Algorithmic Industry Momentum Investing

|
 |  Includes: IDU, IYC, IYE, IYF, IYH, IYJ, IYK, IYM, IYW, IYY, IYZ
by: Emmanuel Marot

If only playing the stock market could be as simple as Momentum Investing. Buy the raising stocks and avoid the ones going down. Because there’s some inertia on the market, wining stocks will continue to appreciate, and you’ll beat the market.

That’s the ‘theory’, anyway. In practice, although the Momentum effect does exist (see Jegadeesh and Titman, 1993), it’s rarely working well enough to compensate for transaction costs, opportunity costs and the like.

Our idea is to focus on whole industries instead of picking up individual stocks. First, industry indices are composed of multiple companies, so their volatility will be lower. Second, we can expect less market discrepancies, such as under-pricing and over-pricing at an aggregate level, meaning an industry performance is driven more by fundamental economic factors that have a slower rate of change. At last, since one can invest in Industries through index ETFs, there will be less transactions than for individual stock-picking, and therefore lower transaction costs.

Note that this idea is not exactly original, see for instance Moskowitz and Grinblatt (1999), Fraulo and Nguyen (2004), Swinkels and Tjong-A-Tjoe (2008), or Jonathan Selsick’s article on Seeking Alpha. But what we propose here is a very simple and efficient way to profit from the Industry Momentum phenomenon.

First, we select 10 iShares ETFs, matching the 10 US industries indices of Dow Jones:

Financials (NYSEARCA:IYF) Health Care (NYSEARCA:IYH)
Technology (NYSEARCA:IYW) Consumer Goods (NYSEARCA:IYK)
Industrials (NYSEARCA:IYJ) Basic Materials (NYSEARCA:IYM)
Consumer Services (NYSEARCA:IYC) Utilities (NYSEARCA:IDU)
Energy (NYSEARCA:IYE) Telecommunications (NYSEARCA:IYZ)
Click to enlarge

We then calculate a ‘momentum’ for each of those ETFs, which is the ratio between two exponential moving averages, one short, one long. Therefore, if the prices are going up, the short EMA will be greater than the long EMA, and the momentum will be greater than 1. We took an approximately monthly period for the short EMA (n=30) and an approximately quarterly period for the long EMA (n=90).

As a market benchmark, we take the Dow Jones U.S. Index (NYSEARCA:IYY).

We start the experiment on the 14th of July, 2000, because it’s the first Bastille Day of the new millennium. No, seriously: that’s the earliest date when we can have all the indices values (we downloaded prices history from Yahoo! Finance). We arbitrarily chose to trade approximately on a monthly basis, every 28 days to be precise (a multiple of 7, so we always trade on the same weekday).

At each trading day, we select the Industry with the highest momentum. If it’s the same than the previous period, then there’s no trading activity, we keep our positions intact. If no industry has an upward momentum, then we simply get out of the market and keep everything in cash.

Simple enough, but does it really work?

Here are the results...

click to enlarge

Click to enlarge

Starting with $100,000 on July 2000, a benchmark ‘market’ investor would have only $107,592 on May 2011. With the industry selection principle described above, an ‘algorithm’ investor would have $266,929. Here is the log of the transactions, each line showing the ending date, the benchmark portfolio value, the algorithm portfolio value, and the last industry choice:

