Hi Tom,
Thought you would find this report from MSN Investor interesting.
Slicing and dicing to find Nasdaq’s best stocks
We've crunched the numbers so you don’t have to, but feel free to follow along as we check our work. Our regression formula comes up with 5 winners and 5 losers.
By Victor Niederhoffer and Laurel Kenner
The big winners in the Nasdaq 100 so far this year have been stocks with low prices relative to earnings and sales, big declines last year and -- surprise -- poor Value Line timeliness rankings.
Start investing with
just $500.
That's our conclusion after considering each stock's earnings, sales, book value, research expenditures and Value Line rankings through of the end of last year.
We distilled our findings on the relative importance of those variables into a regression formula -- so called because it looks back. And we used the formula to forecast, with great trepidation, the five best and five worst Nasdaq 100 ($NDX.X) stocks for the coming five months. This pairs trade appears at the end of today's column, with many caveats.
A few notes on our methodology:
We conducted our study from the vantage point of an investor with access to all published and announced information about prices and fundamentals as of year-end 2000. We calculated our own ratios to make sure they didn't include information that was published after the cutoff -- bias common in commercial databases.
We inverted the standard price-earnings, sales-price and book-price ratios to let us deal properly with negative numbers. The typical study, we are sorry to say, simply throws the negative numbers out, as they can't be ranked on a continuum. An "expensive" stock such as eBay (EBAY, news, msgs) (P/E=166) has a very small E/P: 0.6. (Actually, 0.006, but we’re multiplying our results by 100 to make them easier to handle.) A company that lost money, as Echostar Communications (DISH, news, msgs) did for the 12 months in our study, would have a P/E of -5.7, but an E/P of -5.9.
The research-price ratio is our preferred method of evaluating research spending. It shows how much research a company spends per dollar of price. The more usual research/sales ratio doesn't capture whether investors are already paying richly for expected returns on research spending. The ratio can be computed by dividing the total dollar value of research spending by the total market value of the company's common stock. (Readers with access only to the more usual ratios can arrive at the same result by taking the research/sales ratio and dividing that by the price/sales ratio.
The Nasdaq 100 itself is a modified capitalization-weighted index of the 100 stocks with the largest market values among the 5,000 stocks that trade on the Nasdaq. At present, the most heavily weighted Nasdaq 100 stocks are Microsoft (MSFT, news, msgs), Intel (INTC, news, msgs), Qualcomm (QCOM, news, msgs), Cisco Systems (CSCO, news, msgs) and Oracle (ORCL, news, msgs), together accounting for 28% of the index.
Now that we've explained what we did, here are the results. The following table shows the 10 best and 10 worst performers in the Nasdaq 100 during the first five months of the year, together with the key variables.
Key year-end financials of 10 best-performing Nasdaq 100 companies
(all ratios multiplied by 100)
Company Sales/Price Book/Price Earnings/Price Research/Price Value Line Timeliness % Chg Last Year % Chg This Year
eBay (EBAY, news, msgs) 4.7 11.7 0.6 0.6 2 -47% 98%
Compuware (CPWR, news, msgs) 86.2 0.0 9.1 3.5 4 -83% 59%
Microsoft (MSFT, news, msgs) 9.6 20.8 4.0 1.6 3 -63% 60%
KLA-Tencor (KLAC, news, msgs) 23.1 30.2 7.6 3.9 3 -37% 47%
Electronic Arts (ERTS, news, msgs) 25.1 17.1 1.4 3.9 3 1% 36%
Dell Computer (DELL, news, msgs) 66.6 0.0 5.3 1.1 4 -66% 40%
Apple Computer (AAPL, news, msgs) 149.0 72.5 14.7 7.3 5 -71% 33%
Echostar (DISH, news, msgs) 25.3 0.7 -5.9 0.0 3 -51% 39%
Bed Bath & Beyond (BBBY, news, msgs) 36.0 12.7 2.7 0.0 1 29% 32%
