Ok, first let me say I am not claiming originality here but let's explore this and see where it goes.I have been writing since day one about taking defensive action when the market goes below it's 200 DMA.
$10,000 bought into the S&P 500, and just held, 25 years ago would today be worth $83,800. Had that same $10,000 gone out when the SPX went below it's 200 DMA and went back in when it went back above the 200 DMA would be worth $126,500.
Following this to the letter the investor would have sold in October and still be out. My numbers. BTW, are not scientific, I was looking at a Yahoo chart going ten years at a time so the numbers are close but not exact. Anyone wanting to do it exactly and post the result is more than welcome.
The chart above includes the Swiss franc versus the USD (note it is charted backwards to show the franc going up or down to make the point easier to understand). The chart goes back only as far as Yahoo has currency data. In looking at the two extended periods where SPX went below its 200 DMA (11/2000-3/2003 and 10/2007 through today) the swissi went up 26% and 18% respectively.
So the theory would be stay in SPY while it is above its 200 DMA and swap into the Swiss franc when SPY goes below its 200 DMA. The reality might be a little different. First, so far all we've done is look in the rear view mirror. I would also note that the dollar rallied against the swissi during the relevant periods in 1987, 1994 and 1998.. So either the world changed between 1998 and 2000 with respect to the dollar and or the Swiss franc (this is entirely possible) or the last two times were just a coincidence.
The bigger macro is to explore for the possibility for a better mousetrap. I have unyielding faith in the 200 DMA as an indicator but taken to the extreme of getting out entirely (not practical but this post is about a theory) and finding a currency likely to go up while out of equities is interesting to ponder. Maybe instead of a currency it should be an absolute return product?
Another problem with the S&P 500 or any broad domestic equity proxy is that there is visibility for the US market to lag other markets for a while (nothing apocalyptic but maybe 5% a year instead of 10% for example). Perhaps instead of SPY maybe the long equity exposure should be the iShares All Country World Index ETF (ACWI).
When ACWI is above its 200 DMA then it's all in. When ACWI is below its 200 DMA then it's out of ACWI and into some foreign currency (what about the Sing dollar or the renminbi?) or some sort of absolute return fund?
I find this sort of thing to be fascinating. The point is not should I do this but how can analyzing a theory like this help my actual portfolio.
What do you think would be good proxies for the equity exposure and the defensive exposure?
16 comments:
Morning, Roger--I apologize for being WAY off topic, but I'd love to get your take sometime on the new Vanguard managed payout funds. The more aggressive funds seem to exceed the conventional wisdom of a 4% withdrawal rate that your blog discussed yesterday. I'm trying to sort out what the advantage is over a well-balanced and diversified indivual portfolio. Thanks very much.
I have no opinion on the Vanguard version of this concept versus the Fido version ( so not picking one over the other) but the concept is fantastic. I would expect more participants in the space (is there a third now? don't remember). Annuities are onerous and expensive.
Finding similar attributes (the positive attributes) of annuities in a cheaper, easier to access wrapper could be very important.
It seems to me though that it could take quite a few years for the concept to gain traction and acceptance meaning there may not be a lot of firms in the space. Schwab would seem like an obvious candidate to be working on these in the lab.
A great question on one of my favorite subjects. IMHO, Mebane Faber has done a nify job demonstrating that trend following in five uncorrelated asset classes generates about 3% excess annual returns with the same risk as an all equity portfolio. I believe he uses ETFs for the S&P, MSCI EAFE, commodities, real estate, and 10 year US treasury bonds.
He has a recent post on his site about an "all in" strategy using the three best performing asset classes over the past year. If I recall correctly, it generates returns similar to the trend following moving average approach, but maybe(?) with more volatility.
Interesting topic, thanks!
Doesn't being "all in" or "all out" violate your own rules about not making big bets?
Hi Roger
I like mechanical systems because they are testable and can be executed unemotionally.
About a year ago a spent some time playing with different assumptions of being in market only when above simple moving averages, selling when below. It's an effective way to protect yourself from big market downdrafts, but lots of times you get slowly killed as market just thrashes above and below SMA as it moves sideways. I used the S&P 500 since 1952 since that was easy data to get from Yahoo. A few conclusions:
1. You can vary number of days in moving average (I tested from 2 to 500) but interestingly I got similar results over wide ranges of timeframes. Also, best results were shortest timeframes, but then you do a lot more transactions (52 round trips per year in case of 2 day SMA) so slippage/commission and taxes higher.
2. Sideways market transaction cost (in and out losing money in flat market) can be minimized with a time filter, e.g. only trade once a week, or once a month. But then the magnitude of $ losses goes up for trades that lose money. It's a net wash.
3. I also tried a guard band (i.e. hysteresis - so trade only if above/below SMA but some percentage) that did reduce transaction quantity but did not meaningfully improve compound growth.
