John G links below about mechanical shorting screen he devised.
I ran a variant through q investor. Looks promising. I did not include price book as that eliminated too many stocks before price cash. Only one year of underperformance 1991 c/w post below about not investing during sharp market rebounds from bottom. It is not obvious how to set this up in Q to approximate John G's screen. Appreciate other suggestions especially from the screens originator. April start, 12 week lag c/w recent post suggestions.
Row Modified Filter Rules
1 Price > 5
2 AvgVolume6W > 7000
3 MCap.FY < 200,000,000
4 MCap.FY > 20,000,000
5 %Return26W lowest% 25
6 PBook.FY > 3
7 PSales.FY > 2
8 FreeCashFlow.FY < 0
9 PCash.FY highest 10
Row Buy Date Percent Return (mean)vs S&P 500
1 1997 Apr 8.1% -40.0% 10 4
2 1996 Apr -39.5% -55.0% 9 9
3 1995 Apr 26.7% -2.9% 10 5
4 1994 Apr -18.0% -30.4% 10 7
5 1993 Apr -23.8% -24.8% 10 8
6 1992 Apr -21.4% -31.3% 10 7
7 1991 Apr 17.3% +10.4% 9 2
8 1990 Apr -6.1% -16.5% 6 4
9 1989 Apr -6.5% -20.9% 5 4
10 1988 Apr -25.7% -39.6% 6 6
11 1987 Apr -30.0% -16.1% 7 6
12 Compound -12.9% -25.5%
13 Mean -10.8% -24.3% 8 6
14 StdDev 20.8% 18.2% 2.0 2.1
To: Vol (2107 )
From: John G Monday, Feb 22 1999 9:49PM ET
Reply # of 2121
Vol, shorting the low RS neg cash flow screen does nicely in most
market conditions, but in a correction to a recovery it does
horribly.
You can see this in the two following table, which are actual
screen portfolio performance-to-date as of the close today
screen Russell 2000
6/10/98 (near market peak) -33% -12%
10/7/98 (near market bottom) +52% +19%
Quite simply, you do not want to be short broken stocks in a market rebound.
To: John G (2032 )
From: John G Monday, Jan 4 1999 11:32AM ET
Reply # of 2121
Here are performance results for the low RS/neg cashflow screen that
I've been posting monthly. Positive numbers mean the stocks went up,
i.e., more negative numbers are better for shortselling.
What is desired here is a more negative performance compared to
the Russell 2000.
screen elapsed time performance Russell 2000 difference
* 12 -42 % - 3 % -29 %
* 11 -38 % - 2 % -36 %
* 10 -39 % - 9 % -30 %
* 9 -31 % -13 % -18 %
* 8 -21 % -13 % - 8 %
6/10/98 7 mos. -36 % - 7 % -29 %
* 6 -23 % - 8 % -15 %
* 5 -20 % + 2 % -22 %
* 4 +27 % + 21 % + 6 %
10/7/98 3 mos. +12 % + 27 % -12 %
11/4/98 2 mos. +22 % + 8 % +14 %
12/8/98 1 mo. +11 % + 5 % + 6 %
* date shown with asterisk are based on Telescan's backtesting today,
which should be the same as the results I would have gotten if I
had recorded the data on the date shown.
Recall that small cap stocks had a peak April 21 a low Oct. 8 98
It looks as if this screen tends to produce a high beta portfolio, maybe
because the stocks tend to be lower priced stocks than those in the Russell 2000.
Whether the RUT is up or down, the screen does the same thing, but harder.
I think that the 1 mo. results are too early to be meaningful.
Bottom line:
Over the test period shown here, it looks like the screen
in general underperforms the RUT, as desired.
To: John G (1967 )
From: John G Tuesday, Dec 8 1998 6:38AM ET
Reply # of 2121
Here are the stocks for the Dec 98 low RS / neg cash flow screen.
I did this screen on Dec. 8.
The top 25 stocks were as follows, listed with the lowest 18-week RS first:
cvol,wins,thor,inis,wrdp,kosp,rmii,asvi,alyd,exca,diys,aim,dcti,heb,
mcar,qsri,svi,cnbi,ctt,gumm,uqm,tava,tscn,vinf,ifci,tera
click the link below to see yahoo's detailed report on all 25 stocks:
quote.yahoo.com.
=====================================Boiler plate info: ============
How the scan was done:I used Telescan's Prosearch, available on the
web (for a subscription) at tscn.com
Search components were:
shortability:
price > $5
30 day ave vol > 5 k
small cap:
$20 M < market cap < $200 M
low RS:
rel performance 18 weeks : low as possible
eliminate value stocks:
p/b > 3
p/s > 2
free cash flow < 0
cash/price ratio : low as possible
I optimized the above parameters in May and June 1998, which was soon
after a small cap market peak, by backtesting the results (I played
with the p/s, p/b ratios, the time span for the low RS, etc.) and this
is the set of parameters that did the best over the previous 3-6
months. The returns from this screen are probably best over a 3+
month holding period.
I don't recommend shorting any of the stocks produced by this screen.
This list is intended only to encourage further research. |