Date Benchmark Algorithm Industry
2000-07-14 $100,000 $100,000
2000-08-11 $96,807 $88,670 Technology
2000-09-08 $99,511 $94,795 Technology
2000-10-06 $93,320 $78,413 Technology
2000-11-03 $95,112 $77,828 Utilities
2000-12-01 $86,494 $78,569 Utilities
2000-12-29 $87,227 $85,257 Utilities
2001-01-26 $89,361 $77,660 Utilities
2001-02-23 $81,606 $72,411 Financials
2001-03-23 $74,764 $67,013 Financials
2001-04-20 $81,476 $74,573 Basic Materials
2001-05-18 $85,256 $82,223 Basic Materials
2001-06-15 $80,270 $75,669 Basic Materials
2001-07-13 $79,993 $76,206 Basic Materials
2001-08-10 $78,511 $74,864 Basic Materials
2001-09-07 $71,489 $72,471 Basic Materials
2001-10-05 $70,121 $67,237 Basic Materials
2001-11-02 $71,831 $68,395 Consumer Goods
2001-11-30 $75,529 $69,311 Consumer Goods
2001-12-28 $77,419 $70,171 Consumer Goods
2002-01-25 $75,546 $69,816 Consumer Goods
2002-02-22 $72,564 $72,600 Consumer Goods
2002-03-22 $76,931 $75,833 Consumer Goods
2002-04-19 $75,692 $76,396 Basic Materials
2002-05-17 $73,998 $77,753 Consumer Goods
2002-06-14 $67,742 $75,354 Consumer Goods
2002-07-12 $61,535 $67,507 Consumer Goods
2002-08-09 $61,160 $70,519 Consumer Goods
2002-09-06 $60,297 $70,519 None
2002-10-04 $54,040 $70,519 None
2002-11-01 $60,573 $70,519 None
2002-11-29 $63,180 $70,519 None
2002-12-27 $59,172 $70,519 None
2003-01-24 $58,488 $70,519 None
2003-02-21 $57,559 $70,519 None
2003-03-21 $60,671 $70,519 None
2003-04-18 $60,573 $70,519 None
2003-05-16 $64,581 $70,519 None
2003-06-13 $67,840 $74,637 Financials
2003-07-11 $68,785 $75,521 Financials
2003-08-08 $67,237 $71,040 Technology
2003-09-05 $70,903 $82,631 Technology
2003-10-03 $71,603 $82,744 Technology
2003-10-31 $73,346 $86,280 Technology
2003-11-28 $74,194 $87,717 Technology
2003-12-26 $76,898 $87,642 Technology
2004-01-23 $80,336 $94,411 Technology
2004-02-20 $80,482 $89,854 Technology
2004-03-19 $78,381 $83,463 Technology
2004-04-16 $80,010 $87,292 Energy
2004-05-14 $77,142 $85,429 Energy
2004-06-11 $80,075 $86,257 Energy
2004-07-09 $78,723 $91,224 Energy
2004-08-06 $75,008 $88,896 Energy
2004-09-03 $78,886 $94,588 Energy
2004-10-01 $80,336 $101,780 Energy
2004-10-29 $80,466 $100,176 Energy
2004-11-26 $84,620 $107,472 Energy
2004-12-24 $86,934 $104,626 Energy
2005-01-21 $83,806 $103,953 Energy
2005-02-18 $86,543 $120,201 Energy
2005-03-18 $85,712 $125,272 Energy
2005-04-15 $82,356 $112,698 Energy
2005-05-13 $83,399 $109,645 Energy
2005-06-10 $87,227 $123,306 Energy
2005-07-08 $88,677 $129,929 Energy
2005-08-05 $89,899 $135,258 Energy
2005-09-02 $89,475 $141,519 Energy
2005-09-30 $90,469 $148,608 Energy
2005-10-28 $88,009 $134,172 Energy
2005-11-25 $93,711 $141,364 Energy
2005-12-23 $93,467 $142,089 Energy
2006-01-20 $93,385 $155,387 Energy
2006-02-17 $95,243 $146,746 Energy
2006-03-17 $96,905 $148,660 Energy
2006-04-14 $95,666 $154,248 Energy
2006-05-12 $95,976 $162,026 Basic Materials
2006-06-09 $92,766 $144,418 Basic Materials
2006-07-07 $93,891 $151,444 Basic Materials
2006-08-04 $94,705 $155,556 Energy
2006-09-01 $97,198 $152,435 Energy
2006-09-29 $98,403 $145,895 Energy
2006-10-27 $101,841 $153,256 Telecommunications
2006-11-24 $104,203 $149,279 Telecommunications
2006-12-22 $104,220 $153,315 Telecommunications
2007-01-19 $105,881 $155,927 Telecommunications
2007-02-16 $108,488 $163,465 Telecommunications
2007-03-16 $103,535 $159,072 Telecommunications
2007-04-13 $108,716 $167,738 Telecommunications
2007-05-11 $112,496 $173,713 Utilities
2007-06-08 $112,838 $162,810 Utilities
2007-07-06 $114,516 $168,485 Basic Materials
