|
TSC Technical Forum The Dow Nears Its Jump Zone By Gary B. Smith RealMoney.com Contributor 8/10/2004 8:25 AM EDT URL: http://www.thestreet.com/p/rmoney/techforumrm/10177294.html |
|
|||||||||||||||||||||||||||
Yesterday, I introduced Craig Schulenberg. Part trader, part researcher and -- if you read carefully -- part rocket scientist, Craig has devised a predictive system for the market and Nasdaq 100 Trust (QQQ:Amex) in particular. Today, the rest of his story and his current QQQ reading.
This MTS [Mechanical Trading System] was originally conceived as an end-goal in itself. However, the trading accuracies of each long portfolio average only about 73%. After trading these portfolios for a while, I turned the whole thing into a "preprocessor" and took a second step. Since it is arguably true the stock market possesses many features of a "chaotic system," I began looking at other kinds of chaotic systems. I soon realized that if I viewed the market as a sort of "ideal gas" I could use the data generated by my preprocessor to obtain a "market temperature." This is loosely akin to calculating the temperature of a gas by looking at molecular speeds that follow a Maxwell-Boltzmann distribution. I call this new quantity the Market Timing Indicator, or MTI. The calculation of the MTI concludes the second stage in my current procedure. Although there are additional stages of considerable interest, Stage 3 is the most important. This stage generates explicit buy and sell signals for QQQ, and for 143 other equities that correlate highly with QQQ performance. In addition, I have an entirely separate algorithm that generates overall buy/sell signals for domestic equity mutual funds. The principal output of Stage 3 is a Market Timing Report that shows the complete buy/sell history for QQQ, iShares: Nasdaq Biotech (IBB:Amex) , Semiconductor HOLDRs (SMH:Amex) and iShares: Goldman Sachs Network (IGN:Amex) for the past two years. The QQQ is currently in a "sell" state, and has been in this state continuously since May 26. Normally, the average duration for QQQ longs is about nine days, and for shorts it's about 12 days. The back-tested accuracy of the QQQ prediction is 100% for longs and 91% for shorts over the past 24 months. The annualized gain for longs is currently about 105%. A recent crucial activity was to add a model for inflation-corrected crude oil futures prices. This has been largely responsible for maintaining the "sell" state for QQQ for such a long period. Another interesting feature is a five-day "QQQ Weather Report," which tries to predict the near-term trend in QQQ by using weighted sums of five different neural networks, each tied to different time bases. Each neural network employs 11 independent variables that are formed from the most accurate indicators emerging from the preceding stages. Software development continues at an unabated pace. The availability of an accurate Market Timing Indicator provides many avenues for future research and development. With accuracy, all things are possible.GBS note: As of Monday, Craig was still indicating a sell on QQQ, but he noted in his QQQ "Weather Report" that things were turning positive and thus recommended covering any short positions and going flat. Today, we have the Dow, Office Depot (ODP:NYSE) , Indevus Pharmaceuticals (IDEV:Nasdaq) , Encore Capital (ECPG:Nasdaq) , XM Satellite Radio (XMSR:Nasdaq) and the T2108 chart.
Charts produced by TC2000, which is a registered trademark of Worden Brothers Inc.