Sure, all I'm saying is that it's not as useful to me. On my earnings strategy, I lose 71% of the time, or make 4x my average loss 29% of the time. My standard deviation is pretty big just by the nature of the strategy. That doesn't mean I'd ditch it because it has a low Sharpe ratio. Measuring against the risk-free rate should be an integral part of all trader's performance metrics, however. Particularly option traders (you're "paying for" risk free on every option anyway).
yes , he had a fund , so this info probably available (was) somewhere. I know that everyone will disagrees with me , but I cannot understand how retail trader can use someone's else apps to quantify his own performance. Of shelf apps - one fits it all ? hmmm Take a look how IB displays VaR on individual stocks long ( short ) combo. They are probably ( not sure and don't care) basing it on Beta. What about : bankruptcies buyout 2000 bp overnite drop in vols 5000 ( ! ) one day ramp in vols for bio's and court hearings. So an owner of short combo on XYZ should sleep well because IB's sophisticated VaR shows no more than 3.4% as a "worst case" ?
Portfolio daily P&L is not accurate because of the bid ask spread. Do you use the worst case, or mid as your daily P&L data? It deviates a lot. Say for example, after I just open a position, I will see a drop in my daily p&l because of the bid ask spread. Sharpe is good for a very diversified portfolio, and it is hard to achieve a sharpe ratio of 1 for a very concentrated portfolio. I believe sharpe ratio is a way to measure how well you diversify your porfolio.
I was really surprised to find it (from Taleb): "It's a scam, all of it -- Markowitz, Sharpe, it's just like astrology"
Do you mean a daily sharpe ratio of 1 or annual sharpe ratio of 1? There two differs by a multiple of around 15.
That shouldnât make any difference to âaccuracyâ because the spread after all, is part of trading life. It depends. If itâs a liquid option Iâll take the worst case, if itâs illiquid with wide spreads Iâll take the mid (knowing I can improve on the worst case). Yes you will see a drop, but as I said, thatâs life. The actual drop should be small, unless the new position is massive, in which case youâll want that to show up anyway. After all we are measuring a risk / return. Diversified or concentrated doesnât make any difference because you are measuring a portfolioâs risk adjusted return. You are normalising returns against their associated volatility. In other words, a diversified (less risky) portfolio, would need a larger return than would a concentrated (more risky) portfolio to achieve the same Sharpe. A lot of people think Taleb is a lunatic. I wouldnât go that far, but for all his rants on Sharpe, VaR etc, he has yet to come up with a better alternative ! Calculate a Sharpe using your daily PnL data. The actual time period is of your choosing, but I would suggest using a minimum of 3 months daily data.
You misunderstood my question. I understand you use the daily data. Do you annualize it, and hope the Sharpe ratio > 1? [edit] Using 6 months data, the ratio of 1-day excess mean return to 1-day standard deviation is around 0.1. So the annual Sharpe ratio is around 15*0.1 = 1.5. Since there are around 256 trading days, and 365 days for interest rate, I use 256 as a simplified way to compute the ratio. Do I have it right this time?
yip You take the sandard deviation of your daily returns. Then, since these are daily returns, you multiply it by the square root of 252 (maybe 256 for the yanks). This will then give you the "annualised" volatility.