Applied Materials (AMAT, news, msgs) 2.2 29.0 0.3 3.6 3 -40% 28%
AVERAGE 42.8 19.5 4.0 2.6 3.1 -43% 47%
Key year-end financials of 10 worst-performing Nasdaq 100 companies
Company Sales/Price Book/Price Earnings/Price Research/Price Value Line Timeliness % Chg Last Year % Chg This Year
Ariba (ARBA, news, msgs) 2.7 10.3 -7.7 0.3 n/a -40% -90%
XO Comm. (XOXO, news, msgs) 12.6 21.1 -17.6 0.0 3 -49% -82%
Palm (PALM, news, msgs) 6.9 7.0 0.3 0.5 n/a -70% -80%
Appl.Micro Circ . (AMCC, news, msgs) 2.2 23.3 0.7 0.5 1 136% -75%
Network Appliance (NTAP, news, msgs) 4.4 0.0 0.3 0.6 2 55% -70%
Level 3 Comm . (LVLT, news, msgs) 10.0 33.6 -12.8 0.0 5 -58% -66%
Juniper Networks (JNPR, news, msgs) 1.5 2.1 -0.2 0.2 n/a 122% -68%
I2 Technologies (ITWO, news, msgs) 5.7 34.7 -0.4 1.0 1 12% -63%
McLeod USA (MCLD, news, msgs) 17.7 29.9 -5.7 0.0 4 -26% -59%
Metromedia (MFNX, news, msgs) 3.5 50.7 -7.4 0.0 3 -58% -56%
AVERAGE 6.7 21.3 -5.0 0.3 1.9 7% -71%
Some major differences emerge. Five of the 10 winners had a research-price ratio of 3 or higher, with the average being 2.6. Among the 10 worst, none had a research-price ratio of more than 1, and the average was 0.3.
As for the earnings-price ratio, the average for the 10 winners was 4 (which corresponds to a P/E of 25), with only one, Echostar, showing a deficit. For the 10 losers, the average earnings/price ratio was -5, with seven showing a negative ratio.
The average sales-price ratio for the 10 winners was 42.8 (which corresponds to a price-sales ratio of 2.3), with only three -- Microsoft, Applied Materials and eBay -- below 20 (which corresponds to a price/sales ratio of 5). For the 10 losers, the average was just 6.7, with none above 18 (corresponding to price/sales ratios of 14.9 to 5.5).
Looking at the entire sample of 100 companies, we can calculate the correlations between each of the variables and the subsequent performance of the stock. (As readers may recall from our last column, the correlation coefficient expresses the closeness of the relation between the variables on a scale of -1 to 1. If the variables' moves are unrelated, the coefficient is zero. The greater the tendency to move in the same direction, the closer the number is to 1. The greater the tendency to move in the opposite direction, the closer to -1.)
Correlations between key variables & 2001 YTD return
Variable Correlation w/ 2001 YTD return
Sales-Price 0.29
Book-Price -0.01
Research-Price 0.10
Earnings-Price 0.22
2000 Return -0.20
Value Line Timeliness Ranking 12/00* 0.29
*Only 83 companies ranked by Value Line
Thus, we conclude that the key determinants of superior price performance in 2001 were high sales-price ratio, high earnings-price ratio, high research-price, big decline in price during year 2000 and a poor Value Line Timeliness rating (highest number, 5, is the worst in their system).
We won't draw any conclusions about the value of fading Value Line's ratings so far this year. The sample is too small relative to the thousands of stocks the firm covered -- and the firm is perhaps entitled to five bad months after 40 good years. We did ask Sam Eisenstadt, Value Line's research chairman, to comment. His response: "If you have done well investing in 4s and 5s rather than 1s and 2s, you've been a lucky investor or trader, since the long-term record of those rankings clearly put you at a disadvantage. That's not to say you can't win picking a 4 or 5 -- it's just that the odds are against you."
He also noted the renewed importance of low price/earnings and price/sales ratios in the latter part of 2000 and the early part of this year, which comes after a lag for value stocks of seven or eight years. "After spending most of the ‘90s in the dumps, these stocks have finally come to life. Such a move was long overdue," he said. "But it still leaves this group far behind where they were in the ‘80s. Once the economy shows signs of revival, I suspect you'll see the return of the growth players."
