It seems to me just another "I am better than everyone else" based purely on backtesting, not actual trading results
I also have many systems that give good returns in backtests, but this means nothing when trading them realtime
On 4/16/05, steve_almond <
steve2@xxxxxxxxxxxxxxxxxxxx
> wrote:
>
>
> An article appeared recently in the UK magazine "Investors
> Chronicle" entitled "Useless Tools".
> The gist of the article was that UK fund managers were doing a
> pathetically bad job using their tools (valuations, company
> analysis, business models, knowledge of management etc.) and were
> all (except 3) beaten over each of the last 5 years by the
> author's "Low-Risk Portfolio". This portfolio returned
86.5% (total)
> over the past 5 years.
>
> This low-risk portfolio was quoted as "...based on simple mean-
> variance optimisation, in which the only inputs are the volatilities
> and Betas of the share's quarterly returns since 1992..."
>
> I sent the author an email asking for further details and his reply
> is below. The question is, does anyone use this sort of analysis and
> does Amibroker lend itself to the required calculations?
>
> Steve
>
> >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>
>
> Hello. Thanks for your interest and kind words. I attach a brief
> description of what I do, taken from the website.
> To take this further, I'd recommend Modern Portfolio Theory and
> Investment Analysis, by Edwin Elton and Martin Gruber, who describe
> the process in detail.
> It's a very interesting idea to apply this to US stocks.
> One thing you'll almost certainly find is that historic (in-sample)
> efficient portfolios vastly out-perform fund managers. My portfolio
> shows that this is true out-of-sample too - though of course I've no
> idea whether this will remain the case.
> Best wishes
> Chris
>
> How we form the portfolio
> At the start of every calendar quarter since 1998 I've asked a
> simple question: what set of stocks in the FTSE 100 has had the
> lowest possible volatility of quarterly returns since 1992? This set
> has been our low-risk portfolio.
> To see how stocks get into this portfolio, consider the equation
> that tells us the variance of a portfolio. It is:
>
> Variance = 1/N x (Average variance) + (N-1)/N x Average covariance
>
> The low-risk portfolio is the one that minimizes this variance; to
> find this, I use some software called the Investment Portfolio,
> available from John Wiley & Sons.
> It's clear from this equation that there are only two ways a
> stock
> can get into our portfolio; either it has a low variance, or it has
> a low covariance with other stocks*.
> Note that low volatility on its own is not enough to get into our
> portfolio. Smiths Group has a lower variance than BAT. But it
> doesn't get into our portfolio whereas BAT is a major holding
> because it has a high covariance with other stocks whereas BAT
> doesn't.
> Conversely, low covariances with other stocks are not enough. Rio
> Tinto has these, but it's quite volatile, so it's excluded.
> Note also that equity risk can't be removed entirely. Even the
> lowest possible risk portfolio still has a standard deviation of
> almost half that of the market. This means that you can only remove
> half of equity risk by diversifying among equities. To reduce risk
> by more than this, you need other assets.
> * Actually, I use implied covariances, obtained by multiplying
> stocks' betas by the variance of the All-share index. There's
> a
> simple reason for doing this. Some pairs of stocks might have a
> negative covariance in the past simply because of luck – as one
> had
> bad news at the same time as the other had good.
> It would be imprudent to assume that such chance events will
> continue into the future, so we shouldn't use such covariances to
> measure risk.
> Instead, we ask: what fundamental reason do we have to expect that
> two stocks will move together? If you believe the capital asset
> pricing model – which says that market risk is the only
> systematic
> factor that determines stock returns – the answer can only be
> that
> the two stocks are sensitive to moves in the market. If this is so,
> we can estimate an implied covariance by multiplying their betas by
> the variance in the All-share index.
> The paradox here is that although the CAPM is clearly wrong as a
> guide to asset pricing, it produces a portfolio that delivers great
> risk-adjusted returns.
>
> ------------------------ Yahoo! Groups Sponsor --------------------~-->
> In low income neighborhoods, 84% do not own computers.
> At Network for Good, help bridge the Digital Divide!
>
http://us.click.yahoo.com/EpW3eD/3MnJAA/cosFAA/GHeqlB/TM> --------------------------------------------------------------------~->
>
> Please note that this group is for discussion between users only.
>
> To get support from AmiBroker please send an e-mail directly to
> SUPPORT {at}
amibroker.com
>
> For other support material please check also:
>
http://www.amibroker.com/support.html>
>
> Yahoo! Groups Links
>
> <*> To visit your group on the web, go to:
>
http://groups.yahoo.com/group/amibroker/>
> <*> To unsubscribe from this group, send an email to:
>
amibroker-unsubscribe@xxxxxxxxxxxxxxx>
> <*> Your use of Yahoo! Groups is subject to:
>
http://docs.yahoo.com/info/terms/>
>
--
Cheers
Graham
http://e-wire.net.au/~eb_kavan/