Recently, I wrote about quantifying risk using the Risk Grade metric from the Nasdaq website. After comparing the performance of my portfolio and the associated Risk Grade, I was surprised to see a correlation. However, without knowing more about the Risk Grade algorithm, it was difficult to understand why this is the case.
After searching through the internet, I found that the Risk Grade is based on the Risk Metric methodology. In fact, there’s quite a bit of academic work done on this.
With more digging, I found the revised Risk Metric scheme (1996 Scheme).
Note: Apparently, there is a RM2006 method and I will be reading this when I get some time.
So, after reading the RM1996 technical document, I figured out that the Risk Grade metric uses a decay factor of 151 trading days. The variance is calculated on the day basis with the current day (day 0) having the most relevance and impact, and the 151th day having the least.
This is based on the following equations (excuse the math). The equations are taken from the technical document.
Where lamba = 0.97 and r is the daily gain/loss. 151 days is based on the following:
We are essentially measuring the daily volatility with a decay factor. It is different from the Beta  section 1.3.2 from the technical document.
“Beta measures how much an individual stock is likely to move with the general market. A beta of 1 means that a stock will tend to move lockstep with the general market, while a beta of 1.5 means that, on average, the stock will rise 1.5% for any 1% rise in the stock market, and fall 1.5% with any 1% fall in the stock market, compared to the riskfree interest rate. Beta can be used to compare the systemic risks of various stocks: the higher the beta, the more risk a particular stock is likely to contribute to a portfolio of stocks. While elegant in its simplicity, beta has several limitations which are rooted in its parent theory of the Capital Asset Pricing Model (CAPM). First, it is only a relative risk measure: beta is only a measure of how a stock is likely to move relative to an overall stock index, and gives no indication of the stock’s unique volatility (or the overall stock market’s volatility). Beta can be misleading because two stocks with the same beta generally have a different unique risk. Second, it only measures incremental systemic risk for a perfectly diversified portfolio of stocks (i.e., a stock with a beta of 1 could easily contribute twice as much volatility as the broader stock market, if you have an undiversified portfolio). Third, CAPM focuses only on the risk premium of equities relative to riskfree assets, does not address fixed income and currency investments, and consequently, is difficult to apply across asset classes. In sum, we can say that RiskGrades account for both systematic and unique risk and thus show the whole picture of risk.”
D4s Risk Grade Tool
With the simplicity of the Risk Grade method, I created the following Google Sheets Risk Grade tool. This allows you compare a company’s Risk Grade vs the S&P500 index.
https://docs.google.com/spreadsheets/d/1lMJk6F_bJiI85aiR8IwPJYcGgCT0LE6P2MYhlaT3sE/edit?usp=sharing
With google sheets, you need to download a local copy and then save.
I also changed the lamda=0.99 to increase the time decay to 458 trading days  this should have a better sampling size  as well as making it different from the original algorithm. (I’ve ‘opened’ up the calculations in case you are interested in the math.)
Note: Ideally, the S&P Risk Grade should calculate the variance/covariance of the Risk Grades of all the 500 companies, but I just took the short cut of using the daily swings of .INX.
For the tool, just enter in the yellow cells. Note: .INX is the S&P500 index.
Now to compare this with my previous results.
Results using Nasdaq Risk Grade TM
Results using D4S RiskGrade  Normalized with S&P500 RiskGrade
Okay  the trends are the same. It clearly shows my losers  BBL, NOV and IVZ.
Problems with Numbers
As with all things in life, a number is meaningless. Based on the results, a relative risk grade < 2.0 is clearly preferable. However, it is also clear that Risk Grades move based the time decay factor and when day 0 is. After playing with time Start Date, I do observe at some dates where my IVZ has a score lower than 2.0. Is this a good time to invest?
Maybe… but we need to see the trend.
D4s Risk Grade 10 Yr Chart
Similar to my Yield Theory and P/E charts, ROE/OM, Payout Ratio, FCF charts, a 10 years trend can help to provide addition information.
So, I created the Risk Grade Chart on google sheets  showing you the 10 years trends of the Risk Grades of any company vs the S&P500 index.
https://docs.google.com/spreadsheets/d/1MH1vC8gahOWEf4RQvdVFjJ5hNG9BulEs7NTBXrKIAY/edit?usp=sharing
With google sheets, you need to download a local copy and then save.
For the tool, just enter the two companies to compare in the yellow cell.
BBL vs S&P500 Risk Grade Data & Chart
Let’s compare one of my losers using the 10 years Risk Grade Chart.
BBL

