High quality data is our passion

We strive to build models that are useful to the people deploying at-risk capital in the real world. Consequently, the inner core of the SmartRisk modeling philosophy is empirical evidence. The measuring stick of our model’s success is whether it leads to wise decision making, not whether it is elegant, easy to compute or popular.

Financial reality is characterized by three stylized facts:

  • The normal Gaussian distribution describes financial returns rather poorly and can lead to a severe under-estimation of risk.
  • Assuming a fixed co-movement structure between assets is not only potentially dangerous in practice, but also unnecessary given today’s available computing power.
  • Rare, high-impact, un-expectable events (a.k.a. “Black Swan events”) will continue to happen.

Covisum's tools and modeling approach are designed to better equip the practitioner to deal with this reality.

From a systems standpoint, we have built a scalable cloud-based calculator that handles the number crunching for our users, so they don’t have to. They can simply connect to the risk calculator and leverage its capabilities to power their risk application needs. One can now simply ‘outsource’ their risk calculations by connecting to our systems.

From a modeling point of view, SmartRisk has implemented advanced proprietary models that are absolutely capable of handling heavy-tailed distributions, and are robust enough to be applied to tens of thousands of securities in an unsupervised automated fashion.

Let’s look closer at how the three facts above affect the validity of classic risk measures: Standard Deviation, Beta, Alpha, and the Sharpe Ratio

The most widely taught risk measures of portfolio risks are as follows:

  1. Standard Deviation

A ubiquitous risk measure, the standard deviation is usually presented in percentage form and in annualized format (e.g. 12% p.a.). This statistic doesn’t really mean much to retail investors, and has been shown to be rather poorly understood by practitioners in the real world, as well.

Under the assumption that financial returns are normally distributed, one would expect an investment price to be within 2 standard deviations of the current price with 95% confidence, and within 3 standard deviations with roughly 99.75% confidence. Said differently, under the assumption of a Gaussian distribution, one would expect the value of an investment with a standard deviation of 12%, to be somewhere within +/- 36% of its current value in one year.

In the real world however, financial returns are not normally distributed, not even closely. It thus follows that knowing the standard deviation does not translate into an estimate of potential losses, or risk, since the relation between the number of standard deviations and the confidence interval has broken down.

  1. BETA

Mr. Warren Buffet said it better than we ever could:

“The riskiness of an investment is not measured by beta (a Wall Street term encompassing volatility and often used in measuring risk) but rather by the probability – the reasoned probability – of that investment causing its owner a loss of purchasing-power over his contemplated holding period.”

 There are two main problems with beta as a risk measure.

First, it is of little use for the short-term horizon.  Everyday, virtually every stock, ETF or mutual fund moves either more or less than the day’s market index percentage move multiplied by their respective beta.  Everyday one can find thousands of examples of positive beta stocks and mutual funds that move in the opposite direction of the day’s market move.  This is to be expected as Beta is supposed to measure an average response to a change in index value.

Secondly, and more problematically, beta is also of little value in the long-term, as its value will fluctuate dramatically with the passage of time.  For example, over its lifetime IBM’s beta has ranged from 1.24 to -0.46 (it is currently at 0.69).

Said differently, if an investment has a beta of 0.5, it means (roughly) that on average, it will fluctuate about half as much as the market, but it will do so over an unknown period of time.

  1. ALPHA 

Alpha is supposed to measure the average over or under performance of an investment compared to the market. Unfortunately, since the value of an investment’s alpha depends directly on the value of its beta, it also directly suffers from the same drawbacks and shortcomings as a risk measure as does beta.

  1. Sharpe Ratio

The Sharpe Ratio is the original risk reward coefficient. It is calculated as an investment excess return (over the “risk-free rate”) divided by its standard deviation. The main problem with this measure is that it confuses risk and standard deviation, through its assumption of normality of returns.

Get a 10-day free trial