In Part IV of this series, we looked at what would happen if we bought $200 worth of each of the S&P 500 stocks on April 2005 on October 3, 1980 or when the stock was first issued, whichever came first.  We did not sell any of the stocks.  We did historical testing of an efficiency signal on today’s S&P 500 data going back to the year 1980.   We sold everything on April 4, 2005, which is when our database ended.   Our initial $100,000 in equity became $3,025,960 while paying $31,356 in commissions.  Our gain amounted to a compounded return of 14.89%. 

Now we are going to compare buying and holding two databases.  We have an accurate S&P 500 database (i.e., it is split and dividend adjusted), which only has stocks while they are a member of the S&P 500.  In other words, it drops stocks when they are dropped from the S&P 500 and it adds them when they are added to the S&P500.  Thus, Microsoft doesn’t become part of our database until July 1, 1994 rather than its first publicly traded date of March 17, 1986.  And Dell isn’t included until October 1, 1996 rather than its first publicly traded date of June 24, 1988.  In the previous article (Part IV), Dell produced a huge R-multiple of 1700R, whereas, in the new study it is reduced to 711.6R (where 1R is assumed to be our entire up front investment of $200).

However, our database does not begin until February 1, 1990, so we are missing ten years of data.  As a result, we decided to repeat the study with the April 2005 S&P 500 database beginning on February 1st and then compare that with the accurate S&P 500 data.  Thus, the data for both databases will begin on Feb 1, 1990 and end on April 20, 2005.

Part I:  Buying and Holding the April 2005 S&P 500 (from 1990 or on the date when they first came out as stocks)

In our first study, we simply bought $200 worth of today’s S&P 500 in October 1980 or whenever they came out as stocks.  Thus, we were still purchasing $100,000 worth of stock, but once we bought we didn’t sell unless 1) the stock stopped trading or 2) the database ended on April 20, 2005.   Those were the only two exits.  Thus, this is a real buy and hold situation.  However, we are basically buying the BEST American companies.  We were also buying them either on the start date or when they first came out as stocks.

In Part IV, we started with $100,000 and ended up with $3,025,960.  Our gain amounts to a compounded return of 14.89%.   We made money on 94.78% of our trades and the average gain was 71.35 times the average loss.

The only difference between the study reported in Part IV and this data set is that we started later in this data set.  And the results show that when we start our buy and hold in 1990, our ending equity is $1,745,611.   This is about 1.3 million less than our prior ending equity, but the average yearly compounded return on equity increased from 14.89% to 20.66%.   However, the prior dates include the 1980-1982 bear market and the 1987 crash.   Buying and holding isn’t too profitable during such times even when you have the best stocks in America.  We still had a drawdown of 48.6% on August 31, 2002 and we were still in a drawdown when the data ended in April of 2005.   In this particular run we had 466 wins and 33 losses for a 93.39% win rate.  And the average gain was 27.42 times bigger than our average loss.

Figure 1 shows the equity curve of buying and holding America’s top stocks for 15 years.

Figure 1: Equity Curve over 15 Years

While the results are great from 1990 through 1999, it’s also clear that if you had bought everything in 2000, you still be down by 2005.

Table 1 shows all of the stocks in the database with R-multiples of 30 or more.  And since our risk was 100%, this means that they increase by 30 or more times our original buy-price.  This occurred despite the price drop from 2000 through 2005.  Most of the stocks were purchased in 1990 with those purchased later showing a shaded entry date.  DELL is still the top stock with an R-multiple gain of 711, but that’s a significant drop from 1700 when we started in 1980.

Table 1: Top Stocks from 1990 through 2000
Rank	Symbol	Profit	R-Multiple	Entry Date	Exit Date
1	DELL	$142,318.40	711.59	01/31/1990	4/20/2005
2	CSCO	$42,343.25	211.72	02/21/1990	4/20/2005
3	UNH	$35,535.90	177.68	01/31/1990	4/20/2005
4	EMC	$27,999.20	140	01/31/1990	4/20/2005
5	BBY	$25,704.81	128.52	01/31/1990	4/20/2005
6	TWX	$25,560.48	127.8	03/23/1992	4/20/2005
7	ERTS	$19,364.12	96.82	01/31/1990	4/20/2005
8	APOL	$18,439.69	92.2	12/08/1994	4/20/2005
9	MXIM	$13,854.03	69.27	01/31/1990	4/20/2005
10	APCC	$13,617.49	68.09	01/31/1990	4/20/2005
11	CCU	$12,612.64	63.06	01/31/1990	4/20/2005
12	LLTC	$12,383.72	61.92	01/31/1990	4/20/2005
13	IGT	$11,880.61	59.4	01/31/1990	4/20/2005
14	AMGN	$11,540.08	57.7	01/31/1990	4/20/2005
15	QCOM	$11,539.10	57.7	12/17/1991	4/20/2005
16	DG	$10,955.30	54.78	01/31/1990	4/20/2005
17	ESRX	$10,093.23	50.47	06/11/1992	4/20/2005
18	PAYX	$8,071.74	40.36	01/31/1990	4/20/2005
19	ALTR	$7,953.31	39.77	01/31/1990	4/20/2005
20	CFC	$7,870.83	39.35	01/31/1990	4/20/2005
21	QLGC	$7,858.90	39.29	03/02/1994	4/20/2005
22	HDI	$7,443.30	37.22	01/31/1990	4/20/2005
23	MSFT	$7,427.44	37.14	01/31/1990	4/20/2005
24	BBBY	$6,958.10	34.79	06/09/1992	4/20/2005
25	PHM	$6,758.32	33.79	01/31/1990	4/20/2005
26	AMAT	$6,748.78	33.74	01/31/1990	4/20/2005
27	SCH	$6,523.66	32.62	01/31/1990	4/20/2005
28	VRTS	$6,256.91	31.28	12/13/1993	4/20/2005
29	SPLS	$6,132.63	30.66	01/31/1990	4/20/2005
30	SBUX	$6,045.70	30.23	06/30/1992	4/20/2005

