321gold.com
321gold.com
>
|
Introduction to the
SKI Gold Stock Prediction System Editor's Note. This is a complicated system, but people want to know how SKI comes up with his recommendations. Read it, file it away and we'll be posting SKI's predictions starting Wednesday. There will be 2-5 updates per week depending... as and when... Most gold and gold stock analyses focus on waves, cycles (time), and prices. The SKI system is a unique mathematical system that analyses the interaction of price and time. It was empirically developed in 1985 and has been cross-validated since that time using real-time data on a very broad composite of gold stocks. The goal is to predict the short (1-4 weeks), intermediate (1-2 months), and long-term (years) price fluctuations in gold stocks. Although my original gold stock composite is proprietary, U.S. GoldShares (Symbol: USERX) is used in public analyses. USERX is the oldest gold mutual fund with data available since 1974. Note that the gold stocks comprising USERX have changed dramatically many times over the past 26 years, but the system has continued to perform well. Generally, the system achieves 80% accuracy with the 20% errors involving small (1-3%) losses. This system does not trade frequently. Rather, it was designed to give high probability signals, with the probability of success being known in advance. The goal is to minimize market exposure while risking capital only when the probabilities for success are high and a close stop is provided by the system. Quite a few rises and declines are therefore "missed" by the SKI system. The system does not take into account any "fundamental" factors. It only focuses on the price by time interaction, and the velocity and acceleration of price changes at a specific time. Technical Description The system's base is composed of three indices: the 16-20 day index, the 35-39 day index, and the 92-96 day index. Three longer-term indices were also identified: the 218-222 day index, the 439-443 day index, and the 660-664 index. These indices are computed by comparing the current day's price to each of the index five back prices. For example, looking at the 16-20 index, the current price is compared to the price from 20 days ago, 19 days ago, 18 days ago, 17 days ago, and 16 days ago. If the current price is higher than the price 20 days ago, a +1 is assigned. If the current price is lower than the price 20 days ago, a -1 is assigned. If the current price is equal to the price 20 days ago, a zero is assigned. This is done for each of the 5 days comprising the index and is summed. The value for any day is therefore between -5 and +5. The index is further computed using the sum of the last 5 days of such values and is multiplied by four, yielding an index score between -100 and plus 100. Lastly, yesterday's index score is subtracted from the current day's index score, with difference divided in half and added to the current day's sum to yield the final index score for the day. When the index changes sign (i.e., from positive to negative or from negative to positive) a signal is generated for the close of the following day. This system was designed so that I could obtain buy and sell signals the day before they were to be executed. This means that one usually doesn't have to watch the market, one simply needs closing prices to generate signals to be executed for the close of the next day. The above math was based on the hypothesis that cycles exist, but the math also employs the concepts of the velocity and the acceleration of the daily price changes when compared to prior prices. Some people like to think of each index as being similar to a moving average. That is an acceptable idea, but the important difference is that the set of 5 back prices changes each day. Unlike a moving average that changes slowly, the 5 back prices can change quite dramatically from day to day! By looking at upcoming changes in the back prices, one can get a "feel" and "map" for the future of the market. Another way of viewing these indices is that they are objectively marking Elliot waves and are certainly marking critical points. The 35-39 and 92-96 indices were discovered by searching for indices which maximized profits when one bought as current prices rose over the back prices (i.e., the index sign changes from negative to positive) and vice versa. For example, when current prices rise over the back 35-39 day prices, this is termed a 35-39 index "buy" signal. The 16-20 index was discovered as the best CONTRARY index. When prices rise over the back 16-20 day prices, this is a "sell" signal." The SKI system becomes more complex based on patterns of "buy" and "sell" signals. A "path" is established based upon which of the 3 indices is in "control". The index that buys first is in control and on the path. For example, if the 35-39 index generates a buy signal and then prices rise to the 92-96 index signal, the 35-39 index remains in control until it sells or a higher order index (i.e., the 92-96 index) sells. Bull markets occur when the 92-96 index generates a buy signal that is on the path. Excellent bear market rallies occur when both the 35-39 and the 92-96 are on sells, the market rallies into a 16-20 sell signal, and then falls to a 16-20 index buy signal. The system works best at marking tops in bear markets and bottoms during bull markets. For example, during a bear market, prices will eventually rise to give 35-39 or 92-96 index "buy" signals which will usually come within a day of marking the top of the rally. The system employs an "XXing Out" procedure to label such "buy" signals as actually marking high probability shorting points. Simply stated, when the 35-39 and 