Although I agree that limit orders don't offer the improved execution that they appear to, I disagree with the assertion that they can never offer improved execution.


For one, I don't buy the no-arbitrage argument that you can't benefit from a limit order because if there was a benefit, someone else would already have taken advantage of it. This is basically the old joke about the finance professor who sees a $100 bill lying on the ground but doesn't bother to pick it up because if it really was, someone else would have already picked it up. I know this isn't true from personal experience (and I know some people who have been big multiples on the wrong side of this "trade".) If $100 bills are being dropped, there's a chance you'll be the first one who can grab it. (Furthermore, the no arbitrage argument was decisively refuted by Sanford Grossman in his paper "On the Impossibility of Informationally Efficient Markets" back in 1980.)


The thing is, if you want to improve your execution, you're probably going to need to do your own research. It's kind of like deciding whether or not to play AJo under the gun: just because someone else loses heavily with the hand doesn't mean you should always fold; it's more important to know how well you play compared to your opponents from the flop on. Similarly, with limit orders vs. market orders, you need to consider your situation. Do you pay per share or per trade? Are you willing to adjust your order as the broad market moves (or would you prefer to place an order and leave it)? How does your broker operate?


There are many ways to buy (or attempt to buy) 1000 shares of a thinly traded stock. In addition to placing a market order for 1000, or a fill or kill for 1000, you can (a) submit a limit order above the offer, filling at least some of the order at the current ask, (b) break your order into pieces (you want to pay per share to do this), (c) duplicate the inside bid (if there's a chance you'll fill on a naive internal match), (d) bid below the inside bid at or just above a level where you think there's a lot of support, etc.


Ideally, you would like to do some actual experiments to determine which alternative is best, but, as with poker, you can at least try to apply logic to your knowledge and beliefs about the situation.