Automated strategies for investment management
US006317726B1
(12) United States Patent
(10) Patent N0.: US 6,317,726 B1 (45) Date of Patent: *Nov. 13, 2001
O’Shaughnessy
(54) AUTOMATED STRATEGIES FOR INVESTMENT MANAGEMENT
FOREIGN PATENT DOCUMENTS 2298299 *
(75) Inventor: James P. O’Shaughnessy, Coscob, CT (Us)
Notice:
(GB) ............................ .. G06F/19/00
OTHER PUBLICATIONS
(73) Assignee: Netfolio, Inc., Greenwich, CT (US) (*)
8/1996
Forbes (NPL) “Investment policy; Securities analysis; Fore
Subject to any disclaimer, the term of this patent iS eXtended or adjusted under 35 U~S~C~ 154(k)) by 0 days-
casting techniques; Ratio analysis; Books”. Forbes v158n6 pp:225 Sep. 9, 1996 CODEN: FORBA5 ISSN: 0015—6914. Author: Kennith Fisher.*
This patent is subject to a terminal disclaimer.
* cited by examiner
Primary Examiner—Vincent Millin
(21) Appl' NO‘: 09/361,654 (22) Filed: Jul_ 27’ 1999
Assistant Examiner—Jeffrey PWu (74) Attorney, Agent, or Firm—Elman & Associates
Related US. Application Data (63)
(60)
(57)
Continuation-in-part of application No. 08/995,296, ?led on Dec. 20, 1997. ligg‘gsional application NO- 60/034989: ?led on Dec- 30: ' 7
ABSTRACT
The invention is in the ?eld of using a Computer to Select corporate stocks for investment. Fifty stocks are selected from a database on the basis of certain criteria. The stocks are acquired in equal proportions, and the portfolio is
(51)
Int. Cl.
.................................................... .. G06F 17/60
rebalanced at the end of an annual tenn Strategy 1: market
(52)
US. Cl. ............................... .. 705/36; 705/35; 705/37;
Capitalization greater than $172 million (in?ation adjusted
705/38
?gure for $150 million in 1994 dollars); price-to-sales ratio
(58)
Fleld 0f Search ................................ .. 705/36, 37, 35,
less than 1'5; earnings higher than in previous year; market
705/38,
capitalization greater than market capitalization three
39
.
_
(56)
.
.
.
.
.
months ago; market capitalization greater than market capi
References Clted
taliZation siX months ago; buy stocks With highest one-year
U'S' PATENT DOCUMENTS
stock price appreciation. Strategy II: market capitalization greater than database mean; common shares outstanding
5,132,899 *
7/1992 FOX ...................................... .. 705/36
greater than database mean; CaSh?OW greater than the data
5,761,442 *
6/1998 Barr et al. ............................ .. 705/36
b
5,819,238 * 10/1998 Fernholz .............................. .. 705/36
t.
ase mean' (Crea mg
SET A _
.
t
1
t.
1
)> Pnce' 0'” ‘is m 10 655
th
an
5,946,666 * 8/1999 New, et a1_ 5,978,778 * 11/1999 O’ShaughneSSy
705/36 ---- -- 705/36
average for SET A; sales greater than 1.5 times the average for the database; no utility companies; buy the 25 or 50
670357286
3/2000
' ' ' ' " 705/36
stocks With the best one year stock price appreciation.
6,061,663 6,064,985 *
5/2000 Bloom .................................. .. 705/36 5/2000 Anderson ............................. .. 705/36
6,085,175
7/2000 Gugel ................................... .. 705/36
Fned
''''''''''''''
T
‘
26 Claims, 14 Drawing Sheets
Selem Market SW Cup/tullzdlion‘
5'0“ mom”
,
above $172M and
V
T ,
Me A
wme IO me A
t
H
172
HS
Select Stocks "om Hie A WM)
‘Woe/soles m6 less He D
‘
WC?! 1 5 and wme l _> to me B
I
Select Stocks rmm
we 0 which Irvcvecsed Market Cclpnulizutimn \I'\ my a mums and we
time D
L
we a
175
i-4
19
select Stocks item We a with earnings
“QUE/"'10" wet/160s yeQIS 6m wme to
‘we c Select W File D whlch vrvcvccsed Market
—>
Cupnuhztmon In last 6 months and wme '
V
me E
3
i H]
no
‘Son mm mm Hie‘l ‘E on mgmsv one’
Year Price qppveciuvion and Write in to Flle F m2
1
we F
via
U.S. Patent
Nov. 13, 2001
Sheet 1 0f 14
US 6,317,726 B1
Select Stocks with
Market Capitalization -—} above $172M and -—-}
Stock Database
File A
write to file A
1-1
l-2
1-3
Select Stocks from File A with
price/sales ratio less —} File D
File B
than 1.5 and write .
to file B
i-5
A
1-9
1-4
Select Stocks from
Select Stocks from
File C which
File B with earnings
increased Market
higher than previous _>
Capitalization in last 3 months and write to file D
years and write to file C
File C
1-7
1-6 1-8
Select Stocks from File D which
increased Market _> Capitalization in last
File E
6 months and write tofileE
1-] l
¢ l-lO Fig. i
Sort Stocks from File E on highest one
year price appreciation and write list to File F 1-12
——>
File F
1-1 3
U.S. Patent
Nov. 13, 2001
Sheet 2 0f 14
US 6,317,726 B1
Select Stocks with Market Capitalization
Stock Database
——> greater than
—-}
.
File A
database mean
2]
and write to file A
2_3
+
2-2
-
Select Stocks from File A with more
common shares outstanding than
File D
_>
Fne 5
database mean
and write to file B 2-9
Select Stocks from File C with price to
sales ratio less than average for File C and write to file D
-
2-5 2-4
Select Stocks from File B with cashflow
Fne C
database mean and write to file C
.
26
+
2-8
Select STOCKS from File D with SOleS
QleQTel Th0" 1-5
Select Stocks from File E, eliminate
_> utilities, and write to
times average for the database and
File E
_>
File F
file F 2 12
write to file E
¢
2-7
'
2-13
2-10 2-14 Sort Stocks from File F on highest one
year price appreciation and write list to File G
Fig. 2
2-l5
__>
File 6
U.S. Patent
Nov. 13, 2001
Sheet 3 0f 14
US 6,317,726 B1
Select Stocks with
Market Capitalization greater than
STOCK Database
File A
database mean and write to file A
3-3
+
3-2
Select Stocks from File A with shares
outstanding greater
File B
than the database
File D
mean and write to file B
Select Stocks from File C with sales greater than 1.5 times database mean and write to file D
3-5
3-4
3-9
Select Stocks from File B with cashflow greater than the database mean
——>
File C
and write to tile C 3-7
3-6 3-8 Select Stocks from File D which are not
utility companies
File E
and write to file E 3-H
3-10
Fig. 3
Sort Stocks from File E on lowest price to sales ratio and write list to File F 3-12
File F
U.S. Patent
Nov. 13, 2001
Sheet 4 0f 14
US 6,317,726 B1
Select Stocks with
Market Capitalization greater than
STOCK Database 4-1
File A
database mean and write to file A
+
4-3
4-2
Select Stocks from File A with earnings increase from
File B
previous year and
File D
write to file B 4-5
4-4
4-9
Select Stocks from File C with market
Select Stocks from File B with
capitalization
price/sales raio less than 1.5 and write to tile C
greater than market capitalization three months ago and write to file D
File C
4-7
4-6 M Select Stocks from File D with market
capitalization
File E
greater than market
capitalization six months ago and write to file E
4-l l
+ 440 Sort Stocks from File E on highest one year price
appreciation and
—>
File F
write list to File F 4-12
4-13
U.S. Patent
Nov. 13, 2001
Sheet 5 0f 14
US 6,317,726 B1
Select Stocks with
Stock Database
Market Capitalization greater than $25M and less then $250M
File A
and write to file A
+
5-3
5-2
Select Stocks from File A with price to sales ratio less than one and write to file B
File B
5-5 5-4
Sort Stocks from File 8 on price
appreciation and
File C
write list to File C
5-6
Fig. 5
5-7
U.S. Patent
Nov. 13, 2001
Sheet 6 6f 14
US 6,317,726 B1
U.S. Patent
Nov. 13, 2001
Name
Market
Inc. EVSI Evans WS Inc. PLI SALT Salton Inc. SHRP I FRNT FrontierAirIines Inc. HMGC HMG Worldwide ITGR I I CN Inc. ULTE Ultimate Inc. PR Price National Media CIX Inc. otal NT Inc.
Zones Inc. Inc.
Inc. Shift
-Gottstein Foods Co. Inc.
GI TEC KID HB
Locker Grou TEC Inc. KidoodIe Inc. Inc. Jewelers T
FM
MN
Inc. Inc.
Limit Inc. Inc. Natural Inc.
Int'l
86 61
Limited Inc.
Inc.
Inc. Information
21 1 93 000 50 637 000 84 424 000
Com
55 000 95 719 000
Communications I
40 055 000
Inc.
