Momentum Investing: automated stock selection

ETFs are available which select holdings according to momentum criteria.  One such example is WXM.TO (issued Feb 2012) which replicates the Morningstar Canada Momentum Index (started Dec 2000).

The fund holds 30 equal positions of stocks ranked using a weighted factor:
Trailing Return on Equity 20%
3- Month EPS Estimate Revision (Current Year) 30%
Latest Quarterly Earnings Surprise 10%
Price Change from Month-End 3 Months Ago 10%
Price Change from Month-End 9 Months Ago 10%
Percent Change in Price from 12 Month High 20%

60% of the factor is fundamental, 40% is price momentum.

Results:

During the 2003-2007 bull market, the index returned 40% annually and 22% annually since 2009.

blog figs

The volume on all momentum products is low, currently around 3,000 daily for WXM with spikes over 10,000.  ETF liquidity and spreads are linked to the underlying stocks and therefore should not pose a problem when purchasing.  I would be interested to see the effect of selling a $100K position (approximately the maximum daily volume).

Further information can be found at the following locations:

http://www.exchangetradedforum.com/documents/presentations2012/Wed-1045-BeyondTraditionalIndexing.pdf

http://www.firstasset.com/products/overview/?fund=First+Asset+Morningstar+Canada+Momentum+Index+ETF

Advertisements

Momentum investing – out of sample testing

All of the previous posts used data post tech-bubble recession.  This leaves an unexplored period earlier in the dataset.  Testing from 3/96 to 3/00 gives the results below, which are similar to previous metrics.  This gives confidence that the momentum anomaly the system is designed to capture is robust (also confirming numerous studies).

RULES:

1) Filter out recession dates using RecessionAlert real-time model.

2) Fund universe: FATEX, FBTTX, FDXAX, FEIRX, FLSTX, IBB, IYR, IYE, IYW.

3) Fund selection: highest 52 week ROC, single position.

4) Exit using simple breadth condition or 6x ATR profit stops.

RESULTS:

Annual return 27%, Max. DD 12%, Sharpe 1.5

blog figs

Ticker Trade Date Price Ex. date Ex. Price % chg
FEIRX Long 6/28/1996 9.79 9/13/1996 9.86 0.72%
FEIRX Long 9/20/1996 9.92 11/1/1996 10.13 2.12%
FEIRX Long 11/15/1996 10.51 1/3/1997 10.7 1.81%
FEIRX Long 1/10/1997 10.86 2/7/1997 11.05 1.75%
FEIRX Long (profit) 2/21/1997 11.2 6/13/1997 12.46 11.25%
FEIRX Long 6/20/1997 12.5 7/3/1997 12.77 2.16%
FEIRX Long 7/11/1997 12.78 8/29/1997 12.73 -0.39%
FEIRX Long 9/5/1997 13.06 10/17/1997 13.23 1.30%
FEIRX Long 10/24/1997 13.17 11/7/1997 12.95 -1.67%
FEIRX Long 11/21/1997 13.26 2/6/1998 13.8 4.07%
FEIRX Long 2/13/1998 13.9 3/27/1998 14.97 7.70%
FEIRX Long 4/3/1998 15.26 4/24/1998 15.02 -1.57%
FEIRX Long 5/8/1998 15.11 7/24/1998 15.07 -0.26%
FEIRX Long 8/21/1998 14.13 10/16/1998 13.81 -2.26%
FEIRX Long 10/23/1998 14.03 12/11/1998 14.82 5.63%
FATEX Long 12/24/1998 18.25 2/5/1999 19.59 7.34%
FATEX Long 2/12/1999 19.5 6/11/1999 20.82 6.77%
FATEX Long 6/25/1999 21.75 8/6/1999 21.77 0.09%
FATEX Long 8/13/1999 22.59 9/24/1999 23.38 3.50%
FATEX Long 10/1/1999 23.43 10/29/1999 25.04 6.87%
FATEX Long 11/5/1999 26.44 12/17/1999 32.2 21.79%
FATEX Long 12/23/1999 33.88 1/7/2000 32.64 -3.66%
FATEX Open Long 1/14/2000 34.13 2/25/2000 39.32 15.21%

Notes:

1) Testing this internet bubble period on the technology fund only (FATEX) returns 37% annually!

