Gummy's (Peter Ponzo's) web site Gummy's Tutorial on Mean Regression
Reversion to the mean DOES exist.
Not exact matches
Clearly, adding a small out - of - range segment
to a normally
mean - reverting chart can make it look (at least temporarily) as if the
mean reversion doesn't exist.
I give Grantham credit for coming
to this realization (something he has
done before) but I wonder how his investors feel about it after years of playing the
mean reversion waiting game.
Why
does the prospect of
Reversion compromise Meyer and Berko's ability
to solve their outstanding cases, and what
does that possibility
mean to both of them?
Rights
reversion can be tricky, especially when the contract doesn't give the author a unilateral (
meaning «one - sided») right
to terminate.
So you're seeing a
reversion to the
mean effect — a print success (and hence e-success) in the UK just doesn't affect the US odds very much and vice versa.
I believe # 5, incidentally, from your point of view, though I don't think I could call the erosion of competitive advantages of certain companies
to be
mean reversion, if and when it occurs.
I don't see the VIX being a strong predictor
to mean reversion returns.
They argued that value strategies produce superior returns because most investors don't fully appreciate the
mean reversion phenomenon, which leads them
to extrapolate past performance too far into the future.
Situations in which
mean reversion does not happen are rare enough as
to make a
mean reversion assumption a consistent friend
to the investor.
My previous post The Health of Stock
Mean Reversion: Dead, Dying or Doing Just Fine generated good reader's suggestions on other ways to check on mean reversio
Reversion: Dead, Dying or
Doing Just Fine generated good reader's suggestions on other ways
to check on
mean reversionreversion health.
LSV frame their Contrarian Investment, Extrapolation and Risk findings in the context of «contrarianism,» arguing that value strategies produce superior returns because most investors don't fully appreciate the phenomenon of
mean reversion, which leads them
to extrapolate past performance too far into the future.
But I consider value, momentum, and
mean -
reversion effects
to be givens, while I try
to analyze what industries and companies will
do well.
In all of my years of
doing quantitative analyses of equity and debt markets, as well as the economy as a whole, my models have shown me that there is a tendency toward
mean -
reversion, but it is a very weak tendency that is swamped by shocks
to the system in the short run.
On reflection, i suspect then that the above graph doesn't just capture
mean -
reversion in CAPE, but also
mean reversion in the other factors contributing
to total return — inflation, dividends, and growth rates.
Mauboussin's research seems
to suggest that, while there exists a strong tendency towards
mean reversion, some companies
do «post persistently high or low returns beyond what chance dictates.»
If
mean reversion does occur in the years ahead, the regression will begin
to show a strong relationship.
Valuation - based tactical asset allocation has proven very hard
to execute over time, for a simple reason: Asset - class valuations
do not exhibit much
mean reversion.
FWIW, I personally trade a more advanced
mean reversion system on and end - of - day basis
meaning I don't need
to sit in front of a screen.
We often predict by extrapolation and
do not consider
reversion to the
mean.
If you didn't use the concept of
mean reversion to your advantage in 2017, now is a great time
to start!
Lakonishok, Shleifer, and Vishny (LSV) argue that value strategies produce superior returns because most investors don't fully appreciate the phenomenon of
mean reversion, which leads them
to extrapolate past performance too far into the future.
Perhaps investors are better off taking into account
mean reversion on a sector by sector basis, given that we
do not seem
to be looking at a scenario of plummeting earnings that will sink all boats.
Mauboussin observes that
reversion to the
mean is a powerful force, and it impacts return on invested capital as it
does many other data series.
This would seem
to somewhat explain
mean reversion of stock prices of low p / b value firms (once Mr. Market realizes he can pay less for income - generating assets), but doesn't explain earnings growth.
I don't quibble with the theory, but, in practice, the strategy seems
to find stocks at the peak of the business cycle (see my summary of Mauboussin in ROIC and
reversion to the
mean: Part 1, 2 and 3).
In Contrarian Investment, Extrapolation, and Risk, Josef Lakonishok, Andrei Shleifer, and Robert Vishny argued that value strategies produce superior returns because most investors don't fully appreciate the phenomenon of
mean reversion, which leads them
to extrapolate past performance too far into the future.
If you normalise the fuel margins the implied multiple can start
to look less attractive, and
to the extent the market
does not expect
mean reversion in terms of fuel margins this could lead
to a nasty surprise.
Such active errors
do not negate however the phenomenon of
reversion to the
mean.
«When I think of long / short business,
to me there's 5 ways
to make money: 2 of those are you either play
mean reversion, which is what a lot of long / short strategies
do, or you can play momentum / trend, and that's typically what I
do.
I know it might be
mean reversion or a supply / demand phenomenon but
do not feel qualified
to say and would enjoy reading your perspective.
It doesn't therefore seem valid
to say that the superior returns
to value are due
to mean reversion when they haven't tested for value.
Of course, it
does imply
mean reversion & economic flexibility — so you always want
to watch out for the exception: A secular / permanent step - change in a country's circumstances, and / or an inability
to adjust
to changing circumstances.
Unlike in the case of the simple return or the return relative
to the market, we
do not find
mean reversion in performance once we control for manager peer - group performance.
The nude
did not
mean a
reversion to academic rigor but disciplined modernization, culminating with Pink Angels in 1945 and a sharp turn
to abstraction.
Central tendency, or
reversion (regression)
to the
mean, says exactly you can subtract problems by adding them together and averaging them away, if you
do it competently.
I haven't
done a quantitative analysis, but my guess is that given an old record, «A», and a new record, «B», the expectation that a year soon after B will be even less than B (a newer record) is probably less the larger the B minus A difference (eg,
reversion to the
mean), BUT, the expectation that a year soon after B will be less than A should increase the larger that difference (eg, there is more confidence in a larger decreasing trend due
to Bayesian updating).