You are here

Thinking About Cypress Stock | Cypress Semiconductor

Thinking About Cypress Stock

Last Updated: 
May 03, 2016


This is an analysis of the historical valuation of Cypress stock. It is not a recommended strategy for trading shares, although the method used to study stock valuation entails the use of hypothetical trading strategies carried out on actual Cypress historical stock price trends.

We have recently had an unprecedented number of calls from individual investors who bought Cypress shares near the all-time-high price of $27.38 per share. Many of the investors ask, “The stock is down 50%; what’s wrong with the company?” Since Cypress met or exceeded Wall Street expectations for earnings per share in every quarter of 1995, the answer relates more to a stock market situation than to a performance problem. The intent of this analysis is to provide data and analytical methods to help shareholders better answer the following questions:

  • Did I pay too high a price for Cypress when I purchased its shares at $20-plus?
  • If I choose to hold my shares, how long will I have to wait to break even?
  • If I trade Cypress shares in the future, how will I know if the shares are high priced—or a bargain?

No analysis can provide answers to questions that require a prediction of the future. However, we can analyze data for Cypress’s 2,515 trading days over the last ten years, and show typical trends for the pricing of Cypress stock, and for the time required historically for an investment to produce a capital gain. Shareholders should not expect that an historical analysis of share price data can be extended accurately into the future (the data itself shows just that), but this ten-year analysis should provide useful added information for making trading decisions.

Semiconductor stock prices declined at the end of 1995 after analysts’ reports on softness in the semiconductor market. Cypress’s share price fell following a series of reports detailing the “crash” in prices of our highest volume product, the 256K-bit static RAM. Despite 47% revenue growth and 82% earnings growth in 1995, our P/E ratio fell to 8.3 at year-end. We believed that our shares had become undervalued, and decided to buy back $70 million of stock with the express intent of reselling those shares at a higher price in the future to raise money for expansion. At year-end, many analysts advised investors to “hold” (usually a euphemism to sell) our shares—while Cypress was buying. Some of the pessimistic news about the semiconductor market was carried on financially oriented cable TV programs which are literally displayed in windows on the computer screens of some stock traders. The problem: How can a small investor who may always be late in reacting to the latest news on the spot-market price for the 256K static RAM hope to compete against institutional investors and stock traders with real-time television hook-ups?


The prices for 256K static RAMs are currently falling—but static RAM price “crashes” have been a constant in our business since the static RAM’s invention in 1970. During this period of dramatic price reductions, the semiconductor industry has grown from $2.6 billion in 1970 to $144 billion in 1995. We can all appreciate how industry sales might grow during a period of dramatic price reductions, but what problem might that present to a company like Cypress, whose price for a 256K-bit static RAM might drop as much as 50% in 1996? (Consider the effect of an equivalent drop in car prices from $20,000 to $10,000 in one year.) The answer is that semiconductor companies routinely absorb significant price reductions with no long-term negative impact.

Price reductions measured in percent per year imply that exponential equations describe our industry’s economics. Exponential equations are notoriously difficult to comprehend. Even slow exponential growth rates like that of population create counter-intuitive results, like this fact caused by exponential growth: “More people are living on earth today than the total number of people who have lived and died in all history.” In an industry that is governed by the implausible outcomes of exponential mathematics, how should an investor make quick, rational decisions—especially if that investor’s position is “under water” and the analysts are strapping on their life jackets?


Cypress’s share price has behaved like that of many good technology companies: up and down, but up over the long haul. Nonetheless, investors sometimes buy Cypress at an inopportune time, just before the share price drops—in many cases, exactly at a time when many analysts are pushing the “strong fundamentals” of Cypress and the semiconductor industry at large. So-called momentum investors “buy high” and hope for the stock to go higher. In some cases, that strategy traps investors into having to hold their shares for years before realizing a profit. Or, if the investor’s patience or courage runs out, the momentum strategy may end up as a “buy high, sell low” strategy—a loser.

The strategy to buy low and sell high is, to use a phrase from GE’s Jack Welch, “simple, but difficult.” “Buying low” often means going against analysts’ warnings, and “selling high” often means selling into a bull market; in other words “buy low, sell high” often translates into “overcome fear, resist greed”—a very difficult task for investors or analysts.


