If you are on the blog page in a web browser from a computer, please subscribe to this using the "Follow by Email" link to the left. If you're on a mobile device you should see something in the frame that allows you to subscribe. Having your email helps me to notify you when Google mucks up email distribution.
If you need to reach me, you can do so using the block at the bottom of the blog or send me an email. I prefer that you use the box at the bottom of this entry, but of course, you can send an email to GreekGodTrading [a t] g m a i l [d ot] c o m (intentionally made difficult for the bots to read and spam me -- so fix it as your intuition tells you).
~~~~~~~~
I trust that everybody is enjoying the long Memorial Day weekend. Let's pause and remember the reason for the holiday ...
~~~~~~~~
I received a question this week that was tied to last week's entry, specifically asking "How many Dividend Champion stocks should I hold to perform better than the S&P 500?"
This isn't as easy a question as it appears.
First of all -- "perform better". In return? In volatility? Both? Some other metric? The question wasn't specific.
First of all -- "perform better". In return? In volatility? Both? Some other metric? The question wasn't specific.
It isn't practical to hold 500 stocks. Commissions would eat you, and management would be a pain in the tail. Contrary to belief, funds that track indexes also do not invest in all the stocks of the index -- they use a subset for this very reason.
Not all stocks in the Dividend Champion list are in the S&P 500, which further complicates the decision tree.
Not all stocks in the S&P 500 pay dividends, further complicating the answer.
Reformulating the question: Using just stocks that pass Paul's Greenfield criteria, pay a dividend, and are on the Dividend Champions list, can I beat or outperform the S&P 500 in return and/or volatility?
So, let's start with the S&P 500 stocks. I'm only interested in
2) those stocks that pay a dividend and that are on the Dividend Champions list, and
3) have less or equivalent risk than the S&P 500.
Regarding the first two criteria, I've compiled that list for you, and it is here.
This seems to be a lot of trouble: I *could* simply buy the SPY (an exchange traded fund that mirrors the performance of the the S&P 500) and be done with it. If I were to do that, I'd have the expectation of the performance that I stated last week (5.97% return annually at an average volatility of 17.90%) and I'd presently capture a dividend yield of 1.85% (see this link as an example of how the dividend yield is determined).
Hmmmm.... 1.85% dividend yield.
I'll not get into here but suffice to say, if I want a dividend yield of greater than 1.85% on a basket of stocks, then the "average" of all the yields of the subset of stocks that I invest in has to be greater than 1.85%.
Since I don't *know* what stocks I'm looking for (yet), the way to address this is to only look at stocks that have a yield greater than 1.85%. In doing this I am guaranteed, no matter what, to beat the current dividend yield of the SPY.
I've built the subset of stocks of the S&P 500 that 1) pass my greenfield criteria, and 2) have a dividend yield of greater than 1.85% as of 5/24 closing prices. The stocks on the list are further screened against the May 2019 Dividend Champion CCC tab located here. The final text list is here.
It's still a long list. 76 stocks, to be clear. I have no desire to invest in 76 stocks. Still unmanageable.
Enter Portfolio Optimization
One way to select a basket of stocks is to do so such that some parameter is optimized. Typical parameters are things like 1) maximizing Expected Return (ER), 2) minimizing volatility, 3) minimizing correlation, etc.
If you were paying attention last week, you read that I mentioned the Sharpe Ratio (SR). This is a metric which calculates ER and risk, the latter which is often interchanged with volatility. Hence, for a given equivalent ER between two securities, the one with lower volatility has less risk (and corresponding higher SR).
So, a natural method to select stocks is to maximize SR within the entire portfolio.
But, here's a complicating component: correlation between assets. As one move up, the others may move up, or they may move down. It all depends.
Here's an example that you can use to better understand this. Let's take the exchange traded fund SPY, which I said is a proxy for the S&P 500, and the Vanugard mutual fund VFINX, which also is a proxy for the S&P 500. Theoretically, both should move exactly in lock-step with the S&P 500, and hence, they should both be in sync.
The reality is that yes, the SPY and the VFINX are almost 100% in step. See the picture below:
Click on the image to enlarge
Calculating the SR's of the SPY and the VXF isn't enough -- the correlation between the two needs to be considered. It would not provide any benefit to me to invest in BOTH -- there is no diversification -- and only expenses would go up. There would be no other benefit that is hidden or not apparent.
So, when we are looking to reduce a basket of stocks, we want to choose stocks that lack correlation ... so, we specifically want to maximize SR but minimize correlation. This becomes a multi-variate problem.
As complicated as this sounds, there is software out there to crunch through the numbers. Excel does a great job at this, and I have a great piece of Excel software to help me do this.
This concept of correlation is important in stock selection. It is best to select stocks that are uncorrelated. Stocks that are perfectly correlated have a correlation coefficient of "1". Stocks that are perfectly INVERSELY correlated have a coefficient of "-1". Stocks that are UNCORRELATED have a correlation coefficient of "0". We want a basket of stocks with as low as a correlation coefficient as possible (close to zero as possible).
