I’ve written before about momentum as an asset rotation strategy. I compare the relative momentum of various ETFs (representing diverse asset classes), then “buy” the ETF with the greatest relative momentum. The results are impressive. With only a single holding and about one trade each month, the model is able to vastly outperform the market. In a backtest, from May 24, 2002 until this past week, holding the TSX 60 (XIU) would have grown $10,000 to $17,098, whereas rotating between ETFs would have grown to $74,409 (accounting for $10 trades).
I’ve often wondered whether or not the same thing would work with individual stocks. I’ve tried before, without producing a satisfactory result. However, as I compiled the data (weekly returns for the constituents of the TSX 60), I realized that stock splits hadn’t been accounted for correctly. When a stock split 2:1, the price was reduced by half, causing the return to appear as a loss. When I corrected for stock splits, the results were much more similar to the rotation between ETFs.
I came up with two possibilities that both show encouraging results. First, it is possible to own the single stocks (from the TSX 60) that has the best relative momentum. The outcome is impressive, but the volatility is huge. For example, RIM (RIM) had the best momentum in the middle of 2007, but during that time it lost 64% in a single week. My model suffered from that loss, but recovered by tripling in value early in 2009. While the final result is acceptable, that would not be a fun ride to experience. From $10,000 on January 11, 2002, the backtest results in $134,217 presently.
The other possibility, however, is to produce a portfolio of the five stocks with the strongest relative momentum. The result is much better than the top 10, and not too much less than the top 2 or top 3, but the volatility is much reduced. The other problem that holding only five stocks addresses is turnover. Because we’re looking at momentum, many of the stocks persist (as in the asset rotation model). Turnover is an average of about one round-trip stock trade each week (sell and buy). This keeps trading costs down and, accounting for commission, grows $10,000 from January 11, 2002 to $86,190 presently. The benefit over the ETFs is that all the stocks trade in Canadian dollars.
I plan to begin implementing this model in an RRSP account. Because of the relatively high turnover (1100%), tax accounting would become a nightmare outside of a registered account. Also, to keep the costs down to around 1% of the portfolio, it makes most sense in an account of around $100,000 or more. Not to mention that capital gains are as good as dividends for producing income from a registered account.