4. Bottom line was that long term results very similar to buy and hold, with somewhat lower drawdowns, and less risk in sense that you were out of market for about 30% of the time, so better Sharpe ratio. This was true for wide range of days in SMA, and different guardbands.
The only major caution I would point out is that I do believe starting valuation matters - the 1952 to 2007 data I backtested coincided with overall good economic dynamics and broad reduction in risk premiums. That is unlikely to repeat, and I think there is a good argument that it may reverse going forward. If so, an approach like this which is mechanically more defensive may really outperform in the future.
Michael
anon 6:43 I say a couple of times that this is a theoretical exploration and that it may not be practical (certainly is not practical the way I prefer to do things) so I think you may have missed the point.
Michael I may have met my match in stock market nerd-dom in you, lol. thank you for sharing your results.
Aha... a year ago when I mentioned that a ton of research indicates that you can reduce volatility and slightly improve performance by staying out when the market is below the 200 day MA, I was ridiculed.
five months ago when I said "i am sure roger will eventually come around," my comment was termed a "fine heckle"...
I knew it is wouldn't be too long before it finally happens :)
Roger, both Tom K and I welcome you to the world of trend following.
Isn't this basically what Al Thomas suggested in his book? Except that he also suggests the use of a second trend indicator in/out decision tool to corroborate the 200 DMA for the S&P 500, before making the switch.
One can definitely sleep better at night with this approach as one approaches the end of a wage-earning career.
Tom
I agree with Michael about the "guard band". I did some of my own testing of the 200 DMA a couple of years ago. If you do not have some sort of backup trigger mechanism then you have many periods where you are whipsawed back and forth around the average line, causing unnecessary transaction (and taxation if that applies) costs.
I used +/- 1% as my "guard band". You can read the entire article on my site here - Playing 200 Day Moving Average Defense. There is a comprehensive listing of my findings.
I don't have the tools, data, or smarts to make a meaningful answer to your question. It simply seems logical to me that when one makes a decision to be all in or out, he'd want two assets that are negatively correlated.
Looking at SPY, over just the past year, SDS (-0.99), SHY (-0.88), and TIP (-0.86) show strong negative correlations. Using EFA as an international proxy, the negative correlations for SDS, SHY, and TIP are similar but slightly weaker than vs. SPY.
I'm sure there are a myriad of real estate, commodity, and currency etfs/stocks that could be tested as well and over longer time frames but the logic of swapping to a negatively correlated asset class would remain the underlying principle.
Michael,
Great comment. I particularly think #4 is important - market timing is mostly about reducing risk.
CXO Advisors testing of different moving averages as buy/sell signals vs buy and hold:
http://www.cxoadvisory.com/blog/internal/blog7-19-07/
Hey, Roger, nice post. Simple, straightforward, clear question. I don't have the answer, of course, but it is something I've been thinking about. I just finished reading a paper by Reinhart and Rogoff, "This Time is Different: A Panoramic View of Eight Centuries of Financial Crises" (30-Mar-08), in which it is made abundantly clear that this time is NOT different. This gets me thinking that choosing the safe haven for our funds when the 200DMA trigger is pulled might be best based on an indicator (as yet un-calculated, at least by me) that pulls various economic and financial factors together to give a sort of financial health rating. If the US equities markets have caught a cold, what are the factors that would tend to isolate other economies/markets/currencies from the contagion? Current account balance, for one, comes to mind. A CPI of some sort would be another (are those published somewhere for most currencies?). If we had a good handle on such a rating, it should be quick and easy to figure out which potential haven is likely best and then execute the trade, or, say, pick the top three for some diversification. Is this idea in line with where you're heading?
If I understand your question, a stock market indicator based on economic inputs would be difficult for me to feel confident in because sometime market cycles and economic cycles converge and sometimes the diverge.
Data can be lousy yet stocks go up. Markets move in the direction they should not all the time.
OK, let me back up. For purposes of discussion, you posed the Swiss franc as a potential parking place for funds when the 200DMA trigger got us off the SPY highway, showing that the $/Sf exchange rate tended to appreciate during the SPY downtimes. I'm asking, what are the factors (economic, financial, or other) that might predict that relationship? Why did the Swiss franc appreciate vs the dollar during those times? A better understanding of that might help us pick a better parking spot next time, or at least help us feel more confident in our choice of haven, be it the swissi or whatever. Clear as mud?
oh.
I looked at quite a few currencies before deciding to write about the swissi.
In times of unusual uncertainty money tends to flow into the swissi which is perceived as a safe haven.
this is easily seen in the aussie/swissi pair.
Looking forward maybe there would be an investment product that could capture safehaven currencies.
as far as how to pick something I would think choosing from surplus countries that do not export a lot to the US might be a good place to start looking.
Post a Comment