Where it is

ScreenRisk tab, Return distributionAdvanced

The Risk tab, in Advanced: the Return distribution card, and Skew and Excess kurtosis in the Risk figures card. On a portfolio both read the model since you invested.

Index in this video: Allcap (strategy page).

What it means

The histogram sorts every weekly return of the period into bars one percent wide, from -15% to +15% by default, refitted to the range the returns fill. Markers show the mean week, VaR 95 and VaR 99. Skew measures lopsidedness: below zero the long tail is on the losing side, above zero on the winning side. Excess kurtosis compares the tails with a normal curve: above zero means more extreme weeks than a normal curve would give. The card needs at least 20 weekly returns.

The formula

Count the weeks in each 1 percent band. Skew and excess kurtosis describe the shape: which side the long tail is on, and how heavy the tails are.

As computed in the RupeeCase app
Bars: weekly returns, 1% wide, -15% to +15% (refitted; the ends fold into the end bars)
Markers: Mean, VaR 95, VaR 99
Skew = m3 / m2 ^ 1.5 (population moments of weekly returns)
Excess kurtosis = m4 / m2 ^ 2 - 3
Needs 20 weekly returns

Worked example

Example only

  • 52 weeks: most between -2% and +3%
  • Two weeks at -9% and -11%, none past +6%
  • The long tail is on the left: skew below zero

How to read it

  • Most bars near the middle and a few far left: the falls came in rarer, larger steps.
  • Excess kurtosis above 0: more extreme weeks than a normal curve gives.
  • Read VaR 95 against the bars: about 1 week in 20 lost more.

Watch out

  • The caption says Kurtosis; the value is excess kurtosis, as the tile says.
  • This recording reads month end history, so its bars are months.
  • A few years of weeks is a small sample for the tails.

Transcript

Tap Sound on for voice and music. These are the captions of the video, chapter by chapter.

  1. 0:00Where it is. In Advanced, the Risk tab draws every return as a histogram.
  2. 0:10What it means. Each bar counts the periods in a 1 percent band; markers show the mean and VaR.
  3. 0:42Worked example. Worked example, example only: 52 weeks: most between -2% and +3%. Two weeks at -9% and -11%, none past +6%. The long tail is on the left: skew below zero.
  4. 0:48How to read it. How to read it: Most bars near the middle and a few far left: the falls came in rarer, larger steps. Excess kurtosis above 0: more extreme weeks than a normal curve gives. Read VaR 95 against the bars: about 1 week in 20 lost more.
  5. 0:54Watch out. Watch out: The caption says Kurtosis; the value is excess kurtosis, as the tile says. This recording reads month end history, so its bars are months. A few years of weeks is a small sample for the tails.

Terms in this tutorial

  • Return distributionA bar chart of how many weeks fell into each band of return, from big losses on the left to big gains on the right.
  • SkewA measure of whether the big surprises in returns tend to be gains or losses.
  • Excess kurtosisA measure of how often returns were extreme compared with a normal bell curve.
  • VaRValue at risk: a weekly loss that was matched or exceeded only rarely in the history.

Learn it in depth