Hi,I am working on a research topic and am considering a strategy which is buy (sell) open and sell (close) for SPY. What aspects should I be considering while trying to estimate the capacity of this strategy (just on NYSE). What could be a good way to approximate the capacity.Thanks in advance.

You need a transaction cost model. A simple one using the half spread and a non-linear market impact term might suffice for you.

Thanks skip. I was wondering as my orders would basically be MOO and MOC, is it possible to make some crude approximations for capacity based on recent volumes. Is there a quick dirty way. Thanks.

There's a ton of literature on t-cost modeling. Some people make a career out of it. For an introductory discussion see Grinold & Kahn who present a simple model of transaciton costs. You only have to go a little further to operationalize their model. For something more advanced see maybe Kissell & Glantz, "Optimal Trading Strategies". Keep in mind a couple of things. Market-impact costs are positively related to the volatility of the security being traded, and stock volatility is U-shape intraday. So, that would tend to increase the cost of your strategy since you're trading at the open and close, which are the most volatile times during a normal trading day. MI is negatively related to your participation rate, and average intraday volumes are also U-shaped, which would tend to reduce your costs. MI is also positively related to the immediacy of your trades -- your trading aggressiveness -- which will be very high for any strategy that MUST enter its positions on the open and exit them on the close. The point of all of this is that any model you come up with, especially a simple one, is going to be an approximation, so just keep all of that in mind, of course. That said, I'd go with G&K's simple square-root model for your purposes. You could mitigate some of these concerns by localizing the model to the specific times of day that you'll be trading. What I mean by that is that you would use estimates of the expected volatility and volume during the first 15 minutes or so in your model of MI at the open, and the same during the last 15 minutes in your model of MI at the close. I don't have this data and don't know an inexpensive place to get it, but someone else might. Iteratively increase your trade volume (i.e., your participation rate) in the MI model to see how your net return changes. The optimal capacity (for any particular volatility and volume of the market) would be when the net return flattens out. You might find this paper helpful: http://www.itg.com/news_events/papers/C ... anding.pdf

Thank you skip!