Part II:  Buying and Holding the Real S&P 500 Stocks (from 1990 until they were delisted or until April 2005)

In our second database, we will simply buy $200 worth of the real S&P 500 stocks on February 1, 1990.  We will sell them when they are no longer part of the S&P 500 and buy new stocks when they become part of the S&P 500.  We will then sell everything at the end of our database in April 2005 (we actually had the data through 2007 but cut it off so as to compare both databases for the same time period).

Figure 2 shows the equity curve for this new database.

Figure 2 : Equity Curve of New Data

For this database, our final equity was $771,198 for a 14.36% compounded annual rate of return.  This is much less than the $1.7M and the 20.66% figures of the first database.  Also remember that we were fully invested with the real S&P 500 database, getting the 14% annual ROI,  but not fully invested until near the end with 2005’s S&P 500 database, yet still getting the 20.66% investment.

Something else that really stands out is that the drawdown during the subsequent bear market is much less for the real database.  Compare the two figures.  The worst drawdown for the real S&P 500 was on October 11, 1990 at 21.95%, which certainly says we were not holding the best performing stocks at that time.  And the longest drawdown is from September 1, 2000 to June 16, 2003.

In this database we took 858 trades and rejected 4 because of a lack of money.  We had 636 winners and 222 losers for a win rate of 74.13%.  Our average winner was 7.28 times our average loss.   Of the 858 trades, there were 139 with obsolete symbols (i.e., the stock no longer exists) amounting to 16.2%.   Only 20 of the 139 lost money.

Table 2 shows our summary returns to date with a beginning equity of $100,000.  Also, please note that we were only fully invested from the very beginning in the last study.  All the rest represented a gradual build-up of positions.

Table 2: Summary Results to Date
Study Conditions/Variables	Database	Ending Equity	Annual % ROI	Peak Drawdown
Efficiency  “Close – Close” Smoothing (coding bugs fixed), starts in 1980	2005 S&P 500	$48.347M	28.59%2	18.34%
Efficiency “Close/Close” Smoothing Starts in 1980	Same	$2.835M	14.58%	21.78%
Buy and Hold all Stocks from 1980 or Issue Date through 2005	Same	$3,026M	14.89%	51.27%
Buy and Hold all Stocks from 1990 or Issue Date through 2005	Same	$1.746 M	20.66%	48.60%
Buy and Hold all Stocks from 1990 through 2005 – real S&P 500	Real S&P 500	$0.771 M	14.36%	21.95%

In summary, the efficiency with “close minus close” smoothing is clearly better than buy and hold on the 2005 S&P 500 database.  And even the “close divided by close” smoothing is similar in terms of returns but better in terms of drawdown, suggesting that we could get much better performance with a position sizing algorithm designed to meet whatever our objectives might be.

However, at this point I’m not convinced that the efficiency trades that are being taken automatically by the studies are adequate.  In addition, notice at this point I still have not made position sizing adjustments to see what’s really possible with this sort of trading.  As a result, there is still a lot more research that we’ll do in this series, including:

1.   We’ll  determine what happens when we allow ourselves to take as many as 250 trades (i.e., half the S&P 500 database) at any one time with the two smoothing functions.  With 1% risk and a 25% trailing stop we are limited to 25 trades.  With a 0.1% risk and a 25% trailing stop, we are limited to 250 trades.  We’ll simply increase our starting equity to $1M so that we’ll be investing the same amount ($4000) with each trade. 

2.   As I’ve said, I’m not convinced, given these results, that I’m really buying the stocks I’d normally buy when looking at a chart.  As a result, I plan to look at charts of the 100 trades from both smoothing algorithms to determine how many of them look like “efficient” stocks.  This will give us a good idea to determine if we are looking at efficient stocks or not.  I still have not had the time to do this, so if any of you would like to do that and save me some time, I’d appreciate it.  Please let us know and we’ll send you the data. 

3.   We’ll also try other trend following algorithms including 1) an 180 day channel breakout and 2) linear regression to pick our trades.

All of that is still to come in subsequent articles and it looks like this series might continue for some time.

I think that Mechanica is capable of really answering a number of significant questions about comparing buy-and-hold versus various trend following.  We were fortunate enough to obtain an accurate 17 year database of the S&P 500 that was adjusted for splits and dividends and included numerous stocks that no longer exist.  However, our database was only 17 years (because it goes to Sept 07).  Since the S&P 500 was created in 1957, we’d be interested in knowing if anyone has an accurate database going back that far that we could use or even knows where I might obtain one.  That is, you have prices (dividend and split adjusted) for all stocks in the S&P 500 from 1957 through 2007 while they were members of the S&P 500.

1. If you have some interest in Mechanica, which we are using in these tests, then  go to the Mechanica web site -- http://www.mechanicasoftware.com.  Mechanica is the new windows version of Trading Recipes.

2.  We originally showed the compounded annual return to be 33.3%.  But when I was looking for the data set to fill in the rest of this table, the results were slightly different.

Until next week, this is Van Tharp.