Evaluating Prop Strategies Beyond the Sharpe Ratio
Summary
The document considers what makes an intraday quantitative strategy attractive to a proprietary trading desk, using close-to-close daily equity returns as its premise. One response treats a backtest Sharpe ratio around or above one as an encouraging starting point, then recommends evaluating additional measures: return relative to turnover, transaction costs, asset exposure, and drawdown. It also suggests viewing performance on a rolling basis to assess consistency rather than relying on a single full-period statistic.
Other responses distinguish live records from simulations and stress capacity, liquidity, and revenue potential, especially for high-turnover strategies. They note that firms may not accept backtests alone and that a large bank may prioritize gross profit at the desk level. These are practitioner opinions, not universal hiring thresholds or a formal selection framework. The document gives no underlying datasets or controlled evidence, and the appropriate metrics depend on the strategy, costs, scale, and firm's objectives.
Key ideas
- Sharpe is one measure of risk-adjusted return, not a complete evaluation of a strategy.
- Return relative to turnover, costs, exposure, and drawdown can reveal other practical risks.
- Rolling performance measures can help show whether returns persist across time.
- Live results, capacity, liquidity, and revenue potential influence a strategy's appeal.
- The comments provide practitioner views rather than universal acceptance criteria.
Tags
Full text
# What is an acceptable Sharpe Ratio for a prop desk? # What is an acceptable Sharpe Ratio for a prop desk? What should be the value of a Sharpe Ratio for an intraday quantitative strategy to be accepted by a bank or hedge fund's prop desk? Let's assume the returns are daily changes in account equity, close to close. ## Answer by mpeac (score 9, accepted) https://quant.stackexchange.com/a/21123 A Sharpe ratio of at least 1 in backtesting is a promising start, but that is just one of many statistics of interest. The Sharpe ratio measures return per unit volatility, i.e., return per unit risk. Some other important Sharpe-like measures with different definitions of risk include: - Return per unit turnover (aka yield): A high yielding strategy is more desirable in many ways, since it may also require less liquidity, have less market impact, and be less sensitive to transaction cost modeling errors or future regime changes in transaction costs. - Return per unit transaction costs: Similar to yield, this gives an indication as to how sensitive the strategy performance is to errors in the transaction cost model, and shows how much of an alpha buffer is available for costs which may be hard or difficult to capture in backtesting. - Return per unit max/avg asset exposure: A strategy that is more diversified across assets is less sensitive to event risk, and more likely to have a higher capacity. - Return per unit max drawdown: Drawdown is in many ways a better measure of risk compared to volatility. Day to day Gaussian-like volatility is not likely to wipe out the strategy, while a sustained period of losses will. The leverage of a strategy is likely to be backed out from the max drawdown, based on how much the fund is willing to lose. In addition, any of these numbers are usually not computed just for the entire backtest period, but are instead shown on a rolling basis, to show how consistent the performance is over time, i.e., the performance is not just due to a few lucky trades. ## Answer by MichaelJ (score 2) https://quant.stackexchange.com/a/21136 The required Sharpe ratio depends strongly on whether you are referring to actual profits or a simulation. For actual results, a Sharpe of 1+ over 12+ months is probably the minimum. If the strategy trades liquid markets at liquid times such that it can be scaled up to generate large revenues then, all else being equal, it is more attractive. This is assuming that by intraday you mean just a few trades a day. If it's a high-frequency strategy turning over hundreds or thousands of times per day then Sharpe will likely need to be above four. For high-frequency strategies, if the strategy works the Sharpe is often quite high, routinely above ten. Therefore, larger firms will scrutinize capacity. That is, the total amount of revenue that can be generated from the strategy. There isn't much practical difference between a Sharpe 10 and Sharpe 20 strategy is the latter can't generate any additional revenue. Simulations are a different matter. Many firms won't hire traders based purely on backtested results, regardless of Sharpe. To break in with a backtest you will need either (1) strong credentials, (2) a really good story to go along with the backtest or (3) experience at another firm that is somewhat successful. Keep in mind I am talking about respectable firms here. There are a lot of two bit firms out there that will hire anyone off the street and let them start trading with small limits. Those firms won't offer much in the way of training or infrastructure. ## Answer by madilyn (score 1) https://quant.stackexchange.com/a/21137 Uh-oh, wrong move. There's no point looking at Sharpe here. For a large bank, your job is generally to maximize the gross profit of your desk, because the firm achieves diversification at the operational level (across business units). It seems that you are thinking specifically about the Sharpe ratio of low latency prop trading. In those cases, it gets to a point that you have enough strategies that you just call it off as "infinite Sharpe" and start worrying about your expenses.
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.