Do The StockRanks Work For High Yield Investors?

13th Aug '15 by Ben Hobson

21 comments 24400 reads

Over the past two years Stockopedia’s scoring and ranking of every company in the market has turned up some inspiring results. Shares with the greatest exposure to the key factors of Quality, Value and Momentum have performed very well on average. In fact, a quarterly rebalanced portfolio of the strongest 10% of stocks has delivered a return in excess of 70%. This is plainly an eye-catching performance for any investor focused on capital gains. But for those with a preference for dividend income, the question remains - can StockRanks be used to find dependable high yields too?

Why dividends matter

The impact of dividends on long term ‘total’ portfolio returns’ is well documented. One of the most comprehensive studies was published in a 2002 book called Triumph of the Optimists by UK academics Elroy Dimson, Paul Marsh and Mike Staunton. They showed that while year-to-year investment performance is driven by capital gains, the lion’s share of long-term performance actually comes from reinvested dividends. Their research showed that over 100 years, an investment in the market portfolio with dividends reinvested would have produced 85 times the wealth generated by the same portfolio that only relied on capital gains.

55cc801a6dd71triumph_of_the_optimists.pn

What StockRanks say about dividends

For those new to the concept, each of Stockopedia’s Quality, Value and Momentum Ranks is built using a composite of ratios and data points for each company. However, the only dividend-specific ratio in use across the ranks is the Dividend Yield. This is one of the six measures used in the ValueRank. In this case, the yield is used as a signpost to potentially under- or over-valued shares, and not explicitly as a guide to attractive dividend plays.

But while dividends are deliberately under-represented in the StockRanks, you could argue that the ranks bake in a lot of implied dividend information. The Quality Rank, for instance, uses both the Piotroski F-Score of financial health and the Altman Z-Score of bankruptcy risk. Interpretations of both these scores are used by investment bank Societe Generale as ‘quality’ filters in their Quality Income index - a successful high yield, high quality strategy. In addition, the Quality Rank tends to be higher in stocks with a track record of strong profitability and cashflow generation. These are items that income investors look for.

Equally, 50% of the Momentum Rank is devoted to ‘earnings momentum’ factors like…

Unlock with Email

Unlock with Google

Already have an account?
Login here

LittonOwl

13th Aug '15

Thanks for a thought-provoking article, Ben.

I now live off the income my investments provide, so this article was of great interest to me, as would be a dedicated Stockopedia dividend channel !

My portfolio contains a mixture of large caps and small caps stocks (including a fair few of the stocks you mention), together with fixed income (ORB retail bonds, preference shares and PIBS) and specialized investment trusts (eg. infrastructure/renewables, P2P, Asset Leasing funds etc) and a couple of REITS. It would be great if any such channel could include some of these sorts of investments too.

Regards
Phil

sealeyd

Hi Ben

Good article that resonates with my current selection approach.

I've started with the Winning Growth and Income Screen and then added to that an additional filter for StockRanks > 90, removed any AIM shares, and then excluded any stocks that appear on Short Screens.

In two months the portfolio has returned 5%.

It does bring in a couple of additional requests for Stockopedia functionality:

The ability to screen for changes in StockRank (e.g. 3m change in StockRank) or at least easily check the StockRank over a period of time
The ability to back test a screen

Many thanks

David

pka

Ben, Thanks for an interesting article.

You suggested using a rule that the trailing yield should be greater than 4%. This rule has the advantage of simplicity, but I can see two drawbacks to it. The first is that the trailing yield is not necessarily a good indicator of future yields, because dividends can be cut. The second is that 4% is an arbitrary value; if the stock market should fall significantly then average yields are likely to increase so the number of stocks with yields above 4% would also increase, and conversely if the market rises significantly then the number of stocks with yields above 4% would decrease.

Therefore I suggest that one instead uses a rule that the 'Yield % Rolling 1 year' (which is the forecast yield for the coming year) is greater than (say) 1 times the 'Market Median' value. I think this rule would overcome both of the drawbacks mentioned above.