Let's take the 76 candidate stocks and calculate the correlation to the S&P 500, the Russell 2000 index, and each other. The following figure is a small slice of that 76-stock universe:
Click on the image to enlarge
^GSPC is Yahoo's symbol for the S&P 500. ^RUT is the symbol for the Russell 2000 index. Other stocks are listed as shown.
The green "1"s that form a diagonal are due to the fact that a stock has a perfect correlation with itself.
I've color-coded the figure to show that the closer the correlation, the greater the amount of green. The more neutral, or uncorrelated, we see orange, and the more inversely correlated, we see red.
The S&P 500 (^GSPC) has a 0.908 correlation with the Russell 2000 (^RUT). Both are darker green, so as one moves up, the other also moves up in almost the same manner.
Conversely, you see that WELL is more negatively correlated with the other stocks shown (see the negative numbers in RED).
As I wrote above, this isn't important in the big picture, but it shows you how stock prices move with each other. The goal in portfolio selection is to maximize diversification through as few selections as possible, and this is one way to get there. We want a basket of stocks that has an average correlation as close to "0" as possible".
Portfolio Component Selection
How to choose which stocks? How to ignore some, but select others? How to weight each stock so to maximize some value? The paper here gets into the gory details. What is important here is that there is a trade off between a risk-free investment, a risk-free + risky investment, and a number of risky investments. The goal is to figure out how to get as risk-less as possible but maximize return -- hence optimization of the Sharpe Ratio, while minimizing individual stock correlation.
Tying this back to the correlation thing above, when I crunch through the numbers, here is the correlation matrix for the "solution":
Click on the image to enlarge
Of significance is that there is NOT a great deal of green on the matrix -- most of these stocks are uncorrelated with each other, or probably better-aptly described, do not have a tight correlation. The average correlation to each other is 0.341, which isn't bad. 0.000 would be ideal.
Given that this is the list of stocks with the lowest portfolio correlation, the next challenge is optimizing SR. There is a large number of possible weightings of stocks to give us a return, and when we do that, we play with volatility. It's possible to maximize Expected Return but have terrible volatility; conversely, it's possible to have extremely low volatility but not maximize Expected Return.
Everything has been leading up to this -- the calculations. Don't worry, I won't go through the details. Simply put, the Efficient Frontier and ER/Volatility points for all the stocks shown is here in the following graph:
Click on the image to enlarge.
If you are not familiar with Efficient Frontier go watch this (you REALLY need to watch that video if this is new to you). In the example I've set the S&P 500 expected return to 5.97% (recall that this is the average annual performance of the S&P 500 over history, not including dividends). Any stock below this value is an underperformer in terms of gain; and any stock above this is an overperformer in terms of gain. Further, recall that the S&P 500 volatility over history is about 17.90%, on average, and when you look at the graph above, you see that the majority of stocks have greater volatility than the S&P 500.
There is a unique condition in this process where two volatilities of greater than some number, when combined in the process, actually reduce the overall volatility of the combined set. This means that it is possible to select stocks with higher ER (y-axis) and higher volatility (x-axis) yet come up with a solution that has LOWER volatility but higher ER. This is the Efficient Frontier blue line in the picture.
The red line is the tangent line to the Efficient Frontier. The slope of this line is determined by the risk-free interest rate -- currently a value of 2.37% and zero volatility. The intersection of the straight red line and the blue line is the best the portfolio can be expected to do, given a risk free rate of 2.37% (go here). It is shown as the purple triangle at the intersecting points.
There is a unique condition in this process where two volatilities of greater than some number, when combined in the process, actually reduce the overall volatility of the combined set. This means that it is possible to select stocks with higher ER (y-axis) and higher volatility (x-axis) yet come up with a solution that has LOWER volatility but higher ER. This is the Efficient Frontier blue line in the picture.
The red line is the tangent line to the Efficient Frontier. The slope of this line is determined by the risk-free interest rate -- currently a value of 2.37% and zero volatility. The intersection of the straight red line and the blue line is the best the portfolio can be expected to do, given a risk free rate of 2.37% (go here). It is shown as the purple triangle at the intersecting points.
So, provided you watched the video above and understand the construct, using weights between 0% and 100%, ALL of those stocks define portfolios that are boundaried by the blue line, and because the red line and the blue line are tangent (intersect at 1 point only), there is a theoretical solution to a unique portfolio that provides the optimum ER at the lowest volatility.
That is what I'm seeking. Maximize ER while minimizing volatility.
That is what I'm seeking. Maximize ER while minimizing volatility.
For a reference point, the S&P 500 has a historical Sharpe Ratio of 0.3335, produced from an average yearly return of 5.97% and an average volatility of 17.90% (see last week's blog entry).
The portfolio shown above has an optimal Expected Return of 5.94% yet the volatility has dropped to 13.97%. This is an improved SR that is 0.4252. The method has the potential to produce the same behavior as the S&P 500 yet a a lower volatility, by specifically weighting each stock between 0% and 100%.
What is that weighting? Here you go:
This is a big list of stocks, but you can see that the weights (next to the symbols in light yellow) get smaller and smaller as you go from top to bottom. This is the "perfect solution", e.g., the one where the Efficient Frontier (blue line) and the risk-free interest rate line (red line) intersect.