92-96 indices provide consecutive sell signals, the next 35-39 and 92-96 buy signals are XXed Out. A 16-20 buy signal is XXed Out if it immediately follows a 35-39 or 92-96 sell signal. A second aspect of market prediction involves RUN ANALYSIS. A "run" refers to a pattern of daily up and down market closing prices. If the market has 3 consecutive days of higher closing prices, the run is "3 up". If prices then decline for 2 consecutive days, the run becomes "3 up and 2 down". If prices then close higher the next day, the run changes to "2 down and 1 up". Some people have referred to run patterns as "worms." A run pattern is only completed after the direction of closing prices has changed. I have compiled a listing of every run pattern that has ever occurred and generated probabilities that the end of the run marks a high or a low. These probabilities are moderated and enhanced based upon the current position of the 35-39 and 92-96 indices. For example, a "1 down, 2 up" run has a 98% probability of marking a top during an XXed out 35-39 or 92-96 "buy" signal, but has less than a 40% probability of marking a top during non-XXed out "buy" signals. I was able to go long one day after the August 1998 plunge bottom in gold stocks by noting the dramatic 2 up and 7 down run pattern. Although I have been unable to reliably apply the SKI system work to the stock and bond markets, I occasionally make predictions based upon run patterns and inter-market relationships. When my very broad basket of non-gold stocks declines for exactly five days while falling more than 1 percent per day, a buy signal is generated and vice versa. Since 1985, this rare pattern has generated the following signals: a sell signal at the market top in October 1987, a buy signal at the bottom in 1987 several weeks later, a buy in the Fall of 1990, a buy in April 2000, a sell on 9/5/00, and a buy on 9/24/2001. Bond prices and interest rates are frequently related to gold stock movements with a lag of 6 to 9 months. As gold stocks rise, interest rates will bottom (and bond prices will top) approximately 6-9 months later. Historical Overview The system generated its greatest run analysis buy signal in late August 1976 based upon a run analysis pattern involving a drop of 23% over 10 consecutive days and a fall of over 50% in 2.5 months into a USERX price of 1.13. Although prices rose slowly for years after that, the true bull market 92-96 index buy signal only occurred in March 1979 at 2.18. Prices rose to a high of 10.53 in September 1980 before generating a sell signal in December 1980 at 10.07 and a massive 92-96 sell signal in January 1981 at 7.38. During the 1974-1980 period I was a teenager to young adult, my father had lost his business and the family's small savings in the stock market decline, and I saved all my quarters only to see them eroded/destroyed by inflation. I vowed to never let this recur by learning how to predict the mass human behavior exemplified in financial markets. The system did not generate any major buy signals until late June 1982. Once again the gold stocks yielded a run analysis buy signal by plunging for 10 consecutive days, dropping 22% to USERX 2.94. A valid 92-96 buy signal occurred one month later at 4.01. Prices rose to high of 9.69 in late May 1983 and a 92-96 sell signal occurred one week later at 8.82. A small rise in the market generated the system's first "Triple Sell" in early July 1983 at 8.93. This extremely powerful and long-term pattern occurs when prices rise to a 16-20 sell signal while simultaneously failing to rise enough to stay above the 35-39 and 92-96 indices. Although there were many short and intermediate buys and sells along the way, the next major buy signal was executed on 7/15/1986, one day before the low at 2.96. This was a 16-20 buy signal which predicted a major bear market rally. Prices rose to an XXed out 92-96 "buy" signal (not on the path) which sold the system out at an exact high of 4.76 on 9/5/86. Prices gyrated up and down until the system bought back at a higher price of 4.86 on 2/19/87 and rose to many long-term sell signals at the exact high of 7.91 on 4/14/87. The ensuing years, from 1987 to 1993 failed to yield any clear meaningful buy signals. Prices gradually declined to lows of 1.25 in November 1992 and late January 1993. Please note that although the above historical description focused on buy signals, I have personally made the greatest profits by shorting gold futures and gold stocks from the time I started in 1985 (although that was not the original motivation for developing the system). My greatest gain occurred by shorting gold at $505 on 12/14/87 and pyramiding the profits by adding shorts on every UP closing week until gold fell to $365 on 5/22/89. I personally stopped "trading" gold futures at that time so as to make life less stressful. The system generated its first bull market 92-96 buy signal in more than a decade on 3/15/93 at 1.58. Prices rose without generating a clear sell signal to 2.86 before collapsing in a classic 2 up and 5 down run, with the 5 days down yielding a drop of 18%. That run was another buy signal, but was accompanied by a 92-96 day signal on the day of the bottom at 1.85 (also an exact, to the penny, 61.8% Fibonacci retracement of the rise). Therefore, as prices rose off of that bottom, the next 92-96 buy signal was XXed out. But prices continued to rise, eventually yielding the second historical Triple Sell signal. I point this time period out because the system yielded conflicting long-term signals. Of-course, starting in 1996 the golds began another major decline, with USERX falling to 2.94 after a reverse 10:1 split (an adjusted basis of 29.4 cents). That