Edward Croft

pka - the trouble with forecast dividends of course is that they can easily be very wrong too. In a bear environment or recession broker forecasts will prove over-optimistic. Realised yields can be far less than anticipated. Value investors (of which I take yield investors to be a subset) tend to prefer historic data as it's fact, rather than opinion. Companies are often loathe to cut their dividends as it leads to investor clientele churn and a change in ownership as well as future mistrust of the company. Given the Quality Rank input into the original list this may have helped minimise the exposure to dividend cuts.

Ben pointed out that even though the yields of a few of these stocks were indeed cut, the overall yield still remained high across the portfolio. De La Rue being the classic example - it lost 30% overnight (see chart below) but still the portfolio outperformed. De La Rue still ended up paying a 3.3% yield on cost in spite of the massive cut of the divifrom 40p to 25p. That 3.3% is still a damn good yield compared to the market and contributed to the total return of the portfolio.

55ccdf8cb0983DE_LA_RUE_Share_Price_-_DLA

The point you make about the 4% rule being too static is a very good one. If Ben does decide to go forth with a "Dividend Channel" at Stockopedia I'm sure he'll make sure that's factored in.

Here's a link to the original list and the data from our archives on Google Sheets - https://docs.google.com/spreadsheets/d/1JHL6Eae2ql...

name	ticker	Yield TTM %	Div Cover	Mkt Cap GBP	StockRank
Amlin	AML	5.59	2.27	2327.5	98
Cenkos Securities	CNKS	5.58	1.18	136.5	98
Persimmon	PSN	11.38	0.58	3896.5	98
Catlin	CGL	5.70	2.07	1867.7	98
UK Commercial Property Trust	UKCM	6.44	1.53	975.8	94
Phoenix Group	PHNX	7.58	1.28	1584.1	94
BP	BP.	4.62	1.61	89373.0	94
Laura Ashley Holdings	ALY	7.69	1.08	189.2	94
GVC Holdings	GVC	7.92	0.50	269.3	94
Costain	COST	4.00	1.58	270.6	93
Tui AG	TUI	7.10	1.32	4068.4	92
IG Group	IGG	4.61	1.43	2235.2	92
Andrews Sykes	ASY	4.81	1.53	156.4	91
Londonmetric Property	LMP	5.01	2.86	878.0	91
Newriver Retail	NRR	5.39	2.24	295.8	90
De La Rue	DLAR	5.73	1.11	745.1	90
Berkeley Group	BKG	7.28	1.05	3347.4	89
Real Estate Credit Investments PCC	RECI	4.79	6.02	120.2	89
Intermediate Capital	ICP	5.16	1.70	1621.0	89
Sky	SKY	6.63	1.02	13729.9	88
United Utilities	UU.	4.03	3.00	6099.5	88
Imperial Tobacco	IMT	4.58	0.57	25056.4	88
Highland Gold Mining	HGM	6.89	2.03	230.1	87
J Sainsbury	SBRY	5.51	2.14	6001.7	87
John Menzies	MNZS	4.03	1.89	403.4	86

	Average	5.92	1.74	6635.1	92
	Median	5.58	1.53	1584.1	91

steviej

14th Aug '15

I've done virtually the same, except I've added a spread filter as well (<400). I prefer the approach of The Winning Growth and Income screen, where it looks for yield above x% of market constituents, rather than a fixed number. In fact the screen is a bit of a simplified QVM filter, ranked by Yield.

One of the things I like to do is check which stocks have newly qualified, or the date qualified and then look at the StockReport history - so I second the request for more historic stockrank data.

Ed, you wrote: " the trouble with forecast dividends of course is that they can easily be very wrong too. In a bear environment or recession broker forecasts will prove over-optimistic."

Fair enough, but the only difference between Stockopedia's 'Dividend Dogs' and 'Dividend Dogs (Forecast)' stock screens is that the former uses historic yields whereas the latter uses the rolling 1-year forecast yields. The former has had a +45% return and the latter has had a +81% return to now since January 2012, which suggests that using rolling 1-year forecast yields may be more effective than using historic yields. Admittedly, the period since 2012 has not been a bear environment, the stocks selected by both screens are likely to be different from the proposed StockRanks High Yield screen, and the length of time since 2012 is not very long so the conclusions may not be statistically significant.