33 998 Co
SC DETA
Price/Sales 51 417 065 27 438 197 235 128 439 147 0 36 016 26 881 36 788 70 337 49 590 168 719 29 644 500 137 393 59 1 29 064 67 046 46 58 98 928 25 330 38 465 54 864 59 787 26 971 30 121 126 475 33 614 45 570 856 27 155 40 875 201 329 48 676 40 645
165 539
& K Healthcare
Industries Inc. PSC Inc. Inc.
US 6,317,726 B1
Sheet 7 0f 14
236 658 000
25 010 000 35 005 000
55,519,000
Fig. 7
9 °P Q 0
U.S. Patent
Nov. 13, 2001
Stock Database
Sheet 8 0f 14
Select Stocks in accordance with Strategy ii and write to file A
8-1 8-2
—->'
US 6,317,726 B1
File A
l 8-3
Select Stocks in accordance with Growth Model land write to file A
8-4
Fig. 8
Select Stocks in accordance with
Strategy Ill and write to file A
Stock Database
Select Stocks in accordance with Strategy ii and write to file A
File A
9-3 9-4
Select Stocks in accordance with
Growth Model land write to file A
Fig. 9 9-5
U.S. Patent
Nov. 13, 2001
Sheet 9 0f 14
US 6,317,726 B1
Select Stocks with
Market Capitalization above Si 72M and
Stock Database
File A
write to file A
103
10-2
10-1
Select Stocks from File A with good
trading liquidity and
File B
write to file 13
File D
10-5
10-4
Select Stocks from File C which increased Market Capitalization in last 3 months and write to file D
Select Stocks from File B with earnings
higher than previous years and write to file C
—>
File C
10-7
10-6 10-8 Select Stocks from File D which increased Market
File E
Capitalization in last 6 months and write TO file E
10-11
$ 10-10 Fig. 10
Sort Stocks from File E on highest one
year price appreciation and
File F
write list to File F 10-12
10-13
U.S. Patent
Nov. 13, 2001
Sheet 10 0f 14
US 6,317,726 B1
Select Stocks in accordance with
Value Model 3 and write to file A
STOCK DOTODGSG
Select Stocks in accordance with
Strategy ll and write to file A
Select Stocks in accordance with
Strategy Ill and write to file A Select Stocks in
accordance with Strategy VIII and 11-6
write to file A
Select Stocks in
Stock Database
accordance with Strategy II and write to file A
1 2-3
12-]
1 2-2 Select Stocks in accordance with
Strategy Ill and write to file A
12-4
Fig. 12
U.S. Patent
Nov. 13, 2001
Sheet 11 0f 14
US 6,317,726 B1
Select Stocks in accordance with
Stock Database
.
> value Model In and ——)
FrleA
write to file A i3-i
‘
13-3
13-2
Select Stocks in accordance with
Strategy Ill and write
) to file A
Fig. 13
1 3-4
Select Stocks in accordance with Value Model 3 and write to file A
1 4-2
Select Stocks in
Stock Database
‘
accordance with
—) snmegyu and wme —)
.
FileA
To file A i 4-]
1 4-3 1 4-4
Fig. 14
U.S. Patent
Nov. 13, 2001
Sheet 12 0f 14
US 6,317,726 B1
Select Stocks with
Stock Database
Market Capitalization
—-> greater than
—}
File A
database mean
l5-l
and write to file A
153
+ 152 Select Stocks from File A with more
common shares outstanding than
File D
.._+
File B
database mean and write to file B
15-5
159
Select Stocks from File C with price to
sales ratio less than average for File C and write to file D
+
l5-4
Select Stocks from File B with cashflow
4- greater than the
greater than 1.5 times average for the database and write to file E
1540
File c
157
l5-6
158
Select Stocks from File D with sales
__>
database mean and write to file C
Select Stocks from File E, eliminate utilities; select stocks
, with price-to-sales
——>
File F
ratio below average and write to file F
15-13 15-12
+ 1514
15-15
Sort Stocks from File _
F on highest one
me E
year price appreciation and 1 5_] 1
write list to File 6
Fig. l5
___>
File a
U.S. Patent
Nov. 13, 2001
Sheet 13 0f 14
US 6,317,726 B1
Select Stocks in S&P Stock Database
‘
,
500 and write to file A
16-]
+ 16-2 Select Stocks from
File A highest market CQDiTOIiZQTiO? ___>
Fi|e A
and write to file B
A 1 6-3
Select Stocks in accordance with
> Strategy II and write to file A 1 6-5
Select Stocks in accordance with
_—> Strategy Ill and write to file A 1 6-6
Fig. 16
16-4
U.S. Patent
Nov. 13, 2001
Sheet 14 0f 14
US 6,317,726 B1
T-Bills
/
1 7-2
FUNDS
17-1 \
Stocks selected In accordance wTih strategy XI 1 7-3
Fig. 17
US 6,317,726 B1 1
2
AUTOMATED STRATEGIES FOR INVESTMENT MANAGEMENT
eXplicitly and publicly so anyone With the time, money, data, equipment and inclination can reproduce the results. Third,
CROSS REFERENCE TO RELATED APPLICATION
someone using the same rules and the same reliable database must get the same results. Fourth, the results must be consistent over time; long-term results cannot oWe all their bene?ts to a feW years. Fifth, the rule must be intuitive and
This application is a continuation in part of application
logical and not be derived from the data.
Ser. No. 08/995,296, (?led: Dec. 20, 1997). Application Ser. No. 08/995,296 is a non-provisional application claiming priority under 35 U.S.C. § 119(e) from Provisional US. patent application Ser. No. 60/034,089, ?led Dec. 30, 1996. Both the provisional and non-provisional are incorporated herein by reference.
10
for most strategies), market capitaliZation (generally, requir
FIELD OF THE INVENTION 15
The invention is in the ?eld of using a computer to select corporate stocks for investment.
ing a minimum of $150 million), and annual rebalancing. SiZe of the portfolio. As evaluated, stock portfolios con tained 50 stocks, some of the portfolios in this application contain 10, 25, 30, 40 or 50 stocks. Researchers J. L. Evans
BACKGROUND OF THE INVENTION
Knowing hoW a particular investment strategy performed
The inventor used the S&P Compustat Active and Research Database from 1950 through 1994. The inventor used certain methods to evaluate hoW different rules for constructing portfolios Worked over these periods. Certain choices Were made regarding siZe of the portfolio (50 stocks
20
and S. H. Archer found most of the bene?ts of diversi?cation
risk, variability, and persistence of returns. Before the com
come from as feW as 16 stocks. One Wants to avoid holding too many or too feW stocks. Larger or smaller portfolios are
mencement of the inventor’s Work, there Was no Widely
Within the scope of the inventor’s invention.
historically gives one the vital information one needs on its
available comprehensive guide to Which strategies are long term Winners and Which are not. The inventor had access to 25
the historical S&P Compustat database of United States stock market information: forty-three years of results for Wall Street’s most popular investment strategies. It took the combination of fast computers and huge databases to prove that a portfolio’s returns are essentially
market capitaliZation in eXcess of $150 million (adjusted for
in?ation); it is called All Stocks throughout this application. 30
determined by the factors that de?ne the portfolio. Before computers, it Was almost impossible to determine What
strategy guided the development of a portfolio. The number
of underlying factors (eg price-to-earnings ratio, dividend yield) that an investor could consider seemed endless. The best one could do Was look at portfolios in the most general
35
factors over long periods of time, shoWing What one can 40
beat the S&P 500. Passive indeX fund managers have seen their assets rise as a result, from $10 billion in 1980 to over
$250 billion in 1990.
45
There is no product similar to or the same as the method
or apparatus of the present invention. Since the magnitude of the sums involved and the complexity of the relevant
larger, better-knoWn stocks With market capitaliZations greater than the database average (usually the top 16 percent of the database by market capitaliZation); it is called Large Stocks throughout the application. Annual Rebalancing. The portfolios studied are con structed and rebalanced annually. Stocks are equally Weighted With no adjustment for other variables. For eXample, if $1,000,000 is invested in 50 stocks, a $20,000 investment is made in each stock. Dividends are re-invested
50
SUMMARY OF THE INVENTION
year, all of the stocks may be sold and replaced With another ?fty stocks that meet the criteria of the strategy. Throughout the application, rebalancing refers to this process. Of course, for taX purposes, an investor must be careful in rebalancing
that one does not unnecessarily sell and reacquire shares of
The data presented by the inventor proves that the market
clearly and consistently reWards certain attributes (e.g.,
1995. This ?gure avoids focusing on tiny stocks and focuses only on those stocks Which a professional investor could by Without running into liquidity problems. A stock With a market capitaliZation of $27 million in 1950 is the equiva
in proportion With the original proportions. At the end of the
investment information, it is very desirable to use an objec
tive rule-based strategy and system for automating, to the eXtent practicable, the conduct of this decision-making.