In the semiconductor business, our process yield—the number of chips we ship divided by the number of chips we start on our lines—can vary from 5% for a new product to 90% for a mature product. Our “learning curve,” the rapidity with which we move from 5% yield to 90% yield, separates winners and losers in our business. In the semiconductor struggle, we have learned to act more on what the data presents, and less on our instincts. This analysis of Cypress share price is based solely on data. The data is available to all investors: daily share price and Cypress’s sales for the prior quarter. The analysis eschews the analysis du jour in favor of a statistical look at ten years of daily Cypress trading data. I use this model to help make Cypress’s buy-back decisions, and my personal trading decisions.

The most common stock valuation method is the P/E ratio: the price per share, divided by analysts’ forward-looking earnings per share estimates. The P/E ratio tends to fluctuate dramatically because it not only depends on predictions, but also on fluctuations in profitability. To reduce the volatility inherent in the P/E ratio, I have chosen for a valuation index the “P/S” ratio, the ratio of our price per share, divided by our annualized sales per share, where “annualized sales” equals four times the revenue of the last reported quarter. The P/S ratio also equals Cypress’s market capitalization divided by its sales. As defined, the P/S ratio needs no predictions, and can be calculated and trended by any investor.

P   =          Price/Share               =         Market Capitalizations
S        Annualized Sales/Share                   Annualized Sales

Cypress’s daily P/S ratio for ten years appears below. Our P/S ratio dropped dramatically the week of “black Tuesday” in October 1987. Since that time, the P/S ratio has been relatively stable with a median value of 2.23, and 10th and 90th percentile points of 1.43 and 3.46, respectively. One can then say quantitatively, for example, that Cypress shares were relatively “low” whenever the P/S ratio was below 1.43, because on 90% of the 2148 trading days between 10/26/87 and 1/24/96, the share price was relatively more highly valued.


The portion of the P/S curve from 1992 to 1993 highlights the period during which Cypress bought back about 10 million shares of its stock because—as we said to investors consistently at the time—we felt our shares were undervalued. We had just restructured after our only loss year in 1992, and we had a plan to improve in every quarter of 1993 and 1994. Analysts and investors did not believe in that plan, and our stock remained a great investment opportunity for us at less than $5 per share. The dot on the P/S chart highlights the point at which we sold most of the buy-back shares at a substantial gain in our $110 million convertible subordinated debenture offering. The proceeds were needed to fund Fab IV in Minnesota. In that sale, we did not follow the practice to “sell high,” based on the P/S ratio; we needed the funds for growth and simply took the gain presented to us at the time. The final highlighted section on the P/S graph in late 1995 represents our current $70 million buy-back program—again at “buy low” prices. We are committed to the current buy-back program because we believe in our future.

  • We believe:
  • Total semiconductor sales will grow more in absolute dollars in the next 5 years than they have in all of the last 35 years (as do the majority of semiconductor analysts).
  • Cypress can continue to meet its plan to grow faster than the market, as we have in nine of the last eleven years.
  • Our buy-back program will therefore provide us with the funds for plant and equipment with less dilution.
  • Of course, the challenge for us is to turn our beliefs into reality.


The analysts generally believe:


  • The slow PC market at Christmas, and the problems it caused semiconductor companies, should be a sign of caution.
  • There may be over-capacity in our static RAM market, and in the semiconductor market in general.
  • Despite pervasive bullish five-year forecasts for the total semiconductor market, the worries and rumors surrounding the current market make investment unwise.

Consequently, many analysts have recommended to “hold” our shares recently, which means, we believe, to “sell low.” Cypress is acting on the reverse strategy to “buy low,” based on our plans for the future and an analysis of our share price history.


The time it takes to make the first incremental gain on an investment in Cypress’s shares may be referred to as the “time to money,” or TTM. The time to money for every Cypress trading day is graphed below as a function of the P/S ratio. The insert explains the graph: for buyers who bought Cypress on one of the 8 days in the last eight years when the P/S ratio was 2.23, 3 saw the stock go up the next day, 1 saw the stock go up in two days, and the least fortunate investor waited for 19 days until the stock closed above its purchase price. The complete graph, therefore, represents the time to money for every trading day, sorted by P/S ratio. The time to money has consistently been below 30 days, but waits of hundreds of days for a capital gain have occurred. Long TTM times were more likely to happen to investors who bought at a high P/S ratio; i.e., for momentum investors.

