Skip to main content
Message Font: Serif | Sans-Serif
 
No. of Recommendations: 11
Here's a screen that seems to do well in both
the Zacks backtester and the MI/VL one.

Very simple and beats the S&P every year
in both cases.

I'm sure that someone out there can tweak it
a bit (but not too much) to be even better...

1) Mkt cap >1B
2) dividend yield * 1 year return / (P/E) top 5


 
CAGR in %
Year mi/vl screen Zacks screen S&P
2000 66 58 -11
2001 8 13 -9
2002 10 19 -22
2003 29 49 28
2004 35 55 10
2005 41 45 8
2006 26 43 14
2007 35 56 6


MI/VL screen Zacks screen S&P
Start $ 10,000 $ 10,000 $ 10,000
2000 $ 16,600 $ 15,800 $ 8,900
2001 $ 17,928 $ 17,905 $ 8,099
2002 $ 19,721 $ 21,269 $ 6,317
2003 $ 25,440 $ 31,727 $ 8,086
2004 $ 34,344 $ 49,024 $ 8,895
2005 $ 48,425 $ 71,325 $ 9,606
2006 $ 61,015 $ 102,239 $ 10,951
2007 $ 82,370 $ 159,382 $ 11,608


Also, for 2008, the Zacks screen is currently up 22.5%

In Zacks, I used a 4 week hold, and MI/VL a monthly hold-
might account for some of the differences, the other could
be in the databases accessed (survivorship issues??)

Since no proprietary ratings were used, the current picks from
Zacks and the free MSN screener are pretty much in sync
(excluding ADRs).
Print the post Back To Top
No. of Recommendations: 4
Just one more item, if you leave out the
first critera and have no mkt. cap. restrictions
(something I am slightly against)
you have some pretty amazing results from the Zacks
screen, but not quite as much from the MI/VL side.

Year  CAGR                    $10,000 start port.
2000 44% 14,400
2001 168% 38,592
2002 75% 67,536
2003 154% 171,541
2004 147% 425,423
2005 60% 680,676
2006 23% 837,232
2007 147% 2,067,963
ytd 27% 2,621,864


Over that time period, that works out to a CAGR of 93%
with a max drawdown of -21% (the S&P had a drawdown of -45%)

Note: will be in North Carolina for 10 days playing golf,
so will not be able (or inclined, depending on my score)
to answer any questions.
Print the post Back To Top
No. of Recommendations: 0
Ah, if only your handle was Desmond, rather than Tpoto. Then you could be referred to as Desmond 22. Nothing like a Nobel Peace Prize to add to one's accolades.

vez

http://nobelprize.org/nobel_prizes/peace/laureates/1984/tutu...
Print the post Back To Top
No. of Recommendations: 3
I'm sure that someone out there can tweak it
a bit (but not too much) to be even better...

1) Mkt cap >1B
2) dividend yield * 1 year return / (P/E) top 5
=== === ===
Looks great !!
What is the effect of adding a filter for daily trading volume, say over 500,000 shares/day ?
--BigBunk
Print the post Back To Top
No. of Recommendations: 0
I don't get quite as good using Screen Builder:
0007SBT15XmcpG1000Xtr1yMcdyDcpeT5
CAGD = 25, GSD = 22, Sharpe = 1.04
2000 29 -11
2001 14 -9
2002 14 -22
2003 25 28
2004 47 10
2005 32 8
2006 19 14
2007 21 6

Without the market cap limit, CAGD increases to 29 (Sharpe = 1.16)

How about this 22 year mod:
SBT15XmcpG100XOr13Ar26AcdyCDcpeT5

1) Mkt cap >100M
2) (dividend yield + 13 week return + 26 week return) / (P/E) top 5

CAGD = 37, GSD = 33, Sharpe = 1.11. Note that restricting the timeliness ratings doesn't help.

1986 13 24
1987 -14 6
1988 46 10
1989 56 35
1990 2 -5
1991 137 31
1992 10 7
1993 30 10
1994 54 2
1995 207 39
1996 10 23
1997 49 28
1998 -10 35
1999 30 18
2000 113 -11
2001 20 -9
2002 -11 -22
2003 64 28
2004 62 10
2005 28 8
2006 28 14
2007 51 6
Print the post Back To Top
No. of Recommendations: 0
In the VL universe, if I take the original screen and remove the market cap limitation, and include the trivial checks for cpe>0 cdy>0 - seems to me I should get exactly the definition of YLDEARNYEAR2 .

