Testing Zweig’s 4% Strategy
This is a follow up to CXO’s analysis of Martin Zweig’s 4% Strategy (h/t TWS).
This is a simple weekly strategy: go long at the close on the last day of the week when the Value Line Arithmetic Index (^VAY) closes at least 4% higher than the previous week, and close the position on the last day of the week when it closes at least 4% lower. Note: at end of post I address other variations of the strategy.
Results “trading” the Value Line index (red) versus buy & hold (grey) since 05/1984:
[logarithmically-scaled, growth of $1]
For the moment, I’m being generous and ignoring transaction costs and slippage, and assuming a return on cash of half the nearest 13-week Treasury.
Numbers for the number lovers…
Timing this one particular index, I would characterize Zweig’s 4% rule the same way I would most trend-following strategies. They do little to increase returns, but do a good job at limiting drawdowns.
The next graph makes this more clear. Here I’ve graphed the rolling 1-year max drawdown of Zweig’s strategy (red) versus buy & hold (grey). Note how losses are consistently limited.
The fly in the ointment is that (at least to the best of my knowledge) the Value Line Arithmetic Index isn’t tradable.
So below I’ve tested the same strategy, using the Value Line index to signal trades, but executing those trades on the (dividend-adjusted) S&P 500. The strategy is colored red and buy & hold grey.
[logarithmically-scaled, growth of $1]
The strategy has been considerably less effective under this more realistic scenario.
It did a poor job navigating the 1990’s bull market and 2000-02 bear (unlike most simple trend-following strategies), and while it effectively sidestepped the 2008 crash, it has failed pretty badly to keep up since.
A similar conclusion would be reached trading other market cap weighted indices like the Wilshire 5000, and/or using the Value Line Geometric Index in place of the Arithmetic one (not shown for the sake of brevity).
I assume that at least part of the disparity in performance is because the constituents in the Value Line index are equally weighted (rather than market cap weighted), meaning smaller issues have an outsized impact on performance. Perhaps this strategy would be more effective trading a broad equally-weighted index.
In any case, I’m ending my test here, because IMHO the 4% rule has a fatal flaw relative to a traditional trend-following strategy like the Golden Cross.
Using the 4% rule, the market could steadily march up or down for a long time, like it did in the 1990’s (up) and 2000-02 (down), and never trigger a relevant signal, because the Value Line index never changed by more than 4% in any given week. Whereas with a traditional trend-following strategy, a bull/bear market will always (for better or worse) eventually trigger a corresponding buy/sell signal.
Happy Trading,
ms
P.S. Big ups to CXO and Zweig for the opportunity to test this strategy. I’m a firm believer in vertical blogging.
P.P.S. As reader Eber Terandst points out below, there is an alternative to the Zweig rule that calls for a buy when the Value Line index rises at least 4% from a “recent” low (as opposed to last week). I’ve based this test on the unofficial Zweig website Zweig’s own website (which compares the indicator to the previous week) as well as CXO’s analysis. Other definitions of “recent” low (i.e. lowest low of the last 2 weeks, 3 weeks, etc.) perform even worse than the test presented here. In short, conclusion stands.
. . . . .
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: Trading Strategies, Trend-Following | 5 Comments
Mmm . . . no. This is not the 4 % Zweig rule.
The real rule ( at least as I understand it and have seen it tested) is that the buy signal comes when the index closes at least 4% higher than the recent low point, regardles of when that low was made.
When so defined, the argument that a market could go up a long time without a buy signal coming is no longer true: the rule will give a buy whenever there is a rise of more than 4% from the latest low.
Having said all that, the “correct” (?) version of the rule gives pretty much the same results as in the backtest shown here.
eber
Hello Eber – I’ve seen a similar alternative – I’m basing my test from both Zweig’s own website (see link in post) which clearly states “from the previous week”, as well as CXO’s original test. michael
P.S. had to double-check myself here to make sure I wasn’t sticking my foot in my mouth. All other definitions of “recent low” (i.e. last 2 weeks, 3 weeks, etc.) perform far worse than the simple “last week” definition on Zweig’s website. I’ve also seen others use vague subjective definitions of “recent low” (lines on charts kind of stuff) but I pile those claims in with Dow Theorists. In short, you’re 100% right, that would resolve the “never triggers a buy in bull market” critique, but indicator is still a bust. Thanks for the follow up. michael
Michael, just FYI the link you gave is actually not to Zweig’s site, though it uses his name.
Hello Alex – well color me stupid – good catch…I’m obviously much better at calculating things than reading them. Post has been updated. michael