On the other hand, buyers who bought low, as defined by a P/S ratio equal to or less than the median of 2.23 or less, achieved a time to money of 130 trading days or less with a 99.2% probability, as shown below. The future cannot always be predicted from history. For example, using the P/S ratio trends from May 1986 to October 1987 (before the steep decline on the P/S ratio graph) would not have been good investment predictors after October 1987. But after that point and through January 1996, the P/S ratio was reasonably stable, and simple investment rules would have been effective in making trading decisions. The algorithmic trading rules outlined below probably would have been preferable to momentum investing during the period of the study for the small investor who was unable to react quickly to semiconductor news and rumors on “the street.”

We named our computer HAL, and programmed him to make trades in Cypress shares according to our predetermined rules over an 8.3-year period from 10/26/87 to the most recent trading day in this analysis, 1/24/96. For the first simulation, we gave HAL these instructions:

  • Start with $100,
  • invest the cash at 5% money market,
  • buy low: put all cash into the market for P/S < 1.43 (10th percentile),
  • sell high: put all cash back into the money market for P/S > 3.46 (90th percentile),
  • pay $0.10 for each share traded,
  • calculate annual return based on combined trading gains and money market returns,
  • but do not count the last trade if a sell transaction is pending. (This rule gave roughly equivalent results to a forced sell on the last day of the study.)

With these instructions, HAL made only two trades in eight years:

Action Date Share Price # Shares Value
Deposit 10/26/87 $ 3.13 - $100.00
Withdraw 4/13/92 $ 5.19 - $124.25
Buy 4/13/92 $ 5.19 23.5 $121.91
Sell 5/23/95 $18.250 23.5 $426.53
Deposit 5/23/95 $18.250 - $426.53
Withdraw 1/24/96 $14.375 - $440.29



That trade provided a 49.7% per year capital gain over the holding period, and a 19.7% annualized capital gain over the eight-year period, when blended with the 5% money market rate. The capital gain was great, but few investors would want to follow a strategy that required watching a stock for more than four years before investing for the first time. The computer then ran approximately 1,000 scenarios for rules-based investment identical to the one above, but with different buy/sell criteria.

One successful rules-based strategy was to “buy below average” and “sell above average.” That strategy was designed to buy any time the shares were below the 50th percentile point of P/S = 2.23, and to sell at some higher P/S ratio. The graph below shows HAL’s actual trades for the Buy P/S=2.25, Sell P/S=2.75 strategy, which yielded a 26.4% per year gain by making five high-gain trades in eight years. The table on the next page shows that the average “buy below/sell above average” strategy worked for a wide variety of cases to produce an annual return of approximately 25%. During the same 10/26/87 – 1/24/96 time frame, the most favorable “buy and hold” strategy, which allowed for purchasing shares at the all-time low price of $3.13, yielded 19.7% per year, while the S & P 500 index appreciated 12.9% per year.

A second set of simulations showed that a strategy to buy low and sell high would have yielded a reasonable gain over a very wide variety of buy and sell points in the middle of the P/S range. HAL ran 330 simulations with the Buy P/S between 1.6 and 3.2, and the Sell P/S ratio higher than the Buy P/S ratio by 0.4 to 1.75.

These were large-gain scenarios with the capital gain per trade ranging from more than 10% to over 100%. The total range of annual return figures for these widely varying strategies was stable, with a minimum gain of 16.8%, a maximum gain of 33.0%, and an average gain of 23.7%. These yields are graphed below.

No one can predict the future, but investors should at least consider that in following analysts’ advice, they may sometimes be making the “buy high” mistake when they are purchasing our shares. Investors should also be prepared to hold our shares for an appropriate TTM time to avoid having to sell shares at a loss.

For investors who cannot react rapidly to Wall Street news and rumors, and who are prepared to hold shares for 300-plus days, a simple rules-based “buy low, sell high” trading strategy might be superior to the strategy of reacting real-time to the counter-intuitive and erratic semiconductor market. At a minimum, the analysis presented here provides an independent check on trading decisions.

The author is indebted to Jeff Arenberg and Kevin Murphy for their competent technical support.