How do you pick stocks? The adventuresome among us look for companies in the news, or those subject to takeover and other such rumors. If the news is positive, and the stock is up, they buy the stock, hoping it will go higher. Some of these are lucky, but most tend to overpay and lose money.
Others buy stocks in companies that make products they like. Most of the Apple (AAPL) investors belong to that category, I guess. In the long term they could win big, and if they don't make money, at least they own a company they like.
Some of us, the cool-headed types, are knowledgeable in accounting. They pour through balance sheets, trying to find what a company is worth. This is called value investing. Warren Buffett is known to employ such a technique. A variant of this more conservative style is the dividend investing strategy.
Fundamental analysis is important, however, the common argument against the fundamentals style investing is that most of the time the current stock price already reflects the known fundamentals. Thus, they say that buying one stock has no advantage over the other. Also, if the fundamentals don't change daily, how come the prices do? Thus, those who solely rely on fundamentals are missing a big chunk of information encoded in the daily price movements.
Then there are the chart readers, the technical analysts. They rely on chart patterns and known indicators to find oversold or overbought stocks. From the past patterns, they attempt to project future patterns. But, as the world around them is constantly changing, how could they expect the past pattern to repeat itself? Also, the indicators -- the patterns -- tend to lose effectiveness when too many people trade upon them at the same time.
The Algorithmic Method
We use computers, math and constantly learning algorithms to pick stocks. Markets move in waves, and our algorithms are designed to detect and predict the waves. Many inputs from different sources go into each algorithmic forecast. The output for each stock is the signal, up and down, and the predictability.
The use of math and algorithms eliminates a big problem many of us share. Humans are emotional. When it comes to money decisions, we are not rational and too impulsive. We tend to buy when we should be selling, and sell when we should be buying. Also, most of us can't read balance sheets, can't digest the news and make informed buy or sell decisions.
The Forecast Table Explained
The "heat map" table in Fig. 1 is the forecast issued on the morning September 27, before the market open. Each cell in the table represents a stock or index. Along with the ticker, there are two numbers -- the signal (middle right) and the predictability (bottom left). The stocks are arranged by the signal strength, left to right, and top to bottom.
On the top left corner of the table is the strongest up signal -- American International Group (AIG) -- followed by the KBW banking index (BKX), Orbotech (ORBK), General Cable (BGC), and Goldman Sachs (GS).
The strongest down signal is in the bottom right corner -- Suntech Power (STP) -- followed to the left of it by LDK Solar (LDK), Advanced Micro Devices (AMD), the Ireland index (ISEQ) and Bank of America (BAC).
The signal is the forecast move direction, positive "up," or negative "down." Predictability is the historical correlation between the forecast and the actual move. Thus, it is related to the probability of the predicted move occurring. Cell colors are indicative of the signal -- green for "up" and red for "down". The colors help to identify the predominant market direction.
Case Analysis of the September 27 Forecast
We thought it would be an interesting exercise to find out whether our stock market forecast also could have been anticipated by conventional analysis methods.
Among the "I Know First" last month top 10 stock picks, five long and five short picks (Figure 1), eight went in the predicted direction, and two against the prediction. The forecast we are about to analyze had a one month time horizon.
First, the top five long stocks:
AIG had the strongest up signal of the September 27 forecast. The stock was not in the news headlines then. It gradually went up 13.6 percent to a peak on October 18 before pulling back, ending the month at a 6 percent gain.
We were not the only ones to foresee the AIG climb. A day before our forecast, fellow Seeking Alpha contributor David Zanoni wrote in his September 26 article: "AIG looks like a solid risk-on QE3 investment. It is significantly undervalued and has above average expected earnings growth."
The KBW Banking index went up as predicted (up 3 percent until October 18, with 0.1 percent gain at the end of the month). I could find no reference to the BKX index in the recent articles.
General Cable (BGC), went up 7.1 percent from our forecast to a peak on October 18 on little news at all before pulling back, ending the month at a 2.1 percent gain. Two news items in Forbes emphasized the value in BGC: "General Cable Shares Cross Below Book Value", and this: "General Cable had its numbers increased by Keybank (KEY), as the purchase of Alcan should add to the bottom line. A $37 price target was set with a buy rating."
Goldman Sachs (GS) climbed 10.7 percent to a peak on October 18 (5.6 percent gain by the month's end). I could not find anyone predicting that. GS is a black box to me, and apparently to others, too.
These stocks had no change in visible fundamentals, but the algorithms have predicted their rise. Short of getting into the most inner logic of the algorithm, I can only give a very general explanation: The climb could be attributed to the general stabilization and increased optimism regarding the future of AIG and the U.S. banking system, and the improving conditions for the essential industries, of which BGC is a representative.
Next, for the short stocks recommendations:
The algorithm has predicted that STP and LDK would go down, and down they went. I could find no one on the web who predicted that. The sharp decline of STP (-12.5 percent by October 18, and 9.1% by the month's end), and of LDK (-32 percent at peak, and -29% by the end of the month) is the result of the continued depression in the previously over-hyped, non-essential solar industry. If the budget is tight, who is going to invest in the long payback time pf projects like solar with questionable economics? On top of these solar industry troubles is cheap oil and gas, and the China trade wars with the U.S. Eventually, the solar industry decline will stop. The question is, when?
Bank of America (BAC): In spite of ongoing legal and other troubles, it managed to add 7 percent, bucking our forecast. Possibly the market optimism about the future of the banking sector could be the reason.
As predicted, AMD was down 21 percent by October 18, and 37.6% by the end of the month. As Mike Rest pointed out on September 27: "Anyone who has been long Advanced Micro Devices for the past 6 months may be looking for a large cliff to visit."
Orbotech, for which we had an up signal, climbed 6 percent by October 5, but then went down the next day due to disappointing company guidance for 2012, finishing the month at a 1 percent loss. I could find no mention of ORBK in recent articles.
The stock picking algorithms are not "black magic." If you patiently and follow the fundamentals, the daily price moves, the company reports, the news, the competitors, the suppliers, the patents, the lawsuits, the politics, the weather (yes, this too), you could, at least in theory, predict the stocks' movement. Some of the forecasts have been or could have been anticipated by those who do follow all of these. But the algorithms do it differently, using number crunching computers and cold, heartless logic.
The algorithm doesn't know everything we know. Not every piece of available information is used to make a forecast. It's a tool you can use to focus on opportunities. Combined with other in-depth analysis, such as fundamentals, it's one more tool that helps in making informed investment decisions.