In July 2010, Bianco Reserch highlighted an analysis by Ron Griess at TheChartStore.com of every instance of the death cross (aka dark cross) since 1930.  Below are his results.

The first table shows the performance of the S&P Composite for the time periods listed when the 50-day moving average is falling and crosses a rising 200-day moving average. (Dark Cross)

The second table shows the performance of the S&P Composite for the time periods listed when the 50-day moving average is falling and crosses a falling 200-day moving average. (Death Cross)

The bottom line: There are lots of things to worry about, but this is not one of them — its a low probability indicator:

>

Dark Cross

50-day ma falling across a rising 200-day moving average.

~~~

Death Cross

50-day ma falling across a falling 200-day ma

Category: Markets, Really, really bad calls, Technical Analysis, Trading

Please use the comments to demonstrate your own ignorance, unfamiliarity with empirical data and lack of respect for scientific knowledge. Be sure to create straw men and argue against things I have neither said nor implied. If you could repeat previously discredited memes or steer the conversation into irrelevant, off topic discussions, it would be appreciated. Lastly, kindly forgo all civility in your discourse . . . you are, after all, anonymous.

23 Responses to “Worry About Important Things — Not The Death Cross”

  1. [...] Ritholtz posted some interesting research: Worry About Important Things — Not The Death Cross ← Two Types Of Real Estate Buyers From China No comments [...]

  2. Okay, how about this… During the market’s recent decline the differential of NYSE New 52-week highs and lows was nine times worse than during May 2010, this despite the NYSE Composite Index remaining well above its May 2010 low.

    This same differential has performed pathetically since March ’09 bottom, revealing, first, absolutely no convincing leadership as the market advanced 100%, and second, persistent deterioration since May 2010 leading right up to this year’s top. This is particularly noteworthy given that, during the ’08 swoon a solid, absolute majority of NYSE-listed issues set new 52-week lows. At no time in the market’s recovery did anything even remotely approaching this negative extreme register on the positive (i.e. new 52-week highs). Why is that? Stocks were being distributed to weak hands. Persistently diminishing volume during advancing periods off March ’09 bottom confirms this. Likewise, the recent market swoon is very well explained by this technically conspicuous backdrop and confirms long equity is dominated by weak hands.

    Thus, in this context the “Death Cross” is rather significant.

  3. [...] Ritholtz at The Big Picture, “The bottom line: There are lots of things to worry about, but this [the death cross] is not [...]

  4. craigH says:

    Thanks Barry – Nothing like actual data to head off a bad argument, at least as a standalone indicator.

  5. williamjohndonahue says:

    Everyone always gets the death cross wrong. It’s not based on the 50/200 day DMA, it’s the EMA. Big difference. The real death cross (based on the 50/200 EMA) is an extraordinarily reliable indicator for the S&P500. Don’t believe me though, just look for yourself.

    We did have a death cross last September, but the 50/200 EMA were basically going sideways, and it turned into a golden cross shortly after.

    We don’t have a death cross yet, but will in the next few days if the market doesn’t rally up to the 50/200 EMA (both are about the same level). So far, this potential death cross looks similar to the one in September 2010. If the market rallies up the the 50/200 EMA, both indicators will level off and go sideways. If we don’t see a market rally up to the 50/200 fairly quickly, then I would be concerned.

  6. Excellent analysis on the Dark & Death Crosses, as always, overall portfolio risk management is key, especially as volatility increases.

  7. machinehead says:

    The tables are interesting. (Table 2 should be labeled ‘Death Cross.’) But it’s hard to conclude to anything about the validity of the strategy without knowing when the sell signals were cancelled by offsetting ‘white cross’ buy signals (or ‘truckers little helpers,’ if you prefer).

    A 50-day moving average as the ‘fast indicator’ is probably way too slow. It helps to suppress whipsaws, but at a huge cost in accuracy. There are better ways to make a moving average less twitchy.

  8. [...] Worry About Important Things – Not the Death Cross (Ritholtz) [...]

  9. nofoulsontheplayground says:

    Rather than a low probability indicator, I prefer to refer to it as a lagging indicator. It’s an indication of what has transpired, but has little predictive value.

  10. SINGER says:

    agree with william re: the EMA’s being what’s important…

    it is also a lagging indicator on the time frame it deals with — here the daily — but then its leading on the weekly and monthly frames, i.e. if u have the 50 below the 200 on the daily, then that usually means that has yet to unfold on the larger frames but may do so, and has long since done so on the hourly frames which may even be countertrending up due to the extreme OS nature of a market that gets the 50 under the 200 in that timeframe in the first place…

  11. Moopheus says:

    “We did have a death cross last September, but the 50/200 EMA were basically going sideways, and it turned into a golden cross shortly after.”

    A couple of guys in white shirts, bow ties, and suspenders came to my door once with a similar story. Both named, oddly, “Elder.”

  12. constantnormal says:

    whotta tease … “Worry About Important Things — Not The Death Cross”

    … and then not one word about what those “Important Things” might be … aside from the obvious health, family and friends, I have no firm convictions about those things, other than that the 200 dma seems to be an important after-the-fact indicator that one is screwed (assuming one owns the thing that has gone through the 200 dma). Perhaps I should be looking at ema instead of dma, maybe that will prove to be a more useful tool …

  13. DeDude says:

    RiskAverseAlert;

    Interesting information and opinion. So if I understand you right then you suggest removing the events that do not include a certain amount of 52 week highs in the ?? number of days before the cross. Then it becomes a much more convincing indicator? I guess that is similar to the suggestion that if the cross occurs after a long period of a rather flat market then it is not reliable as a predictor.