The inventor chose $150 million after consulting a trader at a large Wall Street brokerage Who felt it Was the minimum necessary if he Was investing $100 million in 50 stocks in
lent of a $150 million stock at the end of 1994 and each is the equivalent of a stock With a market capitaliZation of $172 adjusted for in?ation. The second stock group includes
Ways. With computers, one can also test combinations of
eXpect in the future from any given investment strategy. History shoWs that traditional active management does not Work. The majority of actively managed funds do not
Market CapitaliZation. The inventor primarily studied tWo groups. The ?rst stock group includes only stocks With a
55
stocks With loW price-to-sales ratios) and clearly and con
sistently punishes others (eg stocks With high price-to-sales
stock in an eXisting portfolio When performing the rebal ancing. A year Was chosen since it is long enough to minimiZe effects of commissions and costs of rebalancing the portfolio. A term as long as tWo years or as short as three
ratios) over long periods of time. A paradoX remains: tests shoW high return predictability, but 80 percent of tradition ally managed mutual funds fail to beat the S&P 500. Models beat human forecasters because they reliably and consis tently apply the same criteria time after time.
60
Stock market decisions and portfolio constructions are served by a methodical scienti?c method. Certain rules help in this process. First, all models must use explicitly stated rules Without ambiguity or alloWance for a private or unique interpretation of the rule. Second, the rule must be stated
65
months could be used as the period after Which one rebal ances the portfolio in accordance With some embodiments of
the present invention. Sharpe Ratios. The inventor uses the Well-knoWn Sharpe ratio of reWard to risk, With higher numbers indicating better risk-adjusted returns. To arrive at the Sharpe ratio, take the average return from a strategy, subtract the risk-free rate of
interest, and then divide that number by the standard devia tion of return.
US 6,317,726 B1 4
3
microcap (capitalization less than $25 million) range. The stocks are too small for mutual fund to buy and far too numerous for an individual to tackle. Large Stocks per formed in a similar fashion to the S&P 500, With slightly
TABLE 1 Standard
Average
Deviation
S&P 500 T-Bills S&P 500 Minus T
14.25% 6.15% 8.10%
12.01% 2.07% 11.68%
Strategy
19.06%
24.37%
T-Bills
Strategy
6.15%
2.07%
12.91%
24.75%
better risk and almost equivalent yield. Computer. The present invention may be utilized on a general purpose computer, such an IBM PC, VAX, Mac or other computer knoWn to those in the art. Additionally, the 10
Minus T
sorting, ?ltering, and criteria could be encoded onto special purpose chips for creating special purpose hardWare for carrying out the present invention. The present invention could be implemented on a Wide area netWork, local area
See Table 1.
The risk-adjust return for the S&P 500 equals 8.10% divided by 12.01% or 67.44.
15
Value Factors include the folloWing: loW price-to
The risk-adjust return for the strategy equals 12.91%
earnings (PE) ratios; loW price-to-book ratios; loW price-to
divided by 24.37% or 52.97.
cash?oW ratios; loW price-to-sales ratios; dividend yields.
Market Capitalization Matters. A comparison of All Stocks (stocks With a market capitalization of more than
20
$150 million) and Large Stocks (stocks With a market capitalization higher than the database average) reveals that
the price by the current earnings per share. The higher the PE, the more investors are paying for earnings, and the 25
of hoW cheap or expensive it is relative to other stocks. $10,000 is invested on Dec. 31, 1951 in the 50 stocks With
the loWest price-to-earnings ratios. The portfolio is rebal
are reinvested, and all variables such as common shares 30
yield, see Table 3 and Table 3A.
Note that Large Stocks With LoW PE outperformed Large 35
stocks’ performance folloWs (in millions ($1M), from an initial investment of $10,000 invested over 43 years) in Table 2 and Table 2A. TABLE 2
anced each year to hold the 50 stocks With the loWest PE
ratios in any given year. For the yield, and risk-adjusted
P. O’Shaughnessy. Published by McGraW-Hill, 1997). A more detailed analysis of hoW capitalization affects
larger the implied expectations for future earnings groWth. A stock’s PE ratio is one of the most common measurements
those strategies using 5-year factors) in millions and the resultant Sharpe Ratio is presented. The portfolio is rebal anced annually. Stocks are equally Weighted, all dividends outstanding are time-lagged to avoid look-ahead bias. For those interested in vieWing more of the underlying data, the inventor suggests that the reader consult his commercially available book, What Works on Wall Street (Author, James
Price-to-Earnings Ratios. For many on Wall Street, buy ing stock With loW price-to-earnings (PE) ratios is a favored indicator. One ?nds a stock’s current PE ratios by dividing
size matters. All Stocks outperformed Large Stocks. For purposes of simplicity in this application, the yield of a SS 10,000 investment over the 43 years (or the 40 years for
netWork, through a dial-up connection to a dedicated machine, through an Internet or Intranet connection.
40
Stocks and had a better Sharpe ratio, While LoW PE All Stocks suffered a Worse return and With Worse risk. Small
companies can have a string of spectacular earnings gains on their Way to becoming large companies. It’s sensible for investors to aWard them With higher PE ratios. Since loW PE ratios indicate loWer investor expectations for earnings groWth, a small company With a loW PE ratio might have
Category
$1M’s
Sharpe Ratio
All Stocks
1.80
47
Large stocks
1.00
45
S&P 500 Cap < $1b 500M < cap < $1b
1.00 .80 .75
44 40
250M < cap < 500M
1.30
$100M < cap < $250M $25M < cap < $100M
1.30 1.70
39 45 42 41
29.10
57
Cap < $25M
very limited prospects. High PE Ratios are dangerous. Both All Stocks and Large Stocks outperform the High PE All Stocks and the High PE Large Stocks. 45
TABLE 3 Category 50
$1M’s
Sharpe Ratio
LoW PE All Stocks LoW PE Large stocks High PE All Stocks
1.23 2.29 .39
37 47 25
High PE Large stocks
.47
29
TABLE 2A Category All Stocks
$1M’s 2.7
Sharpe Ratio
Large stocks
1.6
48
S&P 500 Cap < $1b
1.7 1.6
48 48
500M < cap < $1b 250M < cap < 500M
1.9 3.4
$100M < cap < $250M $25M < cap < $100M
3.4 7.8
47 50 46 48
Cap < $25M
806
TABLE 3A
49
Category
60
64
Although, small cap stocks have been favored in many studies, All Stocks outperforms small caps. A great deal of the bene?t of small cap stocks comes from stocks in the
$1M’s
Sharpe Ratio
LoW PE All Stocks
2.12
40
LoW PE Large stocks High PE All Stocks
3.79 .56
50 27
High PE Large stocks
.65
31
Price-to-book ratios (P/B). Find price-to-book by dividing 65
the current price of the stock by the book value per share. Over the long term, the market reWards stocks With loW price-to-book ratios and punishes those With high ones. See Table 4.
US 6,317,726 B1 6 stock. The result is then multiplied by 100 to make it a TABLE 4 Category
percentage. Thus if a company pays an annual dividend of
$1M’s
Sharpe Ratio
3.59 3.42
47 54
High P/B All Stocks
.29
23
High P/B Large stocks
.56
30
LoW P/B All Stocks LoW P/B Large stocks
$1, and the current price of the stock is $10, the dividend is 10 percent. See Table 7 and Table 7A. TABLE 7 High Yielding Dividends (excluding utilities) 10
TABLE 4A Category
$1M’s
Sharpe Ratio
LoW P/B All Stocks
5.49
49
LoW P/B Large stocks High P/B All Stocks High P/B Large stocks
5.03 .38 .89
56 24 33
20
TABLE 5
LoW P/C All Stocks LoW P/C Large stocks High P/C All Stocks
2.95 3.62 .21
45 53 20
High P/C Large stocks
.55
30
$1M’s
Sharpe Ratio
LoW P/C All Stocks
4.48
47
LoW P/C Large stocks High P/C All Stocks High P/C Large stocks
5.77 .33 .79
56 23 31
annual sales (instead of earnings). Investors Who buy loW PSR stocks buy them because they believe they’re getting a bargain. See Table 6.
30
High PSR All Stocks High PSR Large stocks
$1M’s
Sharpe Ratio
All Stocks
1.6
49
Large stocks
2.90
54
different from their universe With virtually the same risk.
buy higher yielding stocks should stick to large, better knoWn companies, Which usually have the stronger balance sheets and longer operating histories that make higher dividends possible. Small stocks With high dividend yields may be in that position because their prices have fallen. Far from representing a bargain, their high dividend yields may Value Strategy Implications. The forty-three years of data shoW that the stock market methodically reWards certain types of stocks While punishing others. Stocks With loW
40
$1M’s
Sharpe Ratio
5.93 2.55 .07
52 49 11
.41
27
price-to-book, price-to-cash?oW, and price-to-sales ratios dramatically outperform the All Stocks universe. Just as
importantly, those With high price-to-book, price-to 45
cash?oW, and price-to-sales ratios do dramatically Worse. Stocks With loW price-to-earnings ratios and those With high dividend yields fail to beat All Stocks. Buying the 50 stocks With the loWest price-to-sales ratios Was the only strategy
50
The other value strategies came close, With the loW price to-book group matching All Stocks’ Sharpe ratio of 47, and the loW price-to-cash?oW group close behind With a Sharpe ratio of 45. All the Large Stocks value strategies beat the
that beat the All Stocks universe on a risk-adjusted basis.
