I have been developing statistical models over the last two years to
identify stocks that are trending up and to indicate good times to buy
them. This modeling has resulted in a system that I call SIBS, which
stands for Statistically Indicated Buy Signals. Briefly, the SIBS
system does the following:
1) Detects changes from a down-trend to an up-trend in stock price
2) Statistically confirms the new trend
3) Produces signals when continuation of the trend is likely.
I have been testing the SIBS system publicly since May, 1996 on the 3DO
(THDO) thread entitled "3do: Who bought large position late nov./early
december?" (See posts nubers 114, 116, 161, 164, 431, 433, 438, 463, and
470 for a complete history). I am now tranferring the SIBS testing and
discussion to this thread. The following table summarizes cumulative
results of the test so far:
Purchase Sale
Date Stock Price Date Price Wait (d) Gain (Loss) % Annualized %
5/1/96 NSM 15.75 5/20/96 16.875 19 875 5.6 108
5/14/96 NOVL 14.25 5/23/96 15.25 8 1000 6.1 278
5/24/96 LTD 21.50 holding
5/24/96 NSM 16.25 8/29/96 18.125 97 1625 10.0 38
8/5/96 F 33.50 holding
8/8/96 DGN 10.625 8/21/96 11.625 13 750 7.1 234
8/9/96 ERTS 30.50 8/15/96 32.75 6 2000 6.6 401
8/19/96 BORL 8.625 holding
8/23/96 ERTS 32.00 9/6/96 33.00 14 750 2.3 61
8/29/96 GYMB 29.625 9/4/96 30.813 6 938 3.2 193
I will continue updating this table as I observe new SIBS and make
hypothetical transactions. All hypothetical trades involve 1000 shares
and are made at the closing price on the day posted. I generally sell
after a rise of $1.00/share or more. Net gain is calculated as
gain=(sale price - purchase price) X 1000 - 250
The $250 is subtracted to approximately adjust for the bid/ask spread
and brokers fees. The only exception to this rule in the table above
is NOVL, purchased on 4/14/96. For that transaction, I knew the bid
and ask price at the time of purchase and at the time of sale and calculated
the gain exactly.
I look forward to any comments, questions, or discussion that anyone
may have concerning SIBS. |