2007-08-03 $106,940 $156,481 Energy
2007-08-31 $110,003 $164,341 Energy
2007-09-28 $114,158 $177,840 Energy
2007-10-26 $115,070 $181,386 Energy
2007-11-23 $107,918 $174,038 Energy
2007-12-21 $111,779 $186,811 Energy
2008-01-18 $98,941 $164,170 Energy
2008-02-15 $101,515 $171,390 Energy
2008-03-14 $97,181 $172,156 Basic Materials
2008-04-11 $100,147 $180,495 Basic Materials
2008-05-09 $104,660 $189,552 Basic Materials
2008-06-06 $103,584 $192,801 Energy
2008-07-04 $95,666 $190,892 Energy
2008-08-01 $96,155 $168,310 Energy
2008-08-29 $98,094 $168,188 Energy
2008-09-26 $92,506 $168,188 None
2008-10-24 $67,074 $168,188 None
2008-11-21 $59,857 $168,188 None
2008-12-19 $67,221 $168,188 None
2009-01-16 $65,119 $168,188 None
2009-02-13 $63,311 $168,188 None
2009-03-13 $58,227 $168,188 None
2009-04-10 $66,292 $168,188 None
2009-05-08 $72,092 $168,188 None
2009-06-05 $73,330 $168,188 None
2009-07-03 $70,137 $168,188 None
2009-07-31 $77,256 $186,128 Technology
2009-08-28 $80,759 $192,237 Technology
2009-09-25 $82,242 $197,457 Technology
2009-10-23 $85,125 $205,847 Technology
2009-11-20 $85,859 $213,077 Basic Materials
2009-12-18 $87,292 $211,454 Basic Materials
2010-01-15 $90,160 $225,878 Basic Materials
2010-02-12 $85,761 $208,613 Basic Materials
2010-03-12 $92,001 $230,083 Basic Materials
2010-04-09 $95,601 $241,003 Basic Materials
2010-05-07 $88,970 $212,229 Basic Materials
2010-06-04 $85,533 $208,490 Consumer Services
2010-07-02 $82,046 $192,483 Consumer Services
2010-07-30 $88,530 $206,276 Consumer Services
2010-08-27 $85,728 $203,993 Telecommunications
2010-09-24 $92,669 $221,529 Telecommunications
2010-10-22 $95,797 $223,811 Telecommunications
2010-11-19 $97,654 $234,491 Basic Materials
2010-12-17 $101,776 $249,184 Basic Materials
2011-01-14 $105,751 $258,239 Basic Materials
2011-02-11 $109,075 $263,944 Basic Materials
2011-03-11 $107,103 $266,867 Energy
2011-04-08 $109,498 $283,410 Energy
2011-05-06 $110,492 $265,436 Energy
2011-06-03 $107,592 $266,929 Energy
Click to enlarge

Not only does this algorithm provide a much better return, with the investor ending with 148% more money than the market, but it also beats the market, on a monthly basis, in 63% of the cases. It is also less risky, since the maximum draw down is reduced from 49% to 33%.

Moreover, with only 26 transactions on the whole period, this algorithm requires to trade on average only once every 153 days, which represents a low amount of work for any investor willing to use it.

A few caveats, though: first, those simulations do not take into account transaction costs or tax issues, although, as mentioned above, since there are few transactions and only in Index ETFs, those transaction costs are lower than when picking up individual stocks. Second, whereas the volatility of whole Industries tend to be lower than the companies it contains, it is still more volatile than the market as a whole. As a matter of fact, the standard deviation of those returns is slightly higher than the benchmark: 0.057 versus 0.053.

Lastly, not investing when industries’ momentums are all below 1.0 acts as a market timing filter, trying to prevent being long on a bearish market. If we systematically invest in the industry with the highest momentum, even if it’s under 1.0, the results are slightly inferior, ending at $229,988 (instead of $266,929). The investment is also slightly more risky, with a maximum draw down of 38%, but still lower than the market 49%.

Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.