No. of Recommendations: 72
I've taken a look at the internal relationships of various CAGRs within the array of 
backtested screens. If this information is already known to the MI veterans who've 
fashioned these screens, then my apologies to them.

Within the environment of the MIScreenBrowser I've been working on the development 
of a composite MIScreen index that factors in all the backtested screens, with the 
possible exception of Beta. My thinking is that ideally, the index will reflect both 
the overall screen CAGR  for a particular rebalance period and number of picks, as 
well as variations in screen performance across picks.

To explore the feasibility of this approach, I've compiled a set of tables mainly 
using Jamie Gritton's backtester (1986 thru 1999), and supplemented by a few more 
screens from bdfinney's backtester (1986 thru 1998). All values in this data set 
are for monthly rebalance. The tables are:

Table 1 • CAGRs of individual positions
Table 2 • CAGRs of rebalancing groups of positions
Table 3 • Table of various comparisons


Table 1 • CAGRs of Individual Positions (n1…n2):

		 1	 2	 3	 4	 5	 6	 7	 8	 9	10
											
Beta		27 	-5 	24 	13 	22 	39 	25 	 6 	26 	19 
EG		30 	39 	39 	37 	21 	23 	26 	25 	38 	18 
EGRSW		52 	33 	50 	27 	49 	24 	37 	24 	20 	41 
Form90		44 	22 	 6 	42 	29 	18 	32 	-8 	34 	32 
HiPrice		48 	70 	41 	10 	27 	13 	 5 	20 	13 	10 
Key		55 	25 	24 	20 	20 	29 	24 	30 	24 	19 
Key100		58 	34 	25 	 8 	18 	38 	18 	25 	 1 	28 
Blitz		41 	22 	21 	28 	25 	10 	11 	24 	33 	23 
KeyEPS		22 	45 	31 	58 	19 	26 	23 	10 	20 	19 
CAPRS		90 	56 	44 	2 	17 	44 	18 	10 	15 	23 
RSCAP		63 	48 	22 	37 	 6 	28 	31 	 2 	34 	12 
PEG13		50 	84 	48 	31 	38 	13 	26 	7 	-2 	18 
PEG26		18 	36 	86 	39 	-1 	22 	 2 	38 	20 	10 
PEGRSW		41 	55 	57 	61 	11 	36 	38 	 9 	35 	33 
Plow		19 	38 	39 	17 	13 	20 	11 	23 	 9 	15 
PlowRSW		50 	52 	22 	20 	20 	 8 	30 	 6 	12 	 8 
PlowLD		17 	36 	28 	24 	21 	25 	14 	14 	0 	-3 
RS4		51 	42 	41 	12 	34 	23 	25 	29 	18 	11 
RS13		49 	56 	23 	50 	65 	21 	-12 	25 	 9 	21 
RS26		51 	69 	10 	50 	26 	31 	16 	40 	11 	14 
RS52		50 	38 	37 	43 	20 	 7 	 1 	39 	21 	16 
RSFog		40 	47 	37 	36 	33 	41 	24 	6 	-1 	17 
RSibd		43 	27 	-1 	25 	48 	40 	25 	16 	20 	 3 
RSW		44 	49 	29 	 9 	58 	31 	30 	24 	28 	10 
Spark		22 	11 	26 	25 	37 	15 	 9 	30 	 1 	32 
SparkRS		23 	30 	16 	16 	32 	38 	15 	26 	-4 	20 


Table 1 shows variations in CAGR from position to position within each screen and 
among all screens. It was compiled by running ten single position backtest queries 
for each screen with all settings identical except for the position. As expected, the 
general trend is from higher returns in the lower ranks to lower returns in the 
higher ranks. However, I was surprised by some of the more extreme variation between 
positions. For example, position 3 in Form90 and RSibd, or position 4 in CAPRS.


Table 2 • CAGRs of rebalanced groups of positions:

		          Top Ranks [1-n]				Bottom Ranks [n-10]		
		1	1-2	1-3	1-4	1-5	1-10		6-10	8-10	10
											
Beta		27 	16 	23 	23 	25 	29 		32 	23 	19 
EG		30 	41 	45 	47 	44 	41 		35 	34 	18 
EGRSW		52 	50 	56 	52 	54 	48 		39 	36 	41 
Form90		44 	40 	31 	36 	37 	35 		31 	28 	32 
HiPrice		48 	65 	61 	49 	46 	32 		17 	18 	10 
Key		55 	42 	38 	34 	32 	30 		28 	26 	19 
Key100		58 	50 	44 	37 	34 	28 		22 	14 	28 
Blitz		41 	33 	31 	31 	31 	27 		23 	29 	23 
KeyEPS		22 	36 	36 	42 	38 	30 		22 	18 	19 
CAPRS		90 	56 	55 	43 	40 	34 		27 	20 	23 
RSCAP		63 	59 	49 	48 	41 	36 		29 	22 	12 
PEGRS13		50 	75 	71 	64 	60 	39 		19 	12 	18 
PEGRS26		18 	33 	54 	54 	44 	36 		25 	29 	10 
PEGRSW		41 	57 	63 	67 	57 	49 		39 	32 	33 
Plow		19 	32 	37 	33 	31 	26 		20 	19 	15 
PlowRSW		50 	56 	49 	41 	38 	28 		17 	12 	 8 
PlowLD		17 	31 	33 	34 	33 	26 		17 	 9 	-3 
RS4		51 	52 	42 	44 	44 	36 		28 	24 	11 
RS13		49 	60 	52 	55 	59 	40 		20 	23 	21 
RS26		51 	69 	52 	55 	51 	42 		31 	29 	14 
RS52		50 	53 	52 	54 	50 	38 		24 	32 	16 
RSFog		40 	50 	51 	51 	51 	39 		23 	11 	17 
RSibd		43 	43 	31 	32 	38 	34 		28 	19 	 3 
RSW		44 	55 	52 	43 	49 	43 		35 	30 	10 
Spark		22 	20 	23 	25 	29 	27 		23 	26 	32 
SparkRS		23 	28 	26 	24 	27 	27 		25 	20 	20 


