The purpose of the GTI is to measure the performance
of the universe of Gilder stocks -- i.e., the companies
included in the list of "Telecosm Technologies" each
month in the Gilder Technology Report, but excluding
any whose stock is not readily available to individual
It is intended to serve as a benchmark for Gilder
investors to evaluate the performance of their
portfolios. By comparing the GTI to published market
indexes, we can also follow how the Gilder stocks as a
group are faring compared to other market segments.
This is not to say that the GTI is anyone's recommended portfolio, neither mine nor the Gilder group's. Decisions as to which stocks to buy, and when to buy them, are beyond the scope of the GTI.
What I have sought to create in the GTI is a generic
equal-dollar index, one whose day-to-day performance
can be reproduced by anyone. (Note the exception
for "penny stocks" adopted effective May 6, 2002.)
I started it at 10,000 on January 1, 1999. From that
point on, it has changed each time it was updated by
the amount of the average change for the GTI
(Until December 1, 2000, the GTI was updated daily.
It is now updated weekly, after the last session of
To illustrate, suppose that there were 28 companies
on the list. Suppose that Company A was +1.82%,
Company B was -0.38%, and so on down the list. I add
up the 28 individual changes and divide by 28. That
gives me the average change. I then increase (or
decrease) the prior GTI by that amount, and the result
is the new GTI.
Some people think that it's not valid arithmetically to
add up percentage changes and then divide by the
number of companies. They're worried that it's
somehow wrong to add a plus percentage to a minus
percentage, or to add percentages at all. Under other
circumstances, they might have a point. But here
we are working with an equal dollar portfolio, so the
procedure is completely accurate and appropriate.
To illustrate, suppose that in the preceding paragraph
we had instead talked about having $100 invested in
each company. Suppose we then had said that
Company A was +$1.82, Company B was -$0.38, and
so on. Obviously, when we sum the 28 companies'
dollar returns, we get the portfolio's dollar return.
When we then divide by $2,800 (the portfolio's value at
the start of the day), we get the portfolio's percentage
return. Clearly this is a valid procedure.
This topic and others are examined both in FAQ's and
Is It Accurate?
When a new company is added to (or removed from)
the Gilder list, I wait until the end of the week to reflect
the change in the GTI, unless it happens so early in the
week that I can still go back to the end of the preceding
week, which I do when the Gilder Report comes out on a Monday or before the market closes on a Tuesday.
The GTI may understate typical Gilder performance, for
at least two reasons. The first is that every company in
it is weighted equally. Given the freedom to buy with
different weights, we ought to be able to beat a
mechanistic equal-weight portfolio. (I wish I could
weight the GTI stocks by GG's relative enthusiasm for
each, but that would be too subjective to be practical.)
Second, as individuals we are free to time our sales
and purchases according to a stock's changing price
from time to time. That should give us an edge against
the GTI, which must buy as soon as a company is
added to the list, and may not sell until it is removed.
Note that the GTI is not itself a possible portfolio. It
effectively reallocates its holdings after each update,
in order to once again have an equal dollar amount
invested in each company. For a real-world investor,
the transaction costs would be prohibitive.
(And yet . . .)
Rather than a portfolio, the GTI is an index, just as the
NASDAQ and S&P are. Their relative weights are
based on market capitalization, but that is not a useful
measure for Gilder investors. Why would we buy more
of a stock just because the issuer had a greater market
value? Thus there seems to be general consensus
among Gilderites that our best benchmark is an
equal-dollar index. The GTI's virtue is that it doesn't
become unequal over time.
While the GTI began on January 1, 1999, the GTR goes
back to August of 1996. That first year was sporadic,
with only three issues produced, and there was no
explicit list of recommended companies. By early
1997, though, it had evolved into the report we know
and enjoy today.
To complete the record, I have estimated what 1997
and 1998 GTI performance would have been. It is
included in the Year-by-year Performance table on
the home page.