We agree with Sam and note that it's quite possible that the five months of our study were anomalous. We're therefore highly reluctant to use our results as the basis for predicting which companies will be best for the future. Nailing the question down to our own satisfaction would require studies of several entire years, a job involving tens of thousands of hours. Even if the formula proved accurate for the next seven months, and if back-testing confirmed its usefulness in previous years, there would always be the question of whether cycles would be due for a change in future periods.
After a scientific study of this nature, we always feel between Scylla and Charybdis. On one side, we have our editors and headline writers asking, "What's in this for our readers besides a nice description of what happened?" On the other side are our critical readers: "Stop your forecasting now!"
In an effort to balance the two, we have come up with a group of trades for the reader's consideration, based on an equation we arrived at through regression analysis of our data. The formula is the best way of explaining the price changes in the first five months of this year:
Return = -20 + (0.2 x Sales/Price Ratio) -- (0.1 x Year 2000 Price Change) + (0.7 x Earnings/Price Ratio)
Here's how it worked for Apple Computer (AAPL, news, msgs). Apple's sales/price ratio was 149 at the end of 2000. It declined 71% last year. And its earnings/price ratio was 14.7.
Plugging in those three numbers into our formula, we arrive at:
-20 + (0.2x149) -- (0.1 x --71) + (0.7 x 14.7) = 27% return.
Apple's actual return for the first five months was 33%.
For the Nasdaq 100 companies, the return predicted by applying the above formula has a rather high 40% correlation with the actual return. The possibility of that occurring by chance is about 1 in 200.
There are caveats squared in using the formula for forecasting: our own limitations multiplied by the limitations of extrapolating a five-month study based on year-end values. Furthermore, the tendency of stocks to switch course at the start of a new year and maintain momentum throughout the year may defeat our forecasts.
With those warnings in mind, we offer the five stocks that our regression formula predicts will do best over the next five months:
Smurfit-Stone Container (SSCC, news, msgs)
Paccar (PCAR, news, msgs)
Staples (SPLS, news, msgs)
Conexant (CNXT, news, msgs)
ADC Telecommunications (ADCT, news, msgs)
The five companies with the worst forecasts are:
XO Communications (XOXO, news, msgs)
Level 3 Communications (LVLT, news, msgs)
PeopleSoft (PSFT, news, msgs)
IDEC Pharmaceuticals (IDPH, news, msgs)
Genzyme (GENZ, news, msgs)
Recent Articles
--------------------------------------------------------------------------------
• A long, hard summer ahead for investors, 5/31/01
An investing tool you can do without, 5/24/01
• How the Fed's moves shake markets, 5/17/01
more...
We will be happy to send the pertinent data for each Nasdaq 100 stock to all readers requesting.
With all of our caveats in using our study for forecasting, some may wonder why we bother to study such questions at all.
Good question.
At the beginning stages of a science, it is always necessary to create a database that provides an accurate description and classification of the key variables that determine the phenomena to be explained. The examples of Dimitri Mendelev's work in developing the periodic table of the elements in 1869 and Johann Kepler's careful observations on the orbits of planets at the turn of the 17th century are well known to most students of science in this regard.
While studying the determinants of stock price performance in the last five months is not on the same plane of significance as chemistry and orbital science, it did provide in its own right quite a few difficulties in implementation. To come up with prospective variables, for example, we had to go to four original sources: the S&P Stock Guides, the Value Line ratings and reports, and the Bloomberg data banks. The process of collecting and verifying accuracy was quite a daunting task, even for just this one period and 100 stocks. Only in recent years, with the advent of these data files and computers, has such a study become possible. We'd like to see it expanded, so as to be able to appreciate its significance down the road.
We'll report Nov. 8 on the performance of our five-month pairs trade. And we will revisit this topic after the end of the year to see if the variables that determined performance in the first five months continued to be important over the next seven months.
In the meantime, as a service to our loyal readers, we will e-mail our data files on the Nasdaq 100 from year-end 2000 through May 31 to all requesting.
At the time of publication, neither Vic Niederhoffer nor Laurel Kenner owned any stocks mentioned in this column. |