TABLE 6
LoW PSR All Stocks LoW PSR Large stocks
TABLE 7A
The 50 highest-yielding stocks beat the universe 91 percent of the time over all rolling 10-year periods. Investors Who
Price-to-Sales (PSR). Price-to-Sales Ratios is a good measure. The price of the company is measured against
Category
51
35 be an indicator of more trouble to come.
TABLE 5A Category
2.01
The returns of the high yielding large stocks are entirely 25
Sharpe Ratio
39
Large stocks
Category
tion and amortization. The price-to-cash?oW ratio is the market value of the stock divided by total cash?oW. See Table 5.
$1M’s
Sharpe Ratio
High Yielding Dividends (excluding utilities)
by adding income (before extraordinary items) to deprecia
Category
$1M’s 1.11
15
Price-to-cash?oW (P/C). Price-to-cash?oW is yet another measure of Whether a stock is cheap or not. Find cash?oW
Category All Stocks
Large Stocks universe on an absolute and risk-adjusted basis, and they did so at least 88 percent of the time over all 55
rolling 10-year periods. GroWth investors Want high earnings and sales groWth With prospects of more of the same. They usually are not
TABLE 6A
concerned if stock has a high PE ratio, reasoning that a Category
$1M’s
Sharpe Ratio
LoW PSR All Stocks
8.25
LoW PSR Large stocks High PSR All Stocks High PSR Large stocks
3.85 .09 .64
53 52 12 30
60
GroWth investors often aWard high prices to stocks With
rapidly increasing earnings. One-Year-Earnings-Per-Share Percentage Changes. One 65
Dividend yields. Find a stock’s dividend yield by dividing the indicated annual dividend rate by the current price of the
company can groW its Way out of short-term overvaluations.
year-earnings-per-share percentage changes are a poor lone factor upon Which to base investment decisions. See Table 8 and Table 8A.
US 6,317,726 B1 7
8
TABLE 8
TABLE 10A
1-vear earnin_gs-per-share percentage changi Category
$1M’S
Sharpe Ratio
Best All Stocks
.91
34
Best Large stocks
.39
28
gorstin stosctksks Ors
arge
Category 5
Sharpe Ratio
1.08 1.09
37 43
Best pro?t magins All Stocks Best pro?t margins Large stocks
2:
0C
$1M’s
High return on equity (ROE) is a hallmark of a growth
'
10 stock. One ?nds return on equity by dividing common stock
equity into income before extraordinary items (a company’s income after all expenses but before provisions for dividends). One then multiplies them by 100 to express the
TABLE 8A .
term as a percentage. Ilere is used common liquidating
1-vear earnings-per-share percentage changi
_
_
15 equity (called CEQL 1n Compustat) as a proxy for common
Category Best A11 Stocks Best Large stocks
$1M’s 129
Sharpe Ratio 34
equity As With high pro?t margins, many believe that a high
.57
28
return on equity (ROE) is an excellent gauge of hoW
1'1;
2:
effectively a company invests shareholders’ money. The
I
20 higher the ROE, the better the company’s ability to invest one’s money, and presumably, the better an investment the
The implication is that buying stocks simply because they stock will be_ See Table 11 and Table 11A have great earnings gains is a losing proposition. Stocks With the highest 1-year earnings gains almost alWays have the TABLE 11 highest prince-to-earnings ratios, another indicator that poor 25 performance lies ahead. While their returns are slightly
category
$1M’S
Sharpe Ratio
higher than those With the best earning changes, there is no
compelling theory to justify buying stocks With the Worst
East 58g ‘2111 stoiksk
earnings changes.
es
rge S 00 S
Best 5-year earnings gains. Some analysts believe that a 30 1-year change in earnings is meaningless and that one should focus on 5-year groWth rates. Using 5-year earnings
TABLE 11A
gains as the only determinate Will lead to disappointing results. See Table 9 and Table 9A.
Category 35
TABLE 9
Category Best 5_year earnings gains All Stocks Best 5-year earnings gains Large stocks
$1M’s 35 .37
Sharpe Ratio 26 28
$1M’S
_ Sharpe Ram
Best ROE All Stocks
2.51
41
Best ROE Large stocks
1.14
39
ROE provides an excellent example of the importance of 40 looking at the long-term When judging a strategyjs effec tiveness. An investor just out of college at the end of 1964
studying hoW stocks With high return on equity perform ?nd encouraging evidence. The 50 highest ROE stocks from both the All and Large Stocks universe outperformed their
TABLE 9A Category
is '
$1M’S
Sharpe Ratio
.61
23
45 respective benchmarks in the previous decade. Over the
_ _ Best 5-year earnings gains All Stocks
longer period, however, ROE Was a poor sole performance indicator
Best 5-year earnings gains Large stocks
'
Relative Price Strength. Using strong price momentum as _
I
a determinate runs counter to ef?cient market theories. One
NetPro?t marglns are at} fixcenent gauge of a Company 5 50 cannot use past prices to predict future prices, according to
Operanng e?imffncy and ablhty to compéte Successfully Wlth
ef?cient market theory. Conversely, another school of
other ?rms in its ?eld. Thus ‘many bClICVC' that ?rms With
thought Says one Should buy Stocks that have been most
ingctil pro?t gaging“? bettecr) lnvgsténenti’ 5mg: they are tile
battered by the market. In this application, relative strength
6.21.6.3 1‘!
elf 1n us mes‘
me .n S n? pro
margms ,y
and price performers Will be used interchangeably. See Table
dividing income before extraordinary items (A company s . . - 55 12 and Table 12A for comparison of 1-year relative strength
income after all expenses but before provisions for
h
dividends) by net sales. This is then multiplied by 100 to get a percentage. See Table 10 and Table 10A. History shoWs
C anges'
using high pro?t margins as the only determinate in buying a stock Will lead to disappointing results.
TABLE 12
TABLE 10 Category Best pro?t margins All Stocks Best pro?t margins Large stocks
. 1-year relative strength
6O
Category
$1M’S
Sharpe Ratio
.74 .75
34 4O
65
Best All Stocks Best Large stocks Worst All Stocks Worst Large stocks
$1M’s
Sharpe Ratio
3.31 2.98 .03 .49
43 49 3 29
US 6,317,726 B1 9
10
TABLE 12A
PE ratios beloW 20 and positive earnings gains for the year and bought the 50 With the best 1-year price performance, one Would actually earn less than if one bought the loW PE,
1-year relative strength Category
$1M’s
Sharpe Ratio
Best All Stocks
4.11
43
Best Large stocks
4.43
51
Worst All Stocks
.04
Worst Large stocks
.61
5 30
high relative strength stocks alone. The addition of positive earnings gains hurt performance in this instance. More factors do not necessarily mean better performance.
Buying stocks With strong 1-year earnings gains and strong relative price strength outperforms the All Stocks universe. A tWo-factor model that requires stocks from All 10
While All Stocks Best 1-year relative strength had an impressive yield, it had a high standard deviation and therefore high risk that brought the Sharpe ratio to 43, under the All Stocks universe’s 47. Large Stocks Best 1-year relative strength had an impressive yield With slightly more risk than the Large Stocks universe: the resultant 49 Sharpe ratio is higher than the Large Stocks universe’s 45.
also outperforms the All Stocks universe. See Table 14 and Table 14A.
Category a
goal.
- ear 115 toe k s 1y
$1M’s Sharpe Ratio ains a b ove 250/0; 'g g'
earnin s;
Top 50 1-year price Appreciation (all Stocks) ROE>15; Top 50 1-yearPrice 20
performance is dramatically outperformed by the market. See Table 12 and Table 12A. Instead of focussing on the effects of either a single groWth or value factor, using several factors alloWs one to enhance performance or reduce risk, depending on one’s
TABLE 14
15
Price momentum conveys different information about the prospects of a stock and is a much better indicator than
factors such as earnings and groWth rates. Worst 1-year price
Stocks to have 1-year earnings gains exceeding 25 percent and then buys the 50 With the best 1-year price performance
. 63
50
9.4
55
appreciation All Stocks
1.7
47
(Large Stocks) ROE>15; Top 50 1-year Price
2.3
45
appreciation 25
TABLE 14A
From All Stocks With a PE ratio beloW 20, take the 50
stocks With the best 1 year price appreciation. Or from All Stocks With a price-to-book ratio beloW 1, take the 50 stocks With the best 1 year price appreciation. Either multi-factor
analysis outperforms All Stocks, and outperforms either of
30
Category
$1M’s Sharpe Ratio
(all Stocks) 1-year earnings; gains above 25%; Top 50 1-year price Appreciation (all Stocks) ROE>15; Top 50 1-yearPrice
14.38
56
appreciation
the factors that comprise it (i.e. loW PE ratio or best price appreciation). See Table 13 and Table 13A.
