Seeking Alpha
Long only, research analyst, newsletter provider, portfolio strategy
Profile| Send Message|
( followers)  

This is the fourth article of a series. The previous episodes described the market timing effect for three ETF portfolios holding a stock index, sectors and global assets. This one shows that it can be even more profitable for a dividend portfolio.

Dividend investors are usually averse to market timing. They have the feeling to sell low, to buy high and to lose money on each ex-dividend date when a portfolio is in cash. Indeed, timing indicators have inertia. When they trigger a "sell" signal, the market has often already gone down, and on a "buy" signal it has more often already gone up. Moreover, false signals are triggered from time to time. As an example is more convincing than a long theory, this article will show the benefit of a simple market timing rule on a dividend portfolio. I use a simple rule-based strategy, exempt of survivor bias and look-ahead bias (this article explains why).

Strategy Description

The portfolio contains ten S&P 500 stocks in equal weight. It is rebalanced every four weeks applying the following rules and reinvesting the capital, plus dividends, plus gains, minus losses.

- Companies with a short interest above 5% of the float are excluded.

- Companies with an average dividend yield (5 years) or a current yield below 4% are excluded.

- Finally, the ten companies with the lower Total Debt to Equity ratio are selected.

Market timing rule:

The portfolio is sold - or kept in cash - when the S&P500 current year EPS estimate is below its value 15 weeks ago.

In my previous articles, the timing rule was using technical analysis. Here, it is based on a fundamental data: an aggregate EPS estimate. Dividend investors, who are often averse to technical analysis, will probably prefer this one.

Simulation and current portfolio:

The next table gives simulation results on 10 years, with and without market timing, compared with SPY (SPDR S&P 500 ETF Trust). Transaction costs are 0.2%.

10y. starting 8/21/2003

CAGR

Risk (STD*)

Max Drawdown

Sortino Ratio

SPY

7.25%

23.8%

-55.4%

0.19

Strategy without MT

14.3%

19.8%

-40.1%

0.7

Strategy with MT

15.2%

13.6%

-14.3%

1.08

*STD: standard deviation.

The return year-to-date is 16% and the current drawdown 9%. The next table lists the current holdings.

Ticker

Name

Industry

Yield

COP

ConocoPhillips

Oil, Gas & Consumable Fuels

4.17

PEG

Public Service Enterprise Group Inc

Multi-Utilities

4.43

HCP

HCP Inc

REITs

5.21

PNW

Pinnacle West Capital Corp

Electric Utilities

4.01

T

AT&T Inc

Telecommunication Services

5.32

HCN

Health Care REIT Inc.

REITs

5.04

EXC

Exelon Corp

Electric Utilities

4.04

VTR

Ventas Inc.

REITs

4.4

KIM

Kimco Realty Corp

REITs

4.1

TEG

Integrys Energy Group Inc

Multi-Utilities

4.75

The analysis of individual stocks is out of scope here. This list is useful as an example and to explain the current drawdown. The portfolio is 40% in REITs, whose dividends and prices go down when bond yields are rising. On the next rebalancing dates, some REITs may be pushed out by a lower yield or a higher short interest. Another scenario is a turning point in the market. In this case, the market timing rule will force the portfolio to be sold and kept in cash, until the next entry point based on data reflecting the new situation. This example helps understand how this dynamic portfolio works, but it is more interesting to quantify in which extent the timing rule contributes to the performance.

Contribution of market timing

Return and risk:

For this strategy on this time interval, the market timing rule has little influence on the annualized return (14.3% to 15.2%), but reduces the drawdown from -40.1% to -14.3%, and the standard deviation from 19.8% to 13.6%.

Lost dividends:

The cumulative time out of the market is two years, and the longest period is 14 months. Taking an average yield of 4.5%, the missed dividends represent a loss of 9.2% on the total return, or 0.9% annualized. The global performance already discounts it: the annualized return is slightly higher, despite lost dividends.

Transaction costs:

When a market timing signal occurs, the ten stocks are sold or bought together. The portfolio went to cash five times in ten years. The "timing" turn-over is 4% a month and the "normal" turn-over is 10% (a stock permutation in a month on average). Therefore, market timing adds 40% to transaction costs. Once again, the global performance already discounts it.

Possible leveraging:

A low drawdown is a solicitation to leverage. The next table gives the results with a 1.5 factor and a carry cost of 2%.

10y. starting 8/21/2003

CAGR

Risk (STD)

Max Drawdown

Sortino Ratio

Strategy without MT

14.3%

19.8%

-40.1%

0.7

Strategy with MT x1.5

25.1%

20.5%

-21.1%

1.32

The risk is almost the same as without timing, for an additional annualized gain of 11% and a maximum drawdown divided by two.

The main advantage of market timing is not a higher return, but a lower volatility and drawdown. It gives the choice either to take less risk, or to leverage and gain more with the same volatility and a safer drawdown.

And with another strategy?

I wanted this demonstration bias-free and as simple as possible. I use more complicated strategies and timing rules on S&P 500 and Russell 2000 companies. If you want to back test the same timing indicator on your own strategy or portfolio, all signals for 10 years are in the following table (starting date: 8/21/2003). Remember that calculating the risk is as important as the return. Feel free to share your results in a comment.

Cash Period

Sell

Buy

1

11/5/2007

2/25/2008

2

5/19/2008

7/13/2009

3

11/28/2011

1/23/2012

4

6/11/2012

10/31/2012

5

5/13/2013

8/5/2013

Conclusion: it is safer, but not safe enough

Implementing a good dividend strategy and market timing is probably the safest way to invest in stocks. But it is dangerous to invest too much on a single rationale, even the best one. The usual disclaimer says: "past performance is not a guarantee for the future". It seems obvious because historical data are never enough to build a perfect mathematical model. In fact, the truth is worse: even a perfect mathematical model is not a guarantee. Gamblers call this phenomenon "bad luck", scientists call it "non-ergodicity" (explained here by Pr Ole Peters). The best way to avoid bad luck is combining strategies based on non-related rationales with an appropriate allocation vector. Readers wanting more details can find links to my book and website in my Seeking Alpha profile.

Source: How To Add 11% A Year To A Dividend Strategy For The Same Risk