No. of Recommendations: 18
Fabulous table, thanks Zee!
This is a particularly good time to do it, since anything with over
about 18 months of history has seen both good times and bad. "Fair
weather friends" that bite you in the bear have had an excellent chance
to blow up this year. You can't rely on them doing so, but it's
nice that they have had some rope to hang themselves with--I'm always
leery about the screens that are invented during a long bull until
I see what happens next.

Here is one possible re-sorting of your data.
The idea is a "proof so far" metric, which is the product of the
post-discovery outperformance and the number of years post discovery.
So, a "winner" on this list either outperformed by a moderate amount for a
really long time (which makes it pretty believable) or by a tremendous
amount for a more moderate period of time.

The real effect of this is to push newer things down quite a bit,
like Fundamentals. It backtests beautifully, but hasn't got much
history under its belt to counteract accusations of having been overfit.
Maybe yes, maybe no---we'll have a better idea in a few years.

Ranks 1-4             metric
EG 3.614
PEG 2.620
EGRSW 1.985
KEY100 1.853
KEYEPS 1.632
LOWPE 1.592
RS26WK 1.499
Value_EG 1.383
PLOW26WK 1.109
RSW 1.108
RS13WKT12 1.105
PEG-Minimalist 1.094
SLS_RS13 1.081
RS1WK 1.076
GAR4CFS 1.071
PLOWEG5_RS631 1.064
GAR_EG5 1.032
REIT 0.910
RS52WK 0.909
ValueRatio 0.798
Tvalue 0.755
YIELD4 0.706
BLITZ 0.686
LOWPSR 0.623
RS4WK 0.589
RSPS 0.513
PIH_MCP 0.430
HI_DIV 0.355
3PT_Value_SmallCap 0.281
YldDiv 0.274
CDPD 0.205
SPARK 0.110
Fundamentals 0.096

No particular conclusions offered, but EG has certainly held up well,
and is admirably simple too. It's a surprisingly deep screen too:
Top 4 backtests at 34.7%, but even top 20 backtests at 27.9%.
A lot of the perormance is from a preposterously good 1999, 554% for top 4,
so by switching from CAGR to median rolling year, we can reduce that effect.
Median year top 4 is 27.8%, and median year top 20 is 24.8%--pretty close.
Of course, if you look at the screen definition, as the screen depth
approaches 25, it becomes simply RS26WKT12. No bad thing.

One possible use of this list is to build a "mine avoidance" blend, for
those who want the same blend all the time but want the strongest
proof of a screen after the backtest period. For example, you could
use the top 3 stocks of the top dozen screens on this list.
The falloff from the backtest era to the at-least-some-are-real-time
era is quite sharp: The pure backtest era of CAGR 1989-1999 was CAGR 57
versus 19 for the S&P. But, it's 16.6% CAGR 2000 to date versus 0.2% for the S&P.
Still, I'd be happy with beating the S&P by 16% a year. This blend
won't manage that, since this process itself is a backtest, but it
might be a blend that has more "holding up" power than others,
because each of the components has so far resisted the substantial
opportunities it has had to prove itself useless. Post discovery
success isn't a proof of worth, but post discovery failure is
a proof of lack of worth (depending only on your definition of failure).
For example, one of my suggestions, Safety_Bounce has lots of
interesting properties in backtest blends, but has been a total failure.
It fails to show any worth in daily-start testing, and in 45 months
post discovery it has lagged the S&P by about 6.1%/year (-0.1% versus +6.0%).
Don't use it.

Print the post  


What was Your Dumbest Investment?
Share it with us -- and learn from others' stories of flubs.
When Life Gives You Lemons
We all have had hardships and made poor decisions. The important thing is how we respond and grow. Read the story of a Fool who started from nothing, and looks to gain everything.
Contact Us
Contact Customer Service and other Fool departments here.
Work for Fools?
Winner of the Washingtonian great places to work, and Glassdoor #1 Company to Work For 2015! Have access to all of TMF's online and email products for FREE, and be paid for your contributions to TMF! Click the link and start your Fool career.