All Stocks
2.68
49
(Large Stocks) ROE>15; Top 50 1-year Price
3.92
49
appreciation TABLE 13 Category
35
This model Was not able to be tested in the Large Stocks
$1M’s Sharpe Ratio
50 Stocks With PE ratios beloW 20 and best 1
8.6
55
50 stocks With price-to-book ratio beloW 1 and
10.6
61
best 1 year price appreciation All Stocks PSR
(12) United States Patent
(10) Patent N0.: US 6,317,726 B1 (45) Date of Patent: *Nov. 13, 2001
O’Shaughnessy
(54) AUTOMATED STRATEGIES FOR INVESTMENT MANAGEMENT
FOREIGN PATENT DOCUMENTS 2298299 *
(75) Inventor: James P. O’Shaughnessy, Coscob, CT (Us)
Notice:
(GB) ............................ .. G06F/19/00
OTHER PUBLICATIONS
(73) Assignee: Netfolio, Inc., Greenwich, CT (US) (*)
8/1996
Forbes (NPL) “Investment policy; Securities analysis; Fore
Subject to any disclaimer, the term of this patent iS eXtended or adjusted under 35 U~S~C~ 154(k)) by 0 days-
casting techniques; Ratio analysis; Books”. Forbes v158n6 pp:225 Sep. 9, 1996 CODEN: FORBA5 ISSN: 0015—6914. Author: Kennith Fisher.*
This patent is subject to a terminal disclaimer.
* cited by examiner
Primary Examiner—Vincent Millin
(21) Appl' NO‘: 09/361,654 (22) Filed: Jul_ 27’ 1999
Assistant Examiner—Jeffrey PWu (74) Attorney, Agent, or Firm—Elman & Associates
Related US. Application Data (63)
(60)
(57)
Continuation-in-part of application No. 08/995,296, ?led on Dec. 20, 1997. ligg‘gsional application NO- 60/034989: ?led on Dec- 30: ' 7
ABSTRACT
The invention is in the ?eld of using a Computer to Select corporate stocks for investment. Fifty stocks are selected from a database on the basis of certain criteria. The stocks are acquired in equal proportions, and the portfolio is
(51)
Int. Cl.
.................................................... .. G06F 17/60
rebalanced at the end of an annual tenn Strategy 1: market
(52)
US. Cl. ............................... .. 705/36; 705/35; 705/37;
Capitalization greater than $172 million (in?ation adjusted
705/38
?gure for $150 million in 1994 dollars); price-to-sales ratio
(58)
Fleld 0f Search ................................ .. 705/36, 37, 35,
less than 1'5; earnings higher than in previous year; market
705/38,
capitalization greater than market capitalization three
39
.
_
(56)
.
.
.
.
.
months ago; market capitalization greater than market capi
References Clted
taliZation siX months ago; buy stocks With highest one-year
U'S' PATENT DOCUMENTS
stock price appreciation. Strategy II: market capitalization greater than database mean; common shares outstanding
5,132,899 *
7/1992 FOX ...................................... .. 705/36
greater than database mean; CaSh?OW greater than the data
5,761,442 *
6/1998 Barr et al. ............................ .. 705/36
b
5,819,238 * 10/1998 Fernholz .............................. .. 705/36
t.
ase mean' (Crea mg
SET A _
.
t
1
t.
1
)> Pnce' 0'” ‘is m 10 655
th
an
5,946,666 * 8/1999 New, et a1_ 5,978,778 * 11/1999 O’ShaughneSSy
705/36 ---- -- 705/36
average for SET A; sales greater than 1.5 times the average for the database; no utility companies; buy the 25 or 50
670357286
3/2000
' ' ' ' " 705/36
stocks With the best one year stock price appreciation.
6,061,663 6,064,985 *
5/2000 Bloom .................................. .. 705/36 5/2000 Anderson ............................. .. 705/36
6,085,175
7/2000 Gugel ................................... .. 705/36
Fned
''''''''''''''
T
‘
26 Claims, 14 Drawing Sheets
Selem Market SW Cup/tullzdlion‘
5'0“ mom”
,
above $172M and
V
T ,
Me A
wme IO me A
t
H
172
HS
Select Stocks "om Hie A WM)
‘Woe/soles m6 less He D
‘
WC?! 1 5 and wme l _> to me B
I
Select Stocks rmm
we 0 which Irvcvecsed Market Cclpnulizutimn \I'\ my a mums and we
time D
L
we a
175
i-4
19
select Stocks item We a with earnings
“QUE/"'10" wet/160s yeQIS 6m wme to
‘we c Select W File D whlch vrvcvccsed Market
—>
Cupnuhztmon In last 6 months and wme '
V
me E
3
i H]
no
‘Son mm mm Hie‘l ‘E on mgmsv one’
Year Price qppveciuvion and Write in to Flle F m2
1
we F
via
U.S. Patent
Nov. 13, 2001
Sheet 1 0f 14
US 6,317,726 B1
Select Stocks with
Market Capitalization -—} above $172M and -—-}
Stock Database
File A
write to file A
1-1
l-2
1-3
Select Stocks from File A with
price/sales ratio less —} File D
File B
than 1.5 and write .
to file B
i-5
A
1-9
1-4
Select Stocks from
Select Stocks from
File C which
File B with earnings
increased Market
higher than previous _>
Capitalization in last 3 months and write to file D
years and write to file C
File C
1-7
1-6 1-8
Select Stocks from File D which
increased Market _> Capitalization in last
File E
6 months and write tofileE
1-] l
¢ l-lO Fig. i
Sort Stocks from File E on highest one
year price appreciation and write list to File F 1-12
——>
File F
1-1 3
U.S. Patent
Nov. 13, 2001
Sheet 2 0f 14
US 6,317,726 B1
Select Stocks with Market Capitalization
Stock Database
——> greater than
—-}
.
File A
database mean
2]
and write to file A
2_3
+
2-2
-
Select Stocks from File A with more
common shares outstanding than
File D
_>
Fne 5
database mean
and write to file B 2-9
Select Stocks from File C with price to
sales ratio less than average for File C and write to file D
-
2-5 2-4
Select Stocks from File B with cashflow
Fne C
database mean and write to file C
.
26
+
2-8
Select STOCKS from File D with SOleS
QleQTel Th0" 1-5
Select Stocks from File E, eliminate
_> utilities, and write to
times average for the database and
File E
_>
File F
file F 2 12
write to file E
¢
2-7
'
2-13
2-10 2-14 Sort Stocks from File F on highest one
year price appreciation and write list to File G
Fig. 2
2-l5
__>
File 6
U.S. Patent
Nov. 13, 2001
Sheet 3 0f 14
US 6,317,726 B1
Select Stocks with
Market Capitalization greater than
STOCK Database
File A
database mean and write to file A
3-3
+
3-2
Select Stocks from File A with shares
outstanding greater
File B
than the database
File D
mean and write to file B
Select Stocks from File C with sales greater than 1.5 times database mean and write to file D
3-5
3-4
3-9
Select Stocks from File B with cashflow greater than the database mean
——>
File C
and write to tile C 3-7
3-6 3-8 Select Stocks from File D which are not
utility companies
File E
and write to file E 3-H
3-10
Fig. 3
Sort Stocks from File E on lowest price to sales ratio and write list to File F 3-12
File F
U.S. Patent
Nov. 13, 2001
Sheet 4 0f 14
US 6,317,726 B1
Select Stocks with
Market Capitalization greater than
STOCK Database 4-1
File A
database mean and write to file A
+
4-3
4-2
Select Stocks from File A with earnings increase from
File B
previous year and
File D
write to file B 4-5
4-4
4-9
Select Stocks from File C with market
Select Stocks from File B with
capitalization
price/sales raio less than 1.5 and write to tile C
greater than market capitalization three months ago and write to file D
File C
4-7
4-6 M Select Stocks from File D with market
capitalization
File E
greater than market
capitalization six months ago and write to file E
4-l l
+ 440 Sort Stocks from File E on highest one year price
appreciation and
—>
File F
write list to File F 4-12
4-13
U.S. Patent
Nov. 13, 2001
Sheet 5 0f 14
US 6,317,726 B1
Select Stocks with
Stock Database
Market Capitalization greater than $25M and less then $250M
File A
and write to file A
+
5-3
5-2
Select Stocks from File A with price to sales ratio less than one and write to file B
File B
5-5 5-4
Sort Stocks from File 8 on price
appreciation and
File C
write list to File C
5-6
Fig. 5
5-7
U.S. Patent
Nov. 13, 2001
Sheet 6 6f 14
US 6,317,726 B1
U.S. Patent
Nov. 13, 2001
Name
Market
Inc. EVSI Evans WS Inc. PLI SALT Salton Inc. SHRP I FRNT FrontierAirIines Inc. HMGC HMG Worldwide ITGR I I CN Inc. ULTE Ultimate Inc. PR Price National Media CIX Inc. otal NT Inc.
Zones Inc. Inc.
Inc. Shift
-Gottstein Foods Co. Inc.
GI TEC KID HB
Locker Grou TEC Inc. KidoodIe Inc. Inc. Jewelers T
FM
MN
Inc. Inc.
Limit Inc. Inc. Natural Inc.
Int'l
86 61
Limited Inc.
Inc.
Inc. Information
21 1 93 000 50 637 000 84 424 000
Com
55 000 95 719 000
Communications I
40 055 000
Inc.
