Question
on Software for Seasonals
Dear Mr. Smith, I have long admired your work
at FNN. I have previously been a subscriber to CTCR...thanks
for taking it over. I just received your special reprint
of the Nov 1998 issue and indeed it was fascinating! A
good job...I agree with your assessment of Jake Bernstein
(How did Larry ever get mixed up with him?) Right now
I'm interested in combining multiple seasonal factors
for the nearby S&P contract (not spread) or the cash index.
But maybe Jake's the best out there in seasonals...
I've ordered and studied The Stock Trader's Almanac. Traditionally,
they're the experts at stock market seasonals but I find
the data incomplete. The same to a slightly lesser degree
with the Supertrader's Almanac. The newsletter Market
Logic is fairly good but analyzed data only goes to 1994-5.
Have you ever or do you know of any company that makes
(or could make) software that produces the expected daily
% change of the S&P -- taking into account the year of
the presidential cycle, the annual bias, holiday effect,
monthly, five (six) day and 401(k) bulge, day-of-month
and day-of-the week historical record. All those "seasonals"
could be combined into a final grand score (without double
counting) for each and every trading day of any particular
year. For instance you could start with the average yearly
gain of 4.5%, which is four and one-half% or .045 in decimal
form. Divide that by the number of trading days (250),
and you get .00018, which is due to the secular increase;
the monthly % change in March, for example is .01 and
dividing that by 21 (days) results in .00048, etc. I remember
in the early days there was a fellow who predicted the
market on a daily basis for a whole year (or more) ahead.
I believe anyone could do it with reasonbly accurate results
with the available seasonal data and a computer program!
The result, of course, could be profitably used by anyone
according to his or her style as either a standalone or
supplement to other techniques. It's remarkable that by
now such an obvious statistically viable and stable bias
has not been long since discounted by the market but such
seems not the case... (see page four, THE 1999 STOCK TRADER'S
ALMANAC). Since 1950 the only major shift would seem to
be the day of week strengh reversal (from Friday to Monday)
and the 401(k) mid-month bulge as well as an extra end
of month bullish day.
This niche procedure would seem to be a natural and I
have either overlooked it (serious vendors - even after
searching the web) or it has been neglected by vendors
altogether. If the latter is the case, do you think the
data could be customized by a programmer? The bottom line
is that the info can be gathered and processed but it
is very tedious to do it by hand. Would you please comment
and help me?
Thanks, Bill
.
