Feb

4

The professor once performed a beautiful study to see if all the turning points that one could retrospectively select

to be short and long a la birinyi who shows almost 5 times the market drift by getting in and out of the bear and bull markets with 20% being a fuzzy base line , —– and he found it completely consistent with randomness. It was a model of what a good study should be. Perhaps he will share it with us again, or at least tell us the drift. 

Richard Owen writes: 

That would be great to see. It is definitely one of the most mumbocentrically diverse areas of asset analysis and a firm and incestuous friend of the buy and hold debate. The more important corollary to the depressing corollary would therefore be that successful investing is almost entirely about the quality of your liabilities? Would a Japanese salaryman wealth manager with the Professor's report in hand have been able to maintain a career? If not, would he have been right to get a copy of Taleb out to console himself?

Charles "the professor" Pennington writes: 

I have kind of forgotten how that went, but I will see if I can find it.

There was a study of something kind of along those lines from Big Al and/or Kim Zussman not so long ago that was very compelling, covering dozens of possible trading strategies, but only one or two could thread the needle and do better than random.

Russ Sears writes: 

Not as rigorous as the Professor's, but I did a back of envelope study of the Dow from 1900 to 12/31/2012. Not including dividends, just the index.

There were 20 beginning of the years where the Dow was less than it began 10 years (of course these have overlapping decades).

What do do if you retrospectively find yourself in a "Secular Bear Market"?

The next year change in the dow average +14.35% min - 23.5% max 59.6% stdev of 21.1%. Whereas the overall was 4.7% and stdev of 20.9%.

Likewise the next ten years change based on roughly 20% steps of prior 10 years (again overlap) This only covered 1910 to year end 2012 since I needed 10 year periods before and after. There were 2 years 2008 and 2009 that the decade prior was negative, both had positive next years. But we do not know what it will be in 2018 and 2019 so they were not included. Here the overlap does matter since the next 10 year periods are not independent. 

Count  group avg Range for Group    Next 10 years  Range for group 

19     -16.8%     -49.7%       1.4%     108.5%      -3.6%     271.7%
19      16.9%       2.0%      33.0%      82.0%     -39.0%     238.8%
19      60.2%      35.4%      98.1%      95.8%     -15.1%     240.1%
19     137.9%      98.5%     169.4%      75.2%     -39.8%     323.4%
18     240.9%     172.7%     323.4%      96.8%     -49.7%     317.6%

Richard Owen writes: 

Very kind and thoughtful work! Apologies to be very dumb: what periods do the five groupings of 19 counts represent? And group avg [col 2] (I would have thought trailing? But the premise is those periods were negative?) The "excluding divs" heuristic so common for stock analysis is, I guess, one reason why we need King Dimson so badly.

Russ Sears writes: 

The period in the five groupings in the next decade. Hence, may double, triple or more count some years. It takes some time for the "past decade" to move into another grouping.

A Warning that Engendered the Discussion from Victor Niederhoffer: 

Please don't write more as you have threatened about "secular bull markets" or "secular bear markets" that can only be described in retrospect and have no predictive significance, and are mumbo jumbo and depend on random selected starting and ending points and would only lead our fine readers to wallow in absurd, unhelpful charlatanism.


Comments

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search