33 998 Co
SC DETA
Price/Sales 51 417 065 27 438 197 235 128 439 147 0 36 016 26 881 36 788 70 337 49 590 168 719 29 644 500 137 393 59 1 29 064 67 046 46 58 98 928 25 330 38 465 54 864 59 787 26 971 30 121 126 475 33 614 45 570 856 27 155 40 875 201 329 48 676 40 645
165 539
& K Healthcare
Industries Inc. PSC Inc. Inc.
US 6,317,726 B1
Sheet 7 0f 14
236 658 000
25 010 000 35 005 000
55,519,000
Fig. 7
9 °P Q 0
U.S. Patent
Nov. 13, 2001
Stock Database
Sheet 8 0f 14
Select Stocks in accordance with Strategy ii and write to file A
8-1 8-2
—->'
US 6,317,726 B1
File A
l 8-3
Select Stocks in accordance with Growth Model land write to file A
8-4
Fig. 8
Select Stocks in accordance with
Strategy Ill and write to file A
Stock Database
Select Stocks in accordance with Strategy ii and write to file A
File A
9-3 9-4
Select Stocks in accordance with
Growth Model land write to file A
Fig. 9 9-5
U.S. Patent
Nov. 13, 2001
Sheet 9 0f 14
US 6,317,726 B1
Select Stocks with
Market Capitalization above Si 72M and
Stock Database
File A
write to file A
103
10-2
10-1
Select Stocks from File A with good
trading liquidity and
File B
write to file 13
File D
10-5
10-4
Select Stocks from File C which increased Market Capitalization in last 3 months and write to file D
Select Stocks from File B with earnings
higher than previous years and write to file C
—>
File C
10-7
10-6 10-8 Select Stocks from File D which increased Market
File E
Capitalization in last 6 months and write TO file E
10-11
$ 10-10 Fig. 10
Sort Stocks from File E on highest one
year price appreciation and
File F
write list to File F 10-12
10-13
U.S. Patent
Nov. 13, 2001
Sheet 10 0f 14
US 6,317,726 B1
Select Stocks in accordance with
Value Model 3 and write to file A
STOCK DOTODGSG
Select Stocks in accordance with
Strategy ll and write to file A
Select Stocks in accordance with
Strategy Ill and write to file A Select Stocks in
accordance with Strategy VIII and 11-6
write to file A
Select Stocks in
Stock Database
accordance with Strategy II and write to file A
1 2-3
12-]
1 2-2 Select Stocks in accordance with
Strategy Ill and write to file A
12-4
Fig. 12
U.S. Patent
Nov. 13, 2001
Sheet 11 0f 14
US 6,317,726 B1
Select Stocks in accordance with
Stock Database
.
> value Model In and ——)
FrleA
write to file A i3-i
‘
13-3
13-2
Select Stocks in accordance with
Strategy Ill and write
) to file A
Fig. 13
1 3-4
Select Stocks in accordance with Value Model 3 and write to file A
1 4-2
Select Stocks in
Stock Database
‘
accordance with
—) snmegyu and wme —)
.
FileA
To file A i 4-]
1 4-3 1 4-4
Fig. 14
U.S. Patent
Nov. 13, 2001
Sheet 12 0f 14
US 6,317,726 B1
Select Stocks with
Stock Database
Market Capitalization
—-> greater than
—}
File A
database mean
l5-l
and write to file A
153
+ 152 Select Stocks from File A with more
common shares outstanding than
File D
.._+
File B
database mean and write to file B
15-5
159
Select Stocks from File C with price to
sales ratio less than average for File C and write to file D
+
l5-4
Select Stocks from File B with cashflow
4- greater than the
greater than 1.5 times average for the database and write to file E
1540
File c
157
l5-6
158
Select Stocks from File D with sales
__>
database mean and write to file C
Select Stocks from File E, eliminate utilities; select stocks
, with price-to-sales
——>
File F
ratio below average and write to file F
15-13 15-12
+ 1514
15-15
Sort Stocks from File _
F on highest one
me E
year price appreciation and 1 5_] 1
write list to File 6
Fig. l5
___>
File a
U.S. Patent
Nov. 13, 2001
Sheet 13 0f 14
US 6,317,726 B1
Select Stocks in S&P Stock Database
‘
,
500 and write to file A
16-]
+ 16-2 Select Stocks from
File A highest market CQDiTOIiZQTiO? ___>
Fi|e A
and write to file B
A 1 6-3
Select Stocks in accordance with
> Strategy II and write to file A 1 6-5
Select Stocks in accordance with
_—> Strategy Ill and write to file A 1 6-6
Fig. 16
16-4
U.S. Patent
Nov. 13, 2001
Sheet 14 0f 14
US 6,317,726 B1
T-Bills
/
1 7-2
FUNDS
17-1 \
Stocks selected In accordance wTih strategy XI 1 7-3
Fig. 17
US 6,317,726 B1 1
2
AUTOMATED STRATEGIES FOR INVESTMENT MANAGEMENT
eXplicitly and publicly so anyone With the time, money, data, equipment and inclination can reproduce the results. Third,
CROSS REFERENCE TO RELATED APPLICATION
someone using the same rules and the same reliable database must get the same results. Fourth, the results must be consistent over time; long-term results cannot oWe all their bene?ts to a feW years. Fifth, the rule must be intuitive and
This application is a continuation in part of application
logical and not be derived from the data.
Ser. No. 08/995,296, (?led: Dec. 20, 1997). Application Ser. No. 08/995,296 is a non-provisional application claiming priority under 35 U.S.C. § 119(e) from Provisional US. patent application Ser. No. 60/034,089, ?led Dec. 30, 1996. Both the provisional and non-provisional are incorporated herein by reference.
10
for most strategies), market capitaliZation (generally, requir
FIELD OF THE INVENTION 15
The invention is in the ?eld of using a computer to select corporate stocks for investment.
ing a minimum of $150 million), and annual rebalancing. SiZe of the portfolio. As evaluated, stock portfolios con tained 50 stocks, some of the portfolios in this application contain 10, 25, 30, 40 or 50 stocks. Researchers J. L. Evans
BACKGROUND OF THE INVENTION
Knowing hoW a particular investment strategy performed
The inventor used the S&P Compustat Active and Research Database from 1950 through 1994. The inventor used certain methods to evaluate hoW different rules for constructing portfolios Worked over these periods. Certain choices Were made regarding siZe of the portfolio (50 stocks
20
and S. H. Archer found most of the bene?ts of diversi?cation
risk, variability, and persistence of returns. Before the com
come from as feW as 16 stocks. One Wants to avoid holding too many or too feW stocks. Larger or smaller portfolios are
mencement of the inventor’s Work, there Was no Widely
Within the scope of the inventor’s invention.
historically gives one the vital information one needs on its
available comprehensive guide to Which strategies are long term Winners and Which are not. The inventor had access to 25
the historical S&P Compustat database of United States stock market information: forty-three years of results for Wall Street’s most popular investment strategies. It took the combination of fast computers and huge databases to prove that a portfolio’s returns are essentially
market capitaliZation in eXcess of $150 million (adjusted for
in?ation); it is called All Stocks throughout this application. 30
determined by the factors that de?ne the portfolio. Before computers, it Was almost impossible to determine What
strategy guided the development of a portfolio. The number
of underlying factors (eg price-to-earnings ratio, dividend yield) that an investor could consider seemed endless. The best one could do Was look at portfolios in the most general
35
factors over long periods of time, shoWing What one can 40
beat the S&P 500. Passive indeX fund managers have seen their assets rise as a result, from $10 billion in 1980 to over
$250 billion in 1990.
45
There is no product similar to or the same as the method
or apparatus of the present invention. Since the magnitude of the sums involved and the complexity of the relevant
larger, better-knoWn stocks With market capitaliZations greater than the database average (usually the top 16 percent of the database by market capitaliZation); it is called Large Stocks throughout the application. Annual Rebalancing. The portfolios studied are con structed and rebalanced annually. Stocks are equally Weighted With no adjustment for other variables. For eXample, if $1,000,000 is invested in 50 stocks, a $20,000 investment is made in each stock. Dividends are re-invested
50
SUMMARY OF THE INVENTION
year, all of the stocks may be sold and replaced With another ?fty stocks that meet the criteria of the strategy. Throughout the application, rebalancing refers to this process. Of course, for taX purposes, an investor must be careful in rebalancing
that one does not unnecessarily sell and reacquire shares of
The data presented by the inventor proves that the market
clearly and consistently reWards certain attributes (e.g.,
1995. This ?gure avoids focusing on tiny stocks and focuses only on those stocks Which a professional investor could by Without running into liquidity problems. A stock With a market capitaliZation of $27 million in 1950 is the equiva
in proportion With the original proportions. At the end of the
investment information, it is very desirable to use an objec
tive rule-based strategy and system for automating, to the eXtent practicable, the conduct of this decision-making.