bottom, on 8/31/98, was marked perfectly by a classic 7 day down run with prices falling 18% in 7 days. The run analysis buy signal yielded a rise to 4.48 in approximately one month and then the third Triple Sell in history on 11/27/98. Prices declined for months (with the system staying bearish into the Bank of England announcement on 5/7/99) until a clear 16-20 bear market buy signal on 7/13/99 at USERX 3.11 (.311). The system sold on a rise to the next XXed 92-96 buy signal on 8/16/99 at 3.48 (missing the high by two days). The subsequent decline XXed out the next buy signal which immediately preceded the Washington Accord, whereby gold stocks rose 20% in one day into an XXed 92-96 buy signal (a shorting point). Note that there have been several times where XXed out buy signals have been wrong in the short-term, with prices rising after the signals. The XXing Out procedure was designed to avoid losses and therefore misses some rises. Overall, gold stock prices have declined 98% from their 1980s highs, with USERX falling from 10.53 to a current split-adjusted price of .286! Recent Developments The system remained bearish during the year 2000, generating only one 16-20 buy signal in January 2000 that lead to a brief rise in early February before hitting an XXed Out 35-39 shorting signal at the exact top on 2/10/00. Gold stock prices declined into what some analysts view as THE long-term bottom in late November 2000 with USERX at 2.31. The SKI indices and the run analyses did NOT confirm that bottom and did not generate a clear buy signal until the 35-39 index bought on 4/26/01 at USERX 2.68. Prices rose for 17 trading days (as was forecast at that time) into the system's clear sell signal on 5/21/2001 at 3.23 (which was an exact 38.2% Fibonacci retracement of the decline from the 1999 high of 4.71 and the year 2000 low of 2.31). At that time I posted that "prices should decline for months to years." Prices, of-course, did decline into short-term buy signals on 7/9/01 and 8/8/01. The system generated a 218-222 index buy signal on that low of 2.62 on 8/8/01. This index provided a very long-term (years) bull market signal that I remained skeptical of because it had been so many years since such a signal and because I had never seen this long-term index buy at a low. In any case, prices continued to rise, with multiple signals, until a clear sell signal was executed by the 92-96 index on 10/1/01 at 2.82. The subsequent decline into 10/22 and rise back into 11/7/01 generated a clear 16-20 index shorting signal leading to another low on 10/21/01 at 2.61. Now comes the interesting and problematical part. As prices rose again, the 92-96 index generated a bull market buy signal on the path at 2.81 on 12/6/01. The signal was XXed Out, but I was and am concerned that the XXing Out is wrong (a 15% probability). This was the first 92-96 buy signal on the path since 1994. Prices immediately dropped for 3 days but then rose into a 35-39 index buy signal (off the path since the system was on the 92-96 index). If the XXing Out is wrong, this 35-39 signal has always yielded immediate and large price increases. In the last nine trading days, however, prices have remained flat to lower. Unfortunately, I am still betwixt and between right now, but the system says that prices should decline again due to the XXing Out. The back prices are set up properly for a non-XXed 92-96 bull market buy signal on the path if we do get the proper decline over the next two weeks. I continue to be concerned that the XXing Out is wrong and that we will get a significant rise within the next 2 weeks. Stay tuned! We are still on that 218-222 index bull market buy signal from August, but never made a "normal" long-term bottom in the gold stocks (where the run analysis is supposed to give a buy signal one day off of the major bottom). Conclusion As a professor, a scientist, and a practicing clinical psychologist specializing in the prediction of human behavior, I hope to provide an interesting and useful analysis of gold stock movements. Remember that the system is not suited to predicting the movement of any one particular gold stock. My philosophical view on gold is that eventually the price will rise and that all fiat (non-gold based) monetary systems have eventually failed. However, I rely on my system to time such occurrences and to preserve my family's assets via gold and gold stock purchases. I provide these public analyses so as to increase my own investment discipline, to demonstrate that there is some predictability to the human behavior of buying and selling gold-related investments, and perhaps, to gain some notoriety. My almost daily posts will also provide you with some insights into my emotional reactions (i.e., fear and greed, doubt and overconfidence) that impede my own ability to follow the system. I do not have anything to sell and rely on my profession and my own trading for my income. I put 100% of my risk capital into gold stocks long or short because I try not to "invest" in anything that I cannot predict with a reasonable degree of accuracy. Remember, however, that emotions and money management strategies are at least as important as any prediction system for preserving and increasing one's financial assets.
I have followed SKI on the Kitco forum for a couple of years now. His system seems to work but I never quite understood how he came up with what he recommended. So I asked him to give us a background into the system and to allow us to post his recommendations. Remember guys, his system does not come with a money back guarantee, he doesn't make a dime off his suggestions. Life contains risk so act accordingly. Bob Moriarty 321gold Inc |
