vand wrote:hiriskpaul wrote:I thought I would add that volatility is not the only risk metric. Others are available. Value at risk (VaR) is a very commonly used metric. With VaR you calculate what would happen to your portfolio over the next day for a set of historical daily changes. You then line those up from best to worse performance and take the 97th percentile of the distribution. The loss is called the Value at Risk.
Another approach is to consider absolute movements. These tend to be more useful for portfolios that include financial derivatives, but the way it works is you calculate the loss according to specific events. eg how much would I lose if the S&P 500 dropped by 1% tomorrow? Or the risk free rate moved up 0.25%, or for portfolios with financial derivatives, what would happen if implied volatility spiked 5%. Sometimes you would have to model the impacts on particular portfolio constituents. For example if your portfolio has a higher than average proportion of of volatile shares you need to take that into account.
All this can get very involved and is hard if you don't have adequate data. Nonetheless, it is what banks etc. do to manage their risks.
Edit, for some securities volatility is a hopeless risk metric. I am thinking about very illiquid things here, such as mortgage backed securities, micro caps and non-listed shares. A market price, even if you can get one, may not move for weeks on end simply because there have been no trades and/or no MMs to move the price.
VaR just calculates the likely distribution of future loss based upon recent historic volatility to a certain confidence interval - it is still a metric that revolves around volatility.. because at the end of the day, what else do you have to determine risk, other than price data and the change in that timeseries from one time period to another?
Drawdown from peak to trough drawdown
length of drawdown
length x depth of drawdown
skew
Investment risk is about how (un)evenly the returns are distributed over time. With equities the multiples being paid for current and future earnings is always fluctuating wildly; there are time when the market is willing to pay 40 times earnings and other time when it will not pay 10 times - if you pay 40 times earnings and the next day or next year the market decides that it only wants to pay 10 times earnings, then it's going to take a very long time for your investment to pay for itself at the rate that most companies are able to grow their earnings.
There are of other forms of risk of course, but they are not considered investment risk - the behavioral risk that comes with holding an unsuitable portfolio (I would argue this is always the greatest risk any individual investor can face), default risk, inflation risk, geopolitical risk etc etc
A few good question to ask yourself about how much risk you want to take with your current portfolio:
- how would I feel if tomorrow we started a bear market that matched the worst bear market in history
- how well would I honestly rate my investor behaviour through all past bear markets I've been investing through
- Am I living off my portfolio? If not, how close am to retirement? The closer you are to living off your nestegg the less risk you can afford to take due to having less time to recover if portfolio suffers a large drawdown
- How likely is a strategy that worked well in the past likely to persist into the future? The worst thing you can do is to chase return
- What is the downside if I am wrong?
- Will I be happy to carry on working longer if future returns turn out towards the low end of possible outcomes
- Am I happy to adjust my lifestyle if my investments underperform
- How much will it bother me for the rest of my life if I make a big mistake and end up with a poor outcome
VaR goes beyond volatility as it considers actual historical price, interest rate, etc. movements which in reality are fat tailed, or in the case of volatility itself, spiky. A basic volatility calculation and related metrics such as Sharpe averages much of the complexity away and can understate risk. VaR also incorporates actually observed historical correlation in the market. By no means perfect, but a step up from volatility.