The inventor chose $150 million after consulting a trader at a large Wall Street brokerage Who felt it Was the minimum necessary if he Was investing $100 million in 50 stocks in
lent of a $150 million stock at the end of 1994 and each is the equivalent of a stock With a market capitaliZation of $172 adjusted for in?ation. The second stock group includes
Ways. With computers, one can also test combinations of
eXpect in the future from any given investment strategy. History shoWs that traditional active management does not Work. The majority of actively managed funds do not
Market CapitaliZation. The inventor primarily studied tWo groups. The ?rst stock group includes only stocks With a
55
stocks With loW price-to-sales ratios) and clearly and con
sistently punishes others (eg stocks With high price-to-sales
stock in an eXisting portfolio When performing the rebal ancing. A year Was chosen since it is long enough to minimiZe effects of commissions and costs of rebalancing the portfolio. A term as long as tWo years or as short as three
ratios) over long periods of time. A paradoX remains: tests shoW high return predictability, but 80 percent of tradition ally managed mutual funds fail to beat the S&P 500. Models beat human forecasters because they reliably and consis tently apply the same criteria time after time.
60
Stock market decisions and portfolio constructions are served by a methodical scienti?c method. Certain rules help in this process. First, all models must use explicitly stated rules Without ambiguity or alloWance for a private or unique interpretation of the rule. Second, the rule must be stated
65
months could be used as the period after Which one rebal ances the portfolio in accordance With some embodiments of
the present invention. Sharpe Ratios. The inventor uses the Well-knoWn Sharpe ratio of reWard to risk, With higher numbers indicating better risk-adjusted returns. To arrive at the Sharpe ratio, take the average return from a strategy, subtract the risk-free rate of
interest, and then divide that number by the standard devia tion of return.
US 6,317,726 B1 4
3
microcap (capitalization less than $25 million) range. The stocks are too small for mutual fund to buy and far too numerous for an individual to tackle. Large Stocks per formed in a similar fashion to the S&P 500, With slightly
TABLE 1 Standard
Average
Deviation
S&P 500 T-Bills S&P 500 Minus T
14.25% 6.15% 8.10%
12.01% 2.07% 11.68%
Strategy
19.06%
24.37%
T-Bills
Strategy
6.15%
2.07%
12.91%
24.75%
better risk and almost equivalent yield. Computer. The present invention may be utilized on a general purpose computer, such an IBM PC, VAX, Mac or other computer knoWn to those in the art. Additionally, the 10
Minus T
sorting, ?ltering, and criteria could be encoded onto special purpose chips for creating special purpose hardWare for carrying out the present invention. The present invention could be implemented on a Wide area netWork, local area
See Table 1.
The risk-adjust return for the S&P 500 equals 8.10% divided by 12.01% or 67.44.
15
Value Factors include the folloWing: loW price-to
The risk-adjust return for the strategy equals 12.91%
earnings (PE) ratios; loW price-to-book ratios; loW price-to
divided by 24.37% or 52.97.
cash?oW ratios; loW price-to-sales ratios; dividend yields.
Market Capitalization Matters. A comparison of All Stocks (stocks With a market capitalization of more than
20
$150 million) and Large Stocks (stocks With a market capitalization higher than the database average) reveals that
the price by the current earnings per share. The higher the PE, the more investors are paying for earnings, and the 25
of hoW cheap or expensive it is relative to other stocks. $10,000 is invested on Dec. 31, 1951 in the 50 stocks With
the loWest price-to-earnings ratios. The portfolio is rebal
are reinvested, and all variables such as common shares 30
yield, see Table 3 and Table 3A.
Note that Large Stocks With LoW PE outperformed Large 35
stocks’ performance folloWs (in millions ($1M), from an initial investment of $10,000 invested over 43 years) in Table 2 and Table 2A. TABLE 2
anced each year to hold the 50 stocks With the loWest PE
ratios in any given year. For the yield, and risk-adjusted
P. O’Shaughnessy. Published by McGraW-Hill, 1997). A more detailed analysis of hoW capitalization affects
larger the implied expectations for future earnings groWth. A stock’s PE ratio is one of the most common measurements
those strategies using 5-year factors) in millions and the resultant Sharpe Ratio is presented. The portfolio is rebal anced annually. Stocks are equally Weighted, all dividends outstanding are time-lagged to avoid look-ahead bias. For those interested in vieWing more of the underlying data, the inventor suggests that the reader consult his commercially available book, What Works on Wall Street (Author, James
Price-to-Earnings Ratios. For many on Wall Street, buy ing stock With loW price-to-earnings (PE) ratios is a favored indicator. One ?nds a stock’s current PE ratios by dividing
size matters. All Stocks outperformed Large Stocks. For purposes of simplicity in this application, the yield of a SS 10,000 investment over the 43 years (or the 40 years for
netWork, through a dial-up connection to a dedicated machine, through an Internet or Intranet connection.
40
Stocks and had a better Sharpe ratio, While LoW PE All Stocks suffered a Worse return and With Worse risk. Small
companies can have a string of spectacular earnings gains on their Way to becoming large companies. It’s sensible for investors to aWard them With higher PE ratios. Since loW PE ratios indicate loWer investor expectations for earnings groWth, a small company With a loW PE ratio might have
Category
$1M’s
Sharpe Ratio
All Stocks
1.80
47
Large stocks
1.00
45
S&P 500 Cap < $1b 500M < cap < $1b
1.00 .80 .75
44 40
250M < cap < 500M
1.30
$100M < cap < $250M $25M < cap < $100M
1.30 1.70
39 45 42 41
29.10
57
Cap < $25M
very limited prospects. High PE Ratios are dangerous. Both All Stocks and Large Stocks outperform the High PE All Stocks and the High PE Large Stocks. 45
TABLE 3 Category 50
$1M’s
Sharpe Ratio
LoW PE All Stocks LoW PE Large stocks High PE All Stocks
1.23 2.29 .39
37 47 25
High PE Large stocks
.47
29
TABLE 2A Category All Stocks
$1M’s 2.7
Sharpe Ratio
Large stocks
1.6
48
S&P 500 Cap < $1b
1.7 1.6
48 48
500M < cap < $1b 250M < cap < 500M
1.9 3.4
$100M < cap < $250M $25M < cap < $100M
3.4 7.8
47 50 46 48
Cap < $25M
806
TABLE 3A
49
Category
60
64
Although, small cap stocks have been favored in many studies, All Stocks outperforms small caps. A great deal of the bene?t of small cap stocks comes from stocks in the
$1M’s
Sharpe Ratio
LoW PE All Stocks
2.12
40
LoW PE Large stocks High PE All Stocks
3.79 .56
50 27
High PE Large stocks
.65
31
Price-to-book ratios (P/B). Find price-to-book by dividing 65
the current price of the stock by the book value per share. Over the long term, the market reWards stocks With loW price-to-book ratios and punishes those With high ones. See Table 4.
US 6,317,726 B1 6 stock. The result is then multiplied by 100 to make it a TABLE 4 Category
percentage. Thus if a company pays an annual dividend of
$1M’s
Sharpe Ratio
3.59 3.42
47 54
High P/B All Stocks
.29
23
High P/B Large stocks
.56
30
LoW P/B All Stocks LoW P/B Large stocks
$1, and the current price of the stock is $10, the dividend is 10 percent. See Table 7 and Table 7A. TABLE 7 High Yielding Dividends (excluding utilities) 10
TABLE 4A Category
$1M’s
Sharpe Ratio
LoW P/B All Stocks
5.49
49
LoW P/B Large stocks High P/B All Stocks High P/B Large stocks
5.03 .38 .89
56 24 33
20
TABLE 5
LoW P/C All Stocks LoW P/C Large stocks High P/C All Stocks
2.95 3.62 .21
45 53 20
High P/C Large stocks
.55
30
$1M’s
Sharpe Ratio
LoW P/C All Stocks
4.48
47
LoW P/C Large stocks High P/C All Stocks High P/C Large stocks
5.77 .33 .79
56 23 31
annual sales (instead of earnings). Investors Who buy loW PSR stocks buy them because they believe they’re getting a bargain. See Table 6.
30
High PSR All Stocks High PSR Large stocks
$1M’s
Sharpe Ratio
All Stocks
1.6
49
Large stocks
2.90
54
different from their universe With virtually the same risk.
buy higher yielding stocks should stick to large, better knoWn companies, Which usually have the stronger balance sheets and longer operating histories that make higher dividends possible. Small stocks With high dividend yields may be in that position because their prices have fallen. Far from representing a bargain, their high dividend yields may Value Strategy Implications. The forty-three years of data shoW that the stock market methodically reWards certain types of stocks While punishing others. Stocks With loW
40
$1M’s
Sharpe Ratio
5.93 2.55 .07
52 49 11
.41
27
price-to-book, price-to-cash?oW, and price-to-sales ratios dramatically outperform the All Stocks universe. Just as
importantly, those With high price-to-book, price-to 45
cash?oW, and price-to-sales ratios do dramatically Worse. Stocks With loW price-to-earnings ratios and those With high dividend yields fail to beat All Stocks. Buying the 50 stocks With the loWest price-to-sales ratios Was the only strategy
50
The other value strategies came close, With the loW price to-book group matching All Stocks’ Sharpe ratio of 47, and the loW price-to-cash?oW group close behind With a Sharpe ratio of 45. All the Large Stocks value strategies beat the
that beat the All Stocks universe on a risk-adjusted basis.