.INX

Ralative RG
 
7/15/2016

273.24

76.84

3.56

2/14/2016

268.13

90.29

2.97

9/15/2015

203.66

86.59

2.35

4/16/2015

164.05

60.47

2.71

11/15/2014

135.85

54.99

2.47

6/16/2014

104.45

50.01

2.09

1/15/2014

116.73

57.38

2.03

8/16/2013

131.75

56.46

2.33

3/17/2013

133.89

63.18

2.12

10/16/2012

141.98

71.12

2.00

5/17/2012

177.64

88.18

2.01

12/17/2011

207.82

107.50

1.93

7/18/2011

224.06

129.08

1.74

2/16/2011

149.10

68.63

2.17

9/17/2010

176.87

84.83

2.09

4/18/2010

222.32

110.72

2.01

11/17/2009

236.01

108.43

2.18

6/18/2009

304.97

146.48

2.08

1/17/2009

427.01

211.35

2.02

8/18/2008

507.98

252.84

2.01

3/19/2008

236.69

100.93

2.35

10/19/2007

226.11

100.13

2.26

5/20/2007

199.61

74.51

2.68

12/19/2006

159.97

51.39

3.11

7/20/2006

152.32

46.00

3.31

2/18/2006

175.46

57.86

3.03

Median

188.63

80.83

2.17

The chart (orange bars) provides a relative risk of BBL to the S&P index. It is 2.17 more “riskier”.
I can ask myself  is the relative risk worth the dividend yield  which has been cut recently. Maybe I needed to be more careful with BBL back then. Maybe...
JNJ vs S&P500 Risk Grade Data & Chart
Let’s compare one of my winners using the Risk Grade Chart
JNJ

.INX

Relative RG
 
7/15/2016

68.12

76.84

0.89

2/14/2016

91.14

90.29

1.01

9/15/2015

87.97

86.59

1.02

4/16/2015

78.17

60.47

1.29

11/15/2014

72.91

54.99

1.33

6/16/2014

65.84

50.01

1.32

1/15/2014

68.14

57.38

1.19

8/16/2013

63.31

56.46

1.12

3/17/2013

56.76

63.18

0.90

10/16/2012

54.18

71.12

0.76

5/17/2012

63.26

88.18

0.72

12/17/2011

76.28

107.50

0.71

7/18/2011

88.86

129.08

0.69

2/16/2011

66.61

68.63

0.97

9/17/2010

61.63

84.83

0.73

4/18/2010

77.18

110.72

0.70

11/17/2009

75.26

108.43

0.69

6/18/2009

102.38

146.48

0.70

1/17/2009

143.11

211.35

0.68

8/18/2008

180.25

252.84

0.71

3/19/2008

66.87

100.93

0.66

10/19/2007

64.85

100.13

0.65

5/20/2007

57.70

74.51

0.77

12/19/2006

57.68

51.39

1.12

7/20/2006

57.16

46.00

1.24

2/18/2006

58.85

57.86

1.02

Median

67.49

80.83

0.83

Clearly, this has a different trend with a median of 0.83. I consider JNJ a Sleep Well At Night (SWAN) stock.
Criteria Risk Metric Update
With the new tools, I need to modify my “Risk Criteria”  which is to use the 10 year Risk Grade median Threshold < 2.0. Of course >2.0 is okay if the company is worth the “Risk” for investing.
As with all metrics, the Risk Grade metric is just another piece of the puzzle. The puzzle is often incomplete, so please be careful with any investing decisions you make. Please read my disclaimer (and like BrokeDividendInvestor), DON’T FOLLOW ME.
Let me know what do you think of my Risk Grade & Chart tools?
D4s
Disclosure: Long in all the companies above
Disclaimer: Use the tools at your own risk. Please read my disclaimer for more details.
You are a Google Sheets wizard.
ReplyDeleteThis relative risk rating is quite interesting. I used it on what I think of as my riskiest stock, WDR. The relative rating when I purchased was somewhere in the high 2s close to 3.0, but the yield at the time of purchase was >4x the S&P yield.
Great work on figuring this out and thanks for sharing the Google Sheet!
Thanks Papa Div, I am no wizard  I am still learning the nuances of Google Sheets. I am glad the spreadsheets were useful.
DeleteI ran another experiment with 5Yr performance Gain (i.e. Assuming I bought the same companies in my portfolio 5 years ago). My BBL and NOV are still losers. :(
D4s
Meaning that their relative rating 5 years ago was also above 2.0?
DeleteI was using the median relative risk grade (look at the bottom of the table in the risk grade chart summary). Both are above 2.
ReplyDeleteD4s