if you're about to reshuffle, for instance. I don't follow these rules any more, because they have taught me what I needed to know.
A big part of better-than-treasure-bot play is designing your strategy. Look at the cards available. Figure out what combinations are possible, and what work at cross-purposes. Some cards help you make your deck smaller (by getting rid of bad cards via trashing, e.g. chapel, or handing off, e.g. ambassador.) Some cards help you improve the cards you have (by trashing and replacing, e.g. upgrade or remodel). Some cards help you get past cards you don't want in hand (by drawing and discarding them, e.g. warehouse or oasis). These are all valid tactics, but they mostly work at odds -- if you trash your useless stuff, you won't have much to improve or get past; usually, pick one of these and go with it. Similarly, some decks work by always being pretty good; others work by sometimes being terrible and sometimes being great. think about how the cards work together, or don't. In general, if you wouldn't want both cards in hand together, don't buy one of them. And if your strategy relies on getting both cards together, think about whether it's really the right strategy, or work extra hard to make it happen. (Consider that getting two treasure maps together is rare enough that it's worth 4 gold on top of your deck and isn't over-powered. BTW, if you're going to try to make treasure maps work, you need a lot of trashing and/or 3 or more maps.)