TABLE 6
LoW PSR All Stocks LoW PSR Large stocks
TABLE 7A
The 50 highest-yielding stocks beat the universe 91 percent of the time over all rolling 10-year periods. Investors Who
Price-to-Sales (PSR). Price-to-Sales Ratios is a good measure. The price of the company is measured against
Category
51
35 be an indicator of more trouble to come.
TABLE 5A Category
2.01
The returns of the high yielding large stocks are entirely 25
Sharpe Ratio
39
Large stocks
Category
tion and amortization. The price-to-cash?oW ratio is the market value of the stock divided by total cash?oW. See Table 5.
$1M’s
Sharpe Ratio
High Yielding Dividends (excluding utilities)
by adding income (before extraordinary items) to deprecia
Category
$1M’s 1.11
15
Price-to-cash?oW (P/C). Price-to-cash?oW is yet another measure of Whether a stock is cheap or not. Find cash?oW
Category All Stocks
Large Stocks universe on an absolute and risk-adjusted basis, and they did so at least 88 percent of the time over all 55
rolling 10-year periods. GroWth investors Want high earnings and sales groWth With prospects of more of the same. They usually are not
TABLE 6A
concerned if stock has a high PE ratio, reasoning that a Category
$1M’s
Sharpe Ratio
LoW PSR All Stocks
8.25
LoW PSR Large stocks High PSR All Stocks High PSR Large stocks
3.85 .09 .64
53 52 12 30
60
GroWth investors often aWard high prices to stocks With
rapidly increasing earnings. One-Year-Earnings-Per-Share Percentage Changes. One 65
Dividend yields. Find a stock’s dividend yield by dividing the indicated annual dividend rate by the current price of the
company can groW its Way out of short-term overvaluations.
year-earnings-per-share percentage changes are a poor lone factor upon Which to base investment decisions. See Table 8 and Table 8A.
US 6,317,726 B1 7
8
TABLE 8
TABLE 10A
1-vear earnin_gs-per-share percentage changi Category
$1M’S
Sharpe Ratio
Best All Stocks
.91
34
Best Large stocks
.39
28
gorstin stosctksks Ors
arge
Category 5
Sharpe Ratio
1.08 1.09
37 43
Best pro?t magins All Stocks Best pro?t margins Large stocks
2:
0C
$1M’s
High return on equity (ROE) is a hallmark of a growth
'
10 stock. One ?nds return on equity by dividing common stock
equity into income before extraordinary items (a company’s income after all expenses but before provisions for dividends). One then multiplies them by 100 to express the
TABLE 8A .
term as a percentage. Ilere is used common liquidating
1-vear earnings-per-share percentage changi
_
_
15 equity (called CEQL 1n Compustat) as a proxy for common
Category Best A11 Stocks Best Large stocks
$1M’s 129
Sharpe Ratio 34
equity As With high pro?t margins, many believe that a high
.57
28
return on equity (ROE) is an excellent gauge of hoW
1'1;
2:
effectively a company invests shareholders’ money. The
I
20 higher the ROE, the better the company’s ability to invest one’s money, and presumably, the better an investment the
The implication is that buying stocks simply because they stock will be_ See Table 11 and Table 11A have great earnings gains is a losing proposition. Stocks With the highest 1-year earnings gains almost alWays have the TABLE 11 highest prince-to-earnings ratios, another indicator that poor 25 performance lies ahead. While their returns are slightly
category
$1M’S
Sharpe Ratio
higher than those With the best earning changes, there is no
compelling theory to justify buying stocks With the Worst
East 58g ‘2111 stoiksk
earnings changes.
es
rge S 00 S
Best 5-year earnings gains. Some analysts believe that a 30 1-year change in earnings is meaningless and that one should focus on 5-year groWth rates. Using 5-year earnings
TABLE 11A
gains as the only determinate Will lead to disappointing results. See Table 9 and Table 9A.
Category 35
TABLE 9
Category Best 5_year earnings gains All Stocks Best 5-year earnings gains Large stocks
$1M’s 35 .37
Sharpe Ratio 26 28
$1M’S
_ Sharpe Ram
Best ROE All Stocks
2.51
41
Best ROE Large stocks
1.14
39
ROE provides an excellent example of the importance of 40 looking at the long-term When judging a strategyjs effec tiveness. An investor just out of college at the end of 1964
studying hoW stocks With high return on equity perform ?nd encouraging evidence. The 50 highest ROE stocks from both the All and Large Stocks universe outperformed their
TABLE 9A Category
is '
$1M’S
Sharpe Ratio
.61
23
45 respective benchmarks in the previous decade. Over the
_ _ Best 5-year earnings gains All Stocks
longer period, however, ROE Was a poor sole performance indicator
Best 5-year earnings gains Large stocks
'
Relative Price Strength. Using strong price momentum as _
I
a determinate runs counter to ef?cient market theories. One
NetPro?t marglns are at} fixcenent gauge of a Company 5 50 cannot use past prices to predict future prices, according to
Operanng e?imffncy and ablhty to compéte Successfully Wlth
ef?cient market theory. Conversely, another school of
other ?rms in its ?eld. Thus ‘many bClICVC' that ?rms With
thought Says one Should buy Stocks that have been most
ingctil pro?t gaging“? bettecr) lnvgsténenti’ 5mg: they are tile
battered by the market. In this application, relative strength
6.21.6.3 1‘!
elf 1n us mes‘
me .n S n? pro
margms ,y
and price performers Will be used interchangeably. See Table
dividing income before extraordinary items (A company s . . - 55 12 and Table 12A for comparison of 1-year relative strength
income after all expenses but before provisions for
h
dividends) by net sales. This is then multiplied by 100 to get a percentage. See Table 10 and Table 10A. History shoWs
C anges'
using high pro?t margins as the only determinate in buying a stock Will lead to disappointing results.
TABLE 12
TABLE 10 Category Best pro?t margins All Stocks Best pro?t margins Large stocks
. 1-year relative strength
6O
Category
$1M’S
Sharpe Ratio
.74 .75
34 4O
65
Best All Stocks Best Large stocks Worst All Stocks Worst Large stocks
$1M’s
Sharpe Ratio
3.31 2.98 .03 .49
43 49 3 29
US 6,317,726 B1 9
10
TABLE 12A
PE ratios beloW 20 and positive earnings gains for the year and bought the 50 With the best 1-year price performance, one Would actually earn less than if one bought the loW PE,
1-year relative strength Category
$1M’s
Sharpe Ratio
Best All Stocks
4.11
43
Best Large stocks
4.43
51
Worst All Stocks
.04
Worst Large stocks
.61
5 30
high relative strength stocks alone. The addition of positive earnings gains hurt performance in this instance. More factors do not necessarily mean better performance.
Buying stocks With strong 1-year earnings gains and strong relative price strength outperforms the All Stocks universe. A tWo-factor model that requires stocks from All 10
While All Stocks Best 1-year relative strength had an impressive yield, it had a high standard deviation and therefore high risk that brought the Sharpe ratio to 43, under the All Stocks universe’s 47. Large Stocks Best 1-year relative strength had an impressive yield With slightly more risk than the Large Stocks universe: the resultant 49 Sharpe ratio is higher than the Large Stocks universe’s 45.
also outperforms the All Stocks universe. See Table 14 and Table 14A.
Category a
goal.
- ear 115 toe k s 1y
$1M’s Sharpe Ratio ains a b ove 250/0; 'g g'
earnin s;
Top 50 1-year price Appreciation (all Stocks) ROE>15; Top 50 1-yearPrice 20
performance is dramatically outperformed by the market. See Table 12 and Table 12A. Instead of focussing on the effects of either a single groWth or value factor, using several factors alloWs one to enhance performance or reduce risk, depending on one’s
TABLE 14
15
Price momentum conveys different information about the prospects of a stock and is a much better indicator than
factors such as earnings and groWth rates. Worst 1-year price
Stocks to have 1-year earnings gains exceeding 25 percent and then buys the 50 With the best 1-year price performance
. 63
50
9.4
55
appreciation All Stocks
1.7
47
(Large Stocks) ROE>15; Top 50 1-year Price
2.3
45
appreciation 25
TABLE 14A
From All Stocks With a PE ratio beloW 20, take the 50
stocks With the best 1 year price appreciation. Or from All Stocks With a price-to-book ratio beloW 1, take the 50 stocks With the best 1 year price appreciation. Either multi-factor
analysis outperforms All Stocks, and outperforms either of
30
Category
$1M’s Sharpe Ratio
(all Stocks) 1-year earnings; gains above 25%; Top 50 1-year price Appreciation (all Stocks) ROE>15; Top 50 1-yearPrice
14.38
56
appreciation
the factors that comprise it (i.e. loW PE ratio or best price appreciation). See Table 13 and Table 13A.
All Stocks
2.68
49
(Large Stocks) ROE>15; Top 50 1-year Price
3.92
49
appreciation TABLE 13 Category
35
This model Was not able to be tested in the Large Stocks
$1M’s Sharpe Ratio
50 Stocks With PE ratios beloW 20 and best 1
8.6
55
50 stocks With price-to-book ratio beloW 1 and
10.6
61
best 1 year price appreciation All Stocks PSR
