From: John Conover <john@email.johncon.com>
Subject: Re: forwarded message from Dennis Krull
Date: 8 Oct 2000 21:55:01 -0000
BTW, I had a reason to send the attached out over the mailing list.
As it turns out, the elves' pro forma is on par with others that
attempt to time the market.
On average, although they still make money, they do slightly worse
than the stock market indices.
But it is possible to always beat the market indices in the long run
with a very simple portfolio strategy. Just invest in more than ten
stocks, (which ones doesn't seem to matter,) but each and every day,
(or at the most, each and every week,) make sure you have equal asset
allocation in each stock.
The reason it works is that the risk of investing in each of the ten,
or so, stocks adds root-mean-square in the portfolio, while the
average gain in the value of each of the stocks adds linearly.
So, you end up with the portfolio value growth being greater than the
growth in value of any stock in the portfolio-just by shuffling money
around in the portfolio.
As counter intuitive as it sounds, it makes sense. If you look at what
daily/weekly rebalancing does, it is taking money out of the stocks
that are gaining value at any time, and puts it into stocks that are
loosing at any time-a dynamic buy low, sell high timing strategy.
See Appendix 2 and 3 of http://www.johncon.com/ntropix/tsinvest.html
for particulars.
The -i argument to tsinvest uses this method to compute the index,
(the -j uses the sum total of the values of all stocks to calculate
the index,) because it is a very simple and effective strategy to
compare against the stock selection/timing strategies, (-d1 ... -d6.)
John
BTW, if you want to "tweek" tsinvest to do a little better, instead of
having it invest equally in each stock, have the fraction of the
portfolio invested in each stock proportional to the avg / rms of the
stock. Asset allocation is usually done with mathematical programming
techniques-but it can be solved analytically with the correct models.
John Conover writes:
> Attached came from the misc.invest.technical newsgroup ...
>
> John
>
> Content-Type: text/plain; charset=US-ASCII
>
> ------- start of forwarded message (RFC 934 encapsulation) -------
> Message-ID: <Sd2E5.97$ja.172@firefly>
> From: Dennis Krull <denkrull@bluestem.prairienet.org>
> Subject: Wall St Wk Elves Performance Report
> Date: Sun, 08 Oct 2000 17:33:38 GMT
>
>
>
>
> WALL ST. WK. ELVES PERFORMANCE REPORT
>
>
> WALL ST. WK. ELVES 10/6/00 : DJIA = 10596
>
> CURRENT % PERFORMANCE vs. DJIA # SIGNALS
> _____________________________________________________________________
>
> 5/26/95: DJIA = 4369 142.5 0.0 0
>
> Marty Zweig 23.1 -49.2 4
> Bernadette Murphy + 59.5 -34.2 5
> Ralph Acampora + 21.4 -49.9 13
> Laszlo Byrini + 67.7 -30.9 3
> _____________________________________________________________________
>
> 5/17/96: DJIA = 5688 86.3 0.0 0
>
> Frank Cappiello + 46.2 -21.5 4
> Harvey Eisen + 20.9 -35.1 4
> Mike Holland + 51.1 -18.9 11
> Bob Stovall 27.6 -31.5 8
> Mary Farrell + 25.2 -32.8 4
> _____________________________________________________________________
>
> The first group started on 5/26/95, and the second on 5/17/96.
> Each elf is measured against the DJIA starting from the same date.
> Current position is indicated as long(+), short(-), or neutral.
>
> The performance column is the cumulative perfomance of the DJIA
> during the periods which the elf was on a 'buy' signal,
> or the negative of such performance for a 'sell' signal. 'Neutral'
> positions of an elf have no effect on his/her performance.
>
> The next column indicates how much an elf's performance has
> underperformed a simple buy-and-hold of the DJIA over the same
> period. The last column is the number of changes each elf has made
> in his/her stated position.
>
> For example, everyone started with $100. Marty Zweig has returned
> 23 %, giving him $ 123 . $100 invested in the DJIA on the same
> date is now worth $ 243 . Thus Zweig has 123 / 243
> as much money as buy > hold, or 49 % less.
> ------- end -------
--
John Conover, john@email.johncon.com, http://www.johncon.com/