    Mathematically I guess these crosses just indicate a change in the momentum of slope change, so they are great after the fact “predictors”; with build in confirmation bias depending on how many people are true believers. If I were to try improving them I would add the direction of macro-economic data (does the economy have a positive or negative momentum). Also how much of the market is driven by stupid machines (they would swallow death crosses hook, line and sinker). Indeed, based on that the death cross may work this time around.

  14. machinehead says:

    A 50-day EMA will respond a little faster to a large volatile move than a 50-day SMA (simple moving average). But then, so will the 200-day EMA. So the combination of two EMAs is not necessarily more timely than two SMAs.

    To really make it responsive, you need more like a 5 or 10 day MA for the ‘fast average,’ plus a whipsaw filter (for instance, a 1% hysteresis band on either side of the long MA).

  15. Alex says:

    Thanks Barry,

    I will spend more time looking for the Hinderberg Omen, or of course further evidence that aliens are living in my basement (my wife just doesn’t understand!).

  16. JasRas says:

    Ah, I like it when everyone else is looking at the wrong thing! Thank you CNBC for perpetuating this inaccurate indicator. Thank you, thank you, thank you…

  17. Mike C says:

    I’ll take the other side of the argument on this note…..2 main points.

    1. The tables don’t tell you perhaps the most pertinent data points you really need to assess using this which is max drawdown from when a sell signal is issued, and at what price level does the countervailing golden cross buy signal get you back in. I fail to see the relevance of a specific percentage change some arbitrary amount of time later. Case in point. Take the 10/30/00 signal which was down 24% 1 year later. OK. But we know the 00-02 bear market was roughly a 50% drawdown peak to trough although I’m not sure and too lazy to check how far the S&P 500 was already down from peak when the signal was issued on 10/30/00. What you really would want to see is the equity curve of following all the sell and buy signals and comparing that to passive buy and hold. Basically axiomatically, the death cross signal is going to get you out of and keep you out of big bear markets like 29-32, 37-38, 73-74, 00-02, and 07-09. The question is does that make up for all the times you get chopped up everywhere else. I don’t know? It is actually been on my to do list to do this analysis.

    2. Perhaps recency effect, but fact of the matter is this worked for the 00-02, and 07-09 bear markets in keeping you from experiencing the full brunt of those 50% and 58% declines. If you assume the same overarching macro conditions are still at play with large smooth cyclical trends up and down (unlike say the chop from 1976-1982, or 1938-1942) then perhaps this signal is once again warning of a much larger move down. Here’s the million dollar question. Are we just going into a growth slowdown? Or are we going into a recession (Hussman has already called it)? If it is a recession, is it just a mild one, or another deep recession like 07-09 where earnings come crashing down? What I do know for sure is by the time a recession is obvious and forward estimates have come down substantially it will be too late to do anything….the market will have already gotten killed and that will probably be the time to buy. But we got the signal now, and I guess the question is the bottom in at 1120, or are we going to 1000-1050, or below 1000 or even back to the 2002/2009 lows at 700-800.

    Barry, I would be interested to know if you think the crash we just had is more like 1987 where that was it, a sideways base was built, and then the uptrend resumed, or if this is 2008 where the crash led to another substantial downleg.

  18. steve says:

    How about this: http://www.ritholtz.com/blog/2010/11/some-frequently-asked-qa-about-the-us-stock-market/ – is it about context or is it really a low value indicator?

    Ned Davis has done some superb research on Golden Crosses and Death Crosses. In sum, they have predictive value but the degree of value depends on whether the bull or bear market is secular or cyclical. From Ned’s data base we learn that, “Since 1929, S&P 500 Golden Cross signals that have occurred within secular bull markets have been profitable 76% of the time with a median gain of 19.9%.” That is a powerful statistic. However, Ned also notes, “During secular bear markets, Golden Cross signals have only been profitable 58% of the time with a median gain of 1.3%.”

  19. christopher.cosenza says:

    To add to Mike C and Steve’s comments, it does matter if its a secular bull or bear. Barry this aligns itself with David Rosenberg’s comment that secular bear markets are about preserving capital and bull markets growing capital. using the death cross whether ema or sma is virtually irrelevent. Ema will likely chop you up more then sma but ema MAY signal a decline sooner but not always. Another aid would be to look for a failing rally after the signal has been triggered at the 50 or 200 dma to validate the signal. Last summer we never really had that negative sloping 200day for long and along with a concrete test and failing rally. Even if you did follow those triggers you would triggered back in with the golden cross not too much later. Like most things however, people need to be careful using it as a stand alone signal of a pending major decline. But keep it in the toolbox.

  20. macrotrader603 says:

    Great thread. Some very good comments in here.

    I haven’t tested the 50/200 cross recently, but there are much more reliable MA crossover which are shorter in nature.

    They are currently down.

    Anything is possible, but the 50/200 is more of a media event.

  21. Mike C says:

    but there are much more reliable MA crossover which are shorter in nature.

    20/50 EMA?

  22. crunched says:

    Actually, couple the 50/200 cross with markets that have broken a trendline of at least two years and have not made a new market high in four years and the results aren’t pretty.

  23. Thanks for looking at the facts. Since the 50/200 cross is that widely published, it cannot be else that it loses some of it significance.

    However, remember that any long-term bear market has somewhere a 50/200 cross in the beginning. There is no escape from that. Long-term investors can escape the bear market when acting upon a signal like this and avoid the draw-downs of buy and hold investors during those years.

    But sometimes the 50/200 cross is a false bear market signal of course. Nothing is perfect.

    For a long-term investor it is not a problem to act upon a false signal as long he/she can correct quickly after that. If a long-term investor wants to use the 50/200 cross as a sell signal, he/she needs also to look also at a reliable long-term buy indicator to correct for the situations that the signal was false.