-

Mark Simpson

Ben,

Interesting thoughts.

Be careful about quoting the 'bulk of returns come from re-invested dividends' line. It is actually a quirk of compound maths. The actual figures are something like

nominal return with reinvested dividends 11%,
inflation avg. 3.5%,
dividend yield avg. 3.9%

Which they deduce means that capital gains contributed 3.6% and then take off the inflation avg. of 3.5% and conclude the bulk of the real returns came from dividends. Of course the choice to apply the inflation to capital gains is arbitrary. Simply reversing the logic would show that reinvesting dividends in a market without capital growth would lead to very low real returns and capital gains were the driver of the bulk of returns. The truth is that both dividends and capital gains are roughly equal in their contribution to equity returns.

The second fallacy is that assuming that since re-invested dividends do provide a good chunk of long term overall market returns then sorting any set of shares into high and low yield will show long term outperfomance to the high yield ones. This may be true but it needs empirical evidence to show this since it is not implied by logic.

Dremen in 'Contrarian Investment Strategies' showed that dividend yield did produce outperformance but it was much weaker than other 'value' metrics like P/B, P/E with roughly half the outperfomance of these others. (It is sightly better when you look at sharpe ratios etc. since the high yield group also had lower volatility but still had lower sharpe ratios than other simple metrics.)

Given that the Stockrank already includes a contribution from the stronger value metrics the real question to ask in your article is not did the screen outperform the FTSE TR but did it outperform the screen without the yield criteria?

In fact the best way would be to take the set of shares >£100m mkt cap , SR >75 and sort them into Dividend Yield deciles/quartiles and show the total returns and volatities of these deciles/quartiles.

This would really show if yield is worth adding into a mechanical SR strateg or not.

Cheers,

Mark

The former has had a +45% return and the latter has had a +81% return to now since January 2012, which suggests that using rolling 1-year forecast yields may be more effective than using historic yields.

The reason the forecast yields have been more effective is likely because we've been in a recovering economy and broker forecasts have been systematically too low. As a result, both EPS and DPS forecasts have been upgraded up month by month over the last couple of years.

So yes, in a recovering economy I would suggest that using forecast yields is the best approach, but remember - broker forecasts are on average 45% wrong on a 1 year forecast, and something like 95% wrong on a 2 year forecast (see Montier). Much of that difference is due to the enormous slashes in forecasts that come during recessions and bear market periods.

There may be a better, perennial balance in using 'a bit of both' - using rolling current yields which incorporate both historic and forecast yields into a single blended number. If we get any time we'll try and test it.

Ed, you wrote: "So yes, in a recovering economy I would suggest that using forecast yields is the best approach, but remember - broker forecasts are on average 45% wrong on a 1 year forecast, and something like 95% wrong on a 2 year forecast (see Montier). Much of that difference is due to the enormous slashes in forecasts that come during recessions and bear market periods."

Are those figures for the accuracies of broker forecasts for dividends or profits? I would expect forecasts of dividends to be more reliable than forecasts of profits, because companies are reluctant to cut dividends. In any case, I think the important question is which is the more reliable predictor of next year's dividend: is it last year's historic dividend or is it the average of broker forecasts of next year's dividend?

I suggest adding another rule to the proposed Stockopedia High Yield screen, which is that dividend cover (whether historic or forecast) should be greater than, say, 1.2. This is because if dividends are well-covered by profits, I would assume that the chances of them being cut in the future are reduced.

robinleggate

Interesting article. As I am retired and find dividends more reliable than capital gains, I set up my own screen some time ago requiring a good yield within the good rank screen. I also check that the dividend is sustainable, by, among other things, checking that the FCF after capex is usually more than the dividend. I would be interested in a "dedicated dividend channel"

brucepackard

15th Aug '15

Interesting article. My main concern is that you can't just look at dividend cover and assume a dividend is sustainable . By way of example northern rock, RBS and HBOS had well covered dividends up to 2008....until they didn't . Paul Scott has pointed out many companies with weak balance sheets pay out dividends when they should be paying down debt. At this point in the cycle I would strongly suggest avoiding high yield, weak balance sheet stocks. But a thought provoking idea for when the cycle turns

Howard Adams

16th Aug '15

I am interested in a dedicated dividend channel.

