Estimating Capacity and Order Size for Low-Frequency Equity Strategies
Summary
The document considers how to estimate the assets under management a low-frequency equity strategy can support and how to size its orders. It raises average daily volume as a possible liquidity yardstick, while noting that slippage, market impact, and information leakage from displayed or reserve limit orders can constrain execution. Position-sizing rules such as Kelly or fixed-ratio methods may also affect trade size and overall capacity.
The response recommends using a strategy simulator with market data to examine how increased size interacts with liquidity, and says this can provide a minimum-friction estimate when taking liquidity. Providing liquidity is harder to model. Market impact and adverse selection also need consideration, but the response gives no formula, quantitative threshold, or general capacity rule. A credible estimate therefore depends on the strategy’s edge and how execution costs erode it.
Key ideas
- A fixed percentage of average daily volume is only an initial way to think about capacity.
- Order size and capacity depend on slippage, market impact, adverse selection, and information leakage.
- A market-data simulator can estimate liquidity-related effects as strategy size increases.
- Taking liquidity is easier to simulate than providing it, and the former gives a minimum-friction estimate.
- Capacity depends on how much execution costs consume the strategy’s edge.
Tags
Full text
# Determining maximum strategy capacity and optimal order size for low frequency equity strategy # Determining maximum strategy capacity and optimal order size for low frequency equity strategy I have developed a low frequency equity trading strategy that seems to work well with stocks in the S&P 500. Someone asked me about the maximum capacity of the strategy (how much AUM I could handle), and that led me to think about how best to determine how much of each stock I could trade, with the considerations of slippage, market impact, and creating a footprint in the market with a large order in mind. My initial thought was to assume that I could safely trade (that is avoid undue cost and visibility) a small percentage of each stock's ADV, so the maximum capacity of the strategy would be the dollar value of the sum of x % of ADV of all of the stocks. If each order was then roughly say <1% of the stock's ADV, I would probably enter these orders as limit orders with a minimum display size and the rest in reserve, with the expectation that the market will work its way through my limit. While this would minimize slippage, I am concerned that predatory systems could pick up on the reserve quantity, and that that will provide information. I could layer the order in at different price levels, but that could be expensive. Any thoughts on this are appreciated. I am also thinking about adding money management considerations to each order size (Kelly Consideration or Fixed Ratio), which may only further complicate the question of max capacity. Is there a general rule of thumb for determining capacity at a high level? Cheers ## Answer by BGR (score 1) https://quant.stackexchange.com/a/22796 First it would help to know some more details about what you mean by maximum capacity. However here are a few things to consider. - Do you have a simulator you use to simulate your strategy with market data? - If answer to above is yes then you can clearly see the linear impact due to market liquidity constraints for your increased size. - Now 2 is easy only if you take liquidity. If you actually provide liquidity it is more difficult. - That said 2 will atleast give you an estimate of the lower bound ( by lower I mean minimum friction). - you then also have to consider the market impact, adverse selection and all these are far more complicated. So to summarize you need to know the edge you have and how much of the edge gets eaten up by each of the successive steps above. Hope this was helpful.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.