In Table 2 I find the comparison between positions 1-5 and positions 6-10 to be of interest. 
In several instances the differences are not terribly pronounced, (in the case of Beta, 
they're actually inverted). In many cases peak values show up well into the ranks.


In Table 3 below, the columns represent these values:

A: rebalanced returns from positions 6 thru 10 as a percent of those from positions 1 thru 5 
   -- from a low of 32% for PEG13 to a high of 88% for Keystone.
B: averaged individual returns from positions 6 thru 10 as a percent of those from positions 
   1 thru 5 -- from 25% for Peg13  to 88% for Keystone.
C: arithmetic average of individual returns from positions 1 thru 10 as a percent of rebalanced 
   returns from positions 1 thru 10 -- from 68% for PlowLD to 91% for KeyEPS 
D: arithmetic average of individual returns from top 5 positions as a percent of rebalanced returns 
   from top 5 positions -- from 75% for RS52 to an anomalous 106% for CAPRS. 
E: arithmetic average of returns from bottom 5 positions as a percent of rebalanced returns 
   from bottom 5 positions -- from 59% for PlowLD to 100% for CAPRS. 
F: bottom 5 differential as a percentage of top 5 differential -- from 76% for CAPRS to an 
   anomalous 119% for Key100.


Table 3 • Relationships among groups of positions:

		  A		  B		  C		  D		  E		 F
		[6-10]		[6…10]		[1…10]		[1...5]		[6…10]		 E
		-------		-------		------		-----		------		---
		[1-5]		[1...5]		[1-10]		[1-5]		[6-10]		 D
												
Beta		1.28 		1.42 		 .68 		 .65 		 .72 		1.11 
EG		 .80 		 .78 		 .72 		 .75 		 .74 		 .98 
EGRSW		 .72 		 .69 		 .74 		 .78 		 .75 		 .96 
Form90		 .83 		 .76 		 .73 		 .78 		 .71 		 .92 
HiPrice		 .37 		 .31 		 .80 		 .85 		 .72 		 .84 
Key		 .88 		 .88 		 .90 		 .90 		 .90 		1.00 
Key100		 .65 		 .77 		 .90 		 .84 		1.00 		1.19 
Blitz		 .74 		 .74 		 .88 		 .88 		 .88 		 .99 
KeyEPS		 .58 		 .56 		 .91 		 .92 		 .89 		 .97 
CAPRS		 .69 		 .53 		 .93 		1.06 		 .80 		 .76 
RSCAP		 .72 		 .61 		 .79 		 .86 		 .74 		 .85 
PEGRS13		 .32 		 .25 		 .80 		 .84 		 .65 		 .78 
PEGRS26		 .57 		 .52 		 .75 		 .81 		 .74 		 .91 
PEGRSW		 .68 		 .67 		 .77 		 .79 		 .77 		 .98 
Plow		 .65 		 .62 		 .78 		 .81 		 .78 		 .96 
PlowRSW		 .45 		 .39 		 .81 		 .86 		 .75 		 .87 
PlowLD		 .52 		 .40 		 .68 		 .76 		 .59 		 .77 
RS4		 .64 		 .59 		 .79 		 .82 		 .76 		 .93 
RS13		 .34 		 .26 		 .77 		 .82 		 .64 		 .78 
RS26		 .61 		 .54 		 .76 		 .81 		 .72 		 .89 
RS52		 .48 		 .45 		 .72 		 .75 		 .70 		 .93 
RSFog		 .45 		 .45 		 .72 		 .76 		 .76 		1.00 
RSibd		 .74 		 .73 		 .73 		 .76 		 .75 		 .99 
RSW		 .71 		 .65 		 .73 		 .77 		 .70 		 .91 
Spark		 .79 		 .72 		 .77 		 .83 		 .76 		 .91 
SparkRS		 .93 		 .81 		 .79 		 .87 		 .76 		 .88 
												
Average		 .63 		 .59 		 .79 		 .83 		 .76 		 .92 


I want to get this posted, and have run out of time to spend on it at the moment. Hopefully 
the data will prompt further discussion and exploration.

-eremon
Print the post  

Announcements

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.