I’m curious what others think is a satisfactory Sharpe Ratio for a portfolio with a significant # of positions, say 100+? Something with a large enough sample size to be an alternate index to the market-cap weighted indices. Your thoughts?
Anything above 1. Theoretically a Sharpe < 1 means too much risk for too little returns.
OK, but what would your threshold be for just 10 stocks? And if those 10 stocks aren’t enough to build an entire portfolio, or aren’t scalable enough, then what about for a much larger portfolio, where in exchange for more stability and more scalability, one runs the risk of becoming more like the market and therefore might have to accept a lower Sharpe ratio.
Personally I think a Sharpe Ratio > 1 for a portfolio of more than 100 stocks is pretty darn good - maybe bordering on asking too much over the long term and for a scalable portfolio - but curious what others here think.
The best investors with documented histories (ran mutual funds, hedge funds, trading firms, …) over 15-20 year time spans have Sharpe ratios close to .8. These investors ultimately either had to buy larger cap stocks or more positions as their size increased, which is similar to what you are describing. If you can maintain a Sharpe ratio equal to or over 1 over this time period then you will outperform all of them.
Scott
You have to look at:
a) the length of the period you’re looking at (over a 1-5 year period, people might have ridiculously high Sharpe Ratios, but not many do over 10 years), and
b) how an appropriate benchmark did over that same period.
Otherwise, not much meaning to it. If the Bench had a Sharpe of 1.5 in a bull run and you had a Sharpe of 1, that’s not very good. If the bench had a Sharpe of -.2 and you had a Sharpe of 0.2, that’s very good.
0.8 over 10 years, in live trading is amazing. You will likely own some yachts if you can do that in a liquid space.
For example -
Winton Capital management is a premier futures trader. Their Sharpe since launch in 1997 is .8 or so.
Just holding Berkshire Hathaway (BRK.B) has a 0.06 Sharpe since 1/1/1999 according to P123. Some guy named Buffet is running it.
Both people running these firms are billionaires.
For more examples:
PRWCX from TRowe Price is considered a great mutual fund - a leader - by others (I have no opinion). Over 10 years, Sharpe of 0.64.
http://finance.yahoo.com/q/rk?s=PRWCX+Risk
Vanguard Wellington fund is another. 10 year Sharpe of .66.
http://finance.yahoo.com/q/rk?s=VWELX+Risk
Or Royce Small Cap Fund (RYPNX). 10 year Sharpe of .4
I wouldn’t use any Sharpe ‘cut off’ to evaluate any individual strategy…but would look at both Sharpe and Sortino on a total manager level. If I could only use one, would use Sortino.
But, Sharpe in sim’s doesn’t mean much. How did you get to that Sharpe? What rules did you add to, for example, trim tail risk or reduce DD’s? What is the probability they will work out of sample.
mm123,
Sharpe gave a formula, you have different ways to use it, especially regarding the reference rate and the period. You can calculate the ratio on annual returns from 1/1 to 12/31, on returns by rebalancing period, by strategy or combined, etc…
My main account is invested on various ETFs and Stocks P123 models (only screen-based weekly rotations, no optimized parameter, no sell rules, no weight-based ranking, in short: no P123-dependent).
For the stocks part I have 14 positions based on 4 screens, all large caps (8 SP500, 6 Russell1000).
The Sharpe ratio is 2.16 on 15 years on simulated annual returns (total return year by year) with a 3% reference rate and a 0.2% slippage for costs, and the average annual total return (which is not the CAGR) about 39%. I would be satisfied if I could make half of that on the long term in the real world.
For a 100+ port, a Sharpe above 1 is possible with large caps using a combination of many sector-focused models. I did it, at least “in backtest”.
Thanks for all these comments, I’m in agreement.
What this all points to is the difficulty in reaching high risk-adjusted returns with real $$ invested, the context/environment in which we’re investing in, and the size of assets under management. Perhaps one possible conclusion … doubling the market’s risk-adjsuted return (Sharpe or Sortino), meaning by order of magnitude, is a worthy accomplishment for any investor, and may get tougher as more investors become data-driven.
I would add one more consideration that goes beyond my initial question … even the best investors face an ongoing tug-of-war between absolute returns and risk-adjusted returns. It’s what makes our pursuit here so challenging.
mm, If you increase your positions from 10 to 100 to increase the stability of your portfolio, its standard deviation will go down and your Sharpe will go up. You can use Sharpe to compare the trade-off between portfolio variance and portfolio performance as you increase the number of holdings. In that respect the usual number of 20 holdings wasn’t just made up but is often the best trade-off between performance and variance, as shown by the Sharpe ratio.
The standard deviation will not go down much lower beyond 20/25 positions. It might even go up again.
It is more important to diversify model logics than holdings number.
Generally in statistics a higher N will decrease the standard deviation. One reason there are limits to how much you can decrease the standard deviation with stocks is that the individual stocks are not independent. Diversification, as Fred says, can help.
Well these statements about standard deviation are true enough, but it seems to me this is only one-half of the equation … i.e., will the numerator (i.e., the portfolio return) hold at a constant level such that the Sharpe or Sortino Ratio will hold or increase? And there is the related question of whether the diversification is accomplished through similar securities (say, micro- or small-caps, which may or may not reduce standard deviation even as the # of positions increase).
We could go on and on about this topic … anyway good feedback. Thx.