Next to the weights are the closing prices for each of the stocks, as of the 5/24 close.
To the right of the closing prices are the latest quarterly dividends that were paid for the stock. Remember, our initial criteria specified that we had to pick stocks with a dividend yield of at least 1.85%.
To the right of the dividend levels are the number of shares, given a $1,000,000 starting level, rounded to the lowest whole number. Note that the portfolio size is $1,000,000, and yet EQIX, at a close on 5/24 of $496.52, would only have you purchasing 9 shares. You can see that some of these stocks will not be purchased if the starting level is much lower (as is the case with me).
Next to the number of shares are the yearly dividend payments, if everything remains constant. Remember -- these are Dividend Champions -- so we have an expectation of dividend growth. At a minimum, we can expect to see $29,644 over the next year.
The right column shows the actual dollar value committed for each stock.
At the bottom of the table, you can see that we have $29K in expected dividends as well as $999K committed in total portfolio, giving us a 2.97% dividend yield. Note how much better this is than the 1.85% dividend yield cut off. We will certainly beat the S&P 500 in terms of dividends received.
Impact of Starting Smaller
It is highly probable that most of us do not have $1,000,000 to throw at a new strategy.
What if we start with $10,000? What happens then?
Here's the table:
The first thing that should be evident is that there is simply not enough cash to fund all of the positions.
The next thing that is evident is that the share purchases of 1 share probably are not worth the cost of commissions. In fact, I think the lowest number is probably around $200 per stock.
As an example, round-trip costs for me (to buy and to sell a security) are $1.00 each way with TradeStation, so I pay $2.00 round trip. If the stock moves 1% upward on a starting position size of $200 then I'm at break even. If you take a look at the right column you can see that many of these position sizes are below $200, so it probably is not worth attempting to enter this portfolio with $10,000.
Commissions will eat into the portfolio, so it important that you understand the impact of round-trip fees.
If you use Fidelity, Schwab, or E-Trade the commissions are higher and you would be worse off. I would simply buy the SPY and be done with it.
As an example, round-trip costs for me (to buy and to sell a security) are $1.00 each way with TradeStation, so I pay $2.00 round trip. If the stock moves 1% upward on a starting position size of $200 then I'm at break even. If you take a look at the right column you can see that many of these position sizes are below $200, so it probably is not worth attempting to enter this portfolio with $10,000.
Commissions will eat into the portfolio, so it important that you understand the impact of round-trip fees.
If you use Fidelity, Schwab, or E-Trade the commissions are higher and you would be worse off. I would simply buy the SPY and be done with it.
Impact of Starting at $100,000
If $1,000,000 is out of reach, and $10,000 doesn't make sense, what about starting at $100,000?
Here's the table:
This scenario isn't far off the mark.
You could argue that any position that is 2% or greater in size (invested capital ~ $2,000 minimum per position), the impact of commissions could be negligible.
If we choose not to invest in some positions, we throw off the weighting for each stock, and we also have cash left over. The key is to deploy ALL our cash.
So, let's only pick those position sizes above greater than 2%. Here's what the new portfolio looks like:
This portfolio contains 20 positions.
The starting level is $100,000.
You can see that you should receive nearly $3K in dividends the first year if you are invested in this portfolio and hold for the duration (obviously presumes that no company cuts their stock dividend).
The new calculated volatility of the portfolio has increased from 13.97% to 14.06%. Not terrible, but not ideal. Expected Return is constant at 5.94%.
You can also see that some of the position sizes jumped -- and a few others dropped. Nevertheless, this should be an achievable portfolio if you have $100K or more to start.
~~~~~~~~~~~~~~
Summary
1) We started with the stocks comprising the S&P 500.
2) We downselected those stocks to choose only the dividend payers that were yielding greater than 1.85%, the present yield of the exchange traded fund SPY.
3) We further downselected stocks to make sure they were part of the Dividend Champions list.
4) We selected a smaller group of stocks that had the lowest correlation to each other.
5) We further selected weights of stocks to maximize the Sharpe Ratio of a portfolio.
6) We looked at investing $1M, $10K, and $100K in that basket of stocks. It is probably better to simply buy the SPY if you only have $10K to invest.
7) We chose only the stocks of the original optimal SR solution that had a position size of 2% or greater (about $2,000), then we re-optimized. SR dropped a bit on the re-optimization, as you would expect, since we used a subset of the previous optimized universe.
8) The answer to the question is 20 -- you need 20 stocks from the Dividend Champions list in order to perform better than the historical average of the S&P 500. Stock selection matters.
~~~~~~~~
Next week I'll get into the rules of buying/selling and how to use options to get into some of the positions with a lot size of greater than 100 shares.
~~~~~~~~
That's all for now. If you have questions -- ask.
~~~~~~~~~
As with all my ramblings, you are responsible for your own actions and I am not. Nothing I've written here is advice to buy or sell any security, so don't do it unless you absolutely take ownership for your actions.
Regards,
Paul
Posts
