Trading the Golden Cross
This is a test of that most venerable of market timing indicators, the “Golden Cross”. The Golden Cross occurs when the 50-day moving average crosses over the 200-day, and to some technicians signals the start of a bullish bias in the market.
[logarithmically-scaled, growth of $10,000]
In the graph above I’ve assumed a trader went long the S&P 500 at today’s close when the 50-day simple moving average closed above the 200-day (otherwise to cash) from 1930 to present. The strategy is drawn in red versus buy & hold in grey.
Geek note: see end of post for assumptions about return on cash and trade frictions.
As the graph above shows, this simple crossover strategy has more or less kept pace with buy & hold over the last 80 years. But more importantly, as the stats below show, it’s done an even better job reducing drawdowns by limiting losses when the market turned bearish.
This general observation can be applied to all similar trend-following strategies (ex. 10/50-day, 2/10-month, etc)…
Long-term moving average crossovers haven’t been useful for generating outsized returns (for that you’d need a more active strategy), but they have done a good job protecting investors from protracted downturns.
As a hyperactive trader I personally wouldn’t trade such a slow-moving strategy; I think there are just too many opportunities within the trend. However, for investors of the longer-term variety, I think the Golden Cross is a significant improvement over straight buy and hold.
Note: you can track the Golden Cross daily on the free State of the Market report.
[Click for a summary of our posts related to the Golden Cross]
Happy Trading,
ms
Test assumptions: (a) I’ve calculated the moving averages using the S&P 500 cash index (which traders generally use), but calculated returns based on daily dividend-adjusted data (dividends interpolated from quarterly data), (b) results are frictionless (i.e. do not account for transaction costs/slippage) but could be closely reproduced in today’s market using mutual funds less part of an annual expense ratio, and (c) I’ve assumed a return on cash of half the nearest 13-week Treasury.
. . . . .
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 | 10 Comments
Excellent data! Thank you for revisiting this.
One crucial data point missing here is the number of trades. How many times did the moving averages cross over between 1930 and today? That could help compare this to other moving average strategies (e.g. 10-month SMA) and also assess trading friction.
RE to Scott: oops…being a MF trader, I don’t think in these terms. The test above (1930 – 06/2010) traded 1.13 times per year (0.57 round-trips). I’ll compare this to other variations (ex. 10-mo SMA) in a future post. Good comment, michael
Is that DOW30 or S&P500? On http://finance.yahoo.com/echarts?s=^GSPC the available data for S&P500 is since 1950.
RE to blink: the test above is on the S&P 500.
I included the link to the Yahoo quote to make clear that I’m NOT using the cash index (i.e. the returns that most sites like Yahoo report) to calculate returns because those returns are not adjusted for dividends.
michael
hi Michael,
i see you are interpolating dividend data daily.
Does that make sense ? i.,e if you are out of the market during when a dividend is paid (which can happen in any market timing strategy) – dividends should NOT be factored in for those days – which may happen if you interpolate dividends quarterly.
Probably doesnt make too much different, but assume you held for 2 months in the quarter, and your market timing got you out, but the dividend got paid in the third month, you would be accounting for that as part of the two months you were in, correct ?
Its very likely that i am misunderstanding what you are exactly doing.
thanks for the great blog, again !
RE to bgpl: good comment. We’re talking about the S&P 500 here, so dividends are going to be spaced out (so the example above isn’t an issue).
You do bring up a good point however that (as far as I know) there is no way to truly accurately account for dividends daily on the S&P 500 very far back in history.
The method I’ve approached is a SWAG, but comparing the results to something that is accurately adjusted daily (ex. the Vanguard 500 index fund) shows very little deviation. In other words, I think the SWAG is close enough to not materially alter results.
michael
the interesting thing is that this “golden cross” and “death cross” happens all the time on smaller time frames….so we’re really talking about relativity when it comes to this cross (granted that many times on the smaller times frames it too can be a rather accurate trade)…however on the DAILY chart we have not yet seen a cross yet on the /ES so I’m still bearish overall and view this bullish run currently as a retracement in a downtrend. If the cross happens on the daily chart then that would change things for sure but boy I would be shocked f that were to happen. But then again who knows!! We are up to some resistance now on the daily chart so we shall see if we bounce or break there. What do you think Michael?
A bit late for the discussion but… what would be the chart if you go short instead of just staying in cash?
Just something to think about…
Hello Terco: this was tested in another post (which is escaping me at the moment). In a nutshell, this would have been effective since the turn of the century b/c of the very clearly defined bear/bull/bear/bull pattern the market has followed. But for most of the market’s history this has not been an effective approach and has, in my opinion, hurt performance. michael