RollingR

Income is an important matter for many investors and I would support developing a dedicated dividend channel

Chrisfarrell21

17th Aug '15

A dedicated dividend channel would be a great addition, good idea.

Thanks

Chris

fourayes

18th Aug '15

This was a useful discussion for income focussed investors. My view is that 4% yield is at the very top end of my expectations. I prefer to take my long term, >5 yr, Cagr of share value, and add to that my achieved yield, to obtain my actual but inflation boosted return. That way I measure both in one metric.
I would like to see this simple CAGR of price, for say 5 and 10 years, included somehow.

Mintaka1960

I also would like a dividend channel.

jjis

19th Aug '15

Thanks for the interesting article Ben. As some others have commented some other factors and filters can be helpful over and above just a focus on yield either historic or forecast, although obviously the StockRanks seem to help.

I like to focus on security of the dividend via earnings and cash flow coverage as others have suggested. While the quality ranks cover financial security and business quality combined I like to look at these separately to get a feel for the financial security of the balance sheet and the quality of the operations producing the cash flow and dividends.

More controversially, but in common with the Dividend Aristocrat indices I also look at dividend streak and dividend growth both historic and forecast (sorry Ed.) I do this as I am looking to invest in stocks with yields which are well covered, have grown and are forecast to grow and ideally being produced by a quality business with good operating metrics such as a high and stable operating margin and ROIC / ROCE. I then also look for this to be backed up by a strong balance sheet.

On the Momentum side of things I also like to split this up between earnings momentum and price momentum. I then would ideally like to see the earnings forecasts either stable or being upgraded as this should mean forecast growth should then be achieved. I pay some attention to 12 month price momentum, but I'm less influenced by this. I have also set up my own scoring system to rank stocks on this basis.

Twitter: @compoundIncome

Web Site: Compoundincome.org

19th Nov '15

People might be interested that I have just set up two real-life 'systematic' portfolios using StockRanks plus some other selection rules, each containing 20 stocks selected from the FT350. I used the same selection rules for both portfolios, except that for one portfolio I used a rule that the historic dividend yield of all stocks must be non-zero, whereas for the other portfolio I used a rule that the 'Yield % Rolling 1 year' (which is the forecast yield for the coming year) is greater than the 'Market Median' value. There are currently 9 stocks that are in both portfolios. I think it will be interesting to see how these two portfolios perform over the coming year, on a total return basis with dividends reinvested. Of course, the sample size of a 20-stock portfolio is too small and a one-year period is too short to draw any statistically valid conclusions from the results!

21st Nov '15

Hi pka you and others may also be interested to see that I set up a live systemic or mechanical portfolio just over 6 months ago containing 20 stocks based on the Scoring criteria I set out in post 18 above. You can read about it and how to view it here which also detail my own longer term performance from targeting these kind of stocks.

It is early days for the systemic portfolio as it has only been going since April this year but the result so far are quite promising as it is ahead of all of the Guru Income screens over the last 6 months as far as I can tell as it is up by 11.6% against the All share which is down by 3.7% over the same period to the end of October.

However, as you say the sample size of a 20-stock portfolio is too small and a six month period is too short to draw any statistically valid conclusions from the results.

Cheers
Jamie

Hi Jamie, That's an impressive performance from your systematic porfolio so far.

Do the StockRanks work for high yield investors?

Why dividends matter

What StockRanks say about dividends

Unlock the rest of this article with a 14 day trial

About the Author

Ben Hobson

21 comments

Unlock the metrics for all these stocks and more...

Unlock the metrics for all these stocks and more...