However, when I run it in Jamie's backtester
SBT15XcpeG0XcdyG0Xtr1yMcdyDcpeT5
I get CAGR/GSD/Sharpe 28/21/1.15 which is lower than YLDEARNYEAR2 30/21/1.26 .

Can anyone explain the discrepancy?

senco
Print the post Back To Top
No. of Recommendations: 2
sencoman wrote:

In the VL universe, if I take the original screen and remove the market cap limitation, and include the trivial checks for cpe>0 cdy>0 - seems to me I should get exactly the definition of YLDEARNYEAR2 .

However, when I run it in Jamie's backtester
SBT15XcpeG0XcdyG0Xtr1yMcdyDcpeT5
I get CAGR/GSD/Sharpe 28/21/1.15 which is lower than YLDEARNYEAR2 30/21/1.26 .

Can anyone explain the discrepancy?


Remove the timeliness restriction -- "T15X" -- criteria from the SB code, and you'll match the YLDEARNYEAR2 results.

It's funny. I haven't been at the process of screen creation very long, but I've already had several occasions where my "bright" (and independent) ideas end up distilling down either closely or exactly to an already established screen. This stuff ain't easy. :)

At any rate, keep the flow of ideas coming, tpoto.

-Joe
Print the post Back To Top
No. of Recommendations: 0
> Remove the timeliness restriction -- "T15X" -- criteria from the SB code, and you'll match the YLDEARNYEAR2 results.

Joe - yes, I see the difference in the results. Why is it so? As the backtester asks for a range I set it to 1-5 in order to include all stocks, thinking that all stock in VL universe has a grade. Is this not the case?

> At any rate, keep the flow of ideas coming, tpoto.

tpoto, by all means, please do!

Your ideas do inspire exploration, and this specific one definitely motivates looking at a VL stalwart in the Zacks universe.

senco
Print the post Back To Top
No. of Recommendations: 0
> Remove the timeliness restriction -- "T15X" -- criteria from the SB code, and you'll match the YLDEARNYEAR2 results.

Joe - yes, I see the difference in the results. Why is it so? As the backtester asks for a range I set it to 1-5 in order to include all stocks, thinking that all stock in VL universe has a grade. Is this not the case?


No. There are almost always a few stocks that don't have a VL timeliness rating. This is because there are recent or recently-announced events that cast reasonable doubt on whether history is relevant to predictions of the future.

An example of such an event would be a merger transaction where a relatively small, well-run VL-listed company is buying a badly-run company with much higher gross revenues.
Print the post Back To Top
No. of Recommendations: 0
Greetings from a Hampton Inn in North Carolina!
Too HOT to play golf. And it has to be very hot
for me not to play...


Yes, in retrospect that two liner is like yldearnyear2.
I added the mkt cap and did not use timeliness.
Did not realize that it was so close to that
yldearnyear2 screen.

Be carefull when getting picks out of MSN. I normally
add to their screen that the exchange be NY or Nasdaq
(can't remember the exact criteria, not being a my
home machine).

Not sure (need help)- is dyld (div. yield) the same as
cdy ( %current yield) in jamie's backtester?
Seems to give differing answers. Some MI/VL screens
use one, some the other.
Print the post Back To Top
No. of Recommendations: 0
I d/l Zack's 2 week trial just so I could try to recreate this screen. Having not yet tried it out (too busy being lazy!) tpoto, do you think I will be able to recreate your screen?

MZ4 http://flickr.com/photos/sarahracha/160661820/in/set-7215759... (napa valley)
Print the post Back To Top
No. of Recommendations: 1
hxn2 - top 10 from Zacks, no liquidity filter:

CMO
FRO
GNK
FDG
DSX
OCNF
NLY
DRYS
CEL
EXM
SPAL

This is using i26*i122/i72 where
i26 = current yield
i122 = price change 52 weeks
i72 = P/E using current fiscal year estimates

tpoto, are you using the same for P/E?
Print the post Back To Top