2) All trades in 1999 have maximum gains over 10% but the exits allow most of the gain to evaporate.  Would monitoring the market state manually help, or hinder results?  Or is there a better exit trigger than the crude one I am using?

The futility of expecting bond returns over the next decade

Several quants around the blogosphere have discussed the future of portfolios blending equities with bonds to reduce volatility without lowering returns.  The tailwind of interest rates falling from 20% to zero has made these strategies very attractive.  The consensus view is for under-performance as bond yields reach the lower bound.

Now there is a qualitative analysis to back those opinions.  Leuthold Weeden Capital Management (Doug Ramsey, Eric Weigel) recently published an article at advisor perspectives showing that “the correlation between current yields and the subsequent 10-year total return is a stunning 0.96 based on monthly data back to 1930!”.

In summary, bonds are likely a poor home for long-term money over the next decade.

Super high Sharpe ratio strategy

Along the same lines as all the previous tests, just looking at better or simpler exits:

1) Test from 2003 to present (buy the dips), avoiding Nov 2007 to Jul 2009 recession per RecessionAlert real-time model.

2) Select from Fidelity 401k fund universe of FATEX, FBTTX, FDXAX, FEIRX, FLSTX (for reasons discussed previously) using 52 week ROC momentum metric.

3) Exit using profit or trailing stops of ATR multiples (see chart).  No other exits.

Result: Annual return 24%, Max. DD 14%, Sharpe 4, Exposure 54%

This is the effect of varying the main model parameters (note the small z-axis range):

blog figs

Sharpe definition:

The Sharpe ratio (also known as the Sharpe index, the Sharpe measure, and the reward-to-variability ratio) measures the excess return (or risk premium) per unit of deviation in an investment asset or a trading strategy, typically referred to as risk (and is a deviation risk measure), named after William Forsyth Sharpe.

Exploiting the low volatility anomaly

Eric Falkenstein has described the concept for low volatility (low beta) investing through his book, blog and website.  There are several ETFs available such as SPLV.  Personally, I find the concept and results fascinating.

I used Eric’s historical data to test the effect of holding a low beta instrument outside of recessions:  Results are not quite as smooth as the momentum test but bear in mind that this is a single instrument.  There may be an opportunity to improve with either Robeco enhanced funds (if accessible) or a universe of low vol. ETFs ranked by momentum.  There is some yield which boosts returns slightly also.

Annual return 19%, max. peak to trough drawdown 15%, Sharpe 2.

blog figs

‘Timing the dips’ (buying when >1% stocks down >25% in a month) results in:

Annual return 16%, max. peak to trough drawdown 7%, Sharpe 2.6 (Exposure 50%)

Trade Date Price Ex. date Ex. Price % chg
Long (profit) 8/8/2003 99.4 10/31/2003 113.5 14.19%
Long (profit) 4/30/2004 131.1 12/23/2004 155 18.23%
Long (profit) 4/1/2005 163.2 9/16/2005 180.8 10.78%
Long (profit) 10/21/2005 171.4 2/24/2006 197 14.94%
Long (profit) 7/28/2006 181.5 1/26/2007 210.5 15.98%
Long 7/27/2007 224.7 9/28/2007 223 -0.76%
Long 3/13/2009 137 5/8/2009 150.9 10.15%
Long 7/2/2009 156 8/28/2009 165.6 6.15%
Long (profit) 11/6/2009 170 4/16/2010 195.7 15.12%
Long 5/7/2010 185 3/4/2011 208.1 12.49%
Long (profit) 3/18/2011 204.6 7/1/2011 232.1 13.44%
Long 8/19/2011 203.8 11/11/2011 217 6.48%
Long 12/2/2011 214.2 2/17/2012 225.3 5.18%
Long 5/25/2012 233.1 7/6/2012 243.1 4.29%
Long 8/10/2012 242.3 8/31/2012 244.2 0.78%
Open Long 11/2/2012 244.6 3/1/2013 265 8.34%