S T R A T E G I S T   F O R E C A S T

=====================================



=========================================

History of Software/Report Changes

=========================================


This document summarizes all major changes made to the Schulenberg Strategist System(TM) and its various reports since November 10, 2004.


*** New (1/10/07): An enhancement has been made to the PORTSTAT "C" portfolios (Stocks - Conservative) and "E" portfolios (ETFs) to force more careful trading when the MTI level dips below approximately 1.45. Instead of selecting equities strictly on the basis of the Grail System signal, at these low MTI values the portfolio logic now also requires that any newly added equities also be current Long holdings of the Preprocessor. In other words, when it is a bit dicey to add any more "long" positions, the logic demands that a new equity have shown a respectable long-term trend as evidenced by the current Preprocessor holdings.


*** New (1/9/07): We have added a new indicator to our system: “The 95+ Club”. To compute this indicator we look at all of the 1000 Grail System short-term (S/T) timing models, select those with back-tested accuracies of 95.0% or greater (over a 4.5 year period), and then calculate the percentage of those that are in the 'buy' state. The idea is to use the most accurate models as a finer measurement of overall market strength than simply using the total percentage of models that are in the 'buy' state.


*** New (1/8/07):  We have now added 5 new ultra-conservative ETF-only portfolios to our PORTSTAT report.  This now gives us 50 portfolios that are subdivided into 5 different classes that progressively become 'safer': "H", "N", "C", "E" and "U".  The new portfolios (P6U-P15U) will hold from 6 to 15 ETFs, and they are restricted to those ETFs with significant volume (> $ 5 million in value traded per day), and with back-tested trading accuracies of 96% or greater, and back-tested drawdowns of -3.00% or less (over 4.5 years).  Even these 'play it safe' ETF-only portfolios can result in significant gains, however, despite the fact that they spend protracted periods in cash.  Take the P10U (a maximum of 10 ETFs). for example.  This is arguably one of the safest, and least greedy, of our portfolios. The back-tested results are as follows -- assuming that $100,000 was invested on 8/16/02.

On 12/31/02, value = $ 125,511.  Gain from 8/16/02-12/31/02 = 25.5%
On 12/31/03, value = $ 218,074.  Gain for 2003 =  73.7%
On 12/31/04, value = $ 295,082.  Gain for 2004 = 35.3%
On 12/31/05, value = $ 349,023,  Gain for 2005 = 18.2%
On 12/29/06, value = $ 403,096.  Gain for 2006 = 15.5%

*** Remember that these are 'back-tested' gains, and may not be indicative of future performance.  The maximum back-tested drawdown was just under -4.00%.

 

*** New (12/27/06): We have added one more ETF-only portfolio to our PORTSTAT report: P15E will hold a maximum of 15 ETFs, and can be expected to average about 1 buy or sell per day.

 

*** New (12/26/06): We are continuing to increase the accuracy of our 1,000 Grail System stock/ETF models, and thereby the accuracy of our 44 PORTSTAT portfolios. We will discuss this activity in more detail in a few days, and we will address the following topics:

1. Speeding up our software: Following data downloads (stocks and mutual funds data), stock renames, and split adjustments, it has been taking almost 6 hours to run our complete software suite. We have devoted considerable effort to making our software run faster – not only to permit preparation of our reports in a more timely manner, but also to allