Mar

20

The following (copied from Mebane Faber) is so counterintuitive that it's worth considering. I don't think in these terms, and there could be outliers that explain the phenomenon. But (if they did the arithmetic correctly), it is what it is….

Source:

Should You Buy at New Lows? Or New Highs?

So we tested which strategy works better: Buying near 52-week lows… or buying at 52-week highs. We looked at nearly 100 years of weekly data on the S&P 500 Index, not counting dividends. You might be surprised at what we found… After the stock market hits a 52-week high, the compound annual gain over the next year is 9.6%.

That is a phenomenal outperformance over the long-term “buy and hold” return, which was 5.6% a year. On the flip side, buying when the stock market is at or near new lows leads to terrible performance over the next 12 months… Specifically, buying anytime stocks are within 6% of their 52-week lows leads to compound annual gain of 0%. That’s correct, no gain at all 12 months later. Using monthly data, our True Wealth Systems databases go back to 1791.

The results are similar… Buying at a 12-month high and holding for 12 months beats the return of buy-and-hold. And buying at a 12-month low and holding for a year does worse than buy-and-hold. Take a look… 1791 to 2012 All periods 4.3% New Highs 5.5% New Lows 0.9% The same holds true for a more recent time period, this time starting in 1950… 1950 to 2012 All periods 7.2% New Highs 8.5% New Lows 6.0% History’s verdict is clear… You’re much better off buying at new highs than at new lows. You might not agree with it… but it’s true.

Victor Niederhoffer writes: 

That's a shocking result. Heavily weighted one might think to the depression period and the 2008 period, and probably not taking into account durations from hitting the new lows. i.e. the 1st new low in a period or the tenth. Probably even more copacetic to the trend followers with individual stocks.this is how Rocky and I first met, but I don't think he remembered it. A loss of mine was reported in the papers and Rocky wrote to me to memorialize what a woeful idiot I am. I wrote back saying "You seem to take great pleasure in my losses et al". But as you know, you can never win a dispute with Rocky. Now we're friends again.

Scott Brooks writes: 

I have been privileged to buy the low and sell the high on multiple occasions. It's all those other darn trades in between that drag down my return.

I had a friend tell me once that there are 50 perfect days in a year….. a bluebird sky, cool temperature, perfect humidity, occasional slight breeze (you know the kind of day I'm referring too).

Most people make the mistake of living for those perfect days. The key to a great life is to make the best of the perfect days when they arrive. And the way you make the best out of those perfect days is to make the best of the other 315 less than perfect days per year.

It's about having a good positive attitude so you can make the best out of whatever you get. And they way you do that is through practice…..you practice and practice and practice…..until a positive attitude and making the best of things becomes habit.

So make the best out of our less than perfect trades, for they are the ones that are ultimately going to define you as a person and a professional.

Jim Sogi writes: 

Amen to what Scott says. In surfing you got to go on the crappy days so you are in good shape when the big waves come. You can't just wait, like many do, for those rare perfect days. Then they are so out of shape they can't make the drop and have no legs.

Alston Mabry writes: 

I'll assume the data for 1791-1950 is more troublesome, so let's just consider this result:

1950 to 2012

All periods 7.2%

New Highs 8.5%

New Lows 6.0%

The obvious question is: When do you sell?

Jordan Low writes: 

It seems that there is never a good time to sell. You can beat 6% by say investing in short term bonds. It has to be short enough for the turnover of the strategy, so say duration of less than 1Y.

Also, the new high strategy has not really worked since 2000 with the market risk-on/risk-off, so are we in a new "regime"? Or do I keep to the strategy and pray that I will end up ahead 60 years from now — i.e. not a repeated game, you get one dice roll!

Ed Stewart writes: 

I have noted that including historic t-bill rates or alternative short term rate benchmarks as an estimate for return while in cash dramatically alters the return of long term timing models. However, I am not sure if t-bill or similar has been a fair estimate of cash holding returns - I am sure others no much better than I do.

With regard to the article idea, It does seem to be the logic of a simple trend model - something like Long on first close in top X% of range 52 week range, Flat when close in bottom X% of 52 week range. A bunch of rule sets similar to this (some type of very long term trend indicator or look-back) seem to give similar results - and like was mentioned much of the benefit comes from missing a small number of significant market declines.

In other periods (like the 90's) the models can trigger whip-saws that would likely have frustrated many "believers" at that time into giving up on them - which of course means they would have missed the benefits that accrued since 2000.

In thinking about timing models, one real benefit is that they provide a framework for the panic instinct while including a signal to get back in. The problem with the public is that they can panic, become traumatized, then never get back in until years have passed (if ever). In other words, even if one is skeptical about the future performance of timing models, such models might be a useful tool if the realistic alternative is very poor money-weighted returns with a near certainty (rather than the theoretical return of buy and hold).

A commenter writes: 

 I take the view that when any sign is known to the market, it will start to disappear; and when it is no longer a sign, it will start to reappear.

I would think it applies to this case as well. The advantage of buying at market high is not news. When was it first exploited? Were the turtles first known to the public for doing this? In the 90'es?

But anyhow, I think a plot of the returns across the time span is more meaningful (and clearly more revealing) than the average. With that, I presume that we will see the advantage of buying at market high is diminishing in the recent decade. More meaningful I think would be how much it has diminished so that we can anticipate the future when it returns.

Russ Sears writes: 

I suspected that the results depended on the period looked at. Kim gave the 250 day period results. But what happens in other periods. I looked at the S&P index from 1950-2013, with cut-off dates determined by period's length. I defined it a "first new high" if there were X day high within X days. and looked at the next X days log normal returns.

5 day period
avg     0.14%    Stdev    2.18%
         count  avg next period      T
new low  1006   0.06%                (1.24)
New high 1004   0.29%                 2.14

25 day period
avg      0.71%   Stdev    4.77%
         count  avg next period      T
new low   208    0.99%                 0.85
New high  179    0.86%                 0.44

50 day period  
avg     1.40%   Stdev     6.75%
        count   avg next period      T
new low  100     1.16%                (0.36)
New high  96     0.87%                (0.78)

100 day period  
avg      2.78%  Stdev     9.67%
        count   avg next period      T
new low  44      3.69%               0.63
New high 35      4.51%               1.06

500 day period  
avg     13.42%  Stdev    21.96%
        count avg next period      T
new low    8      8.33%              (0.66)
New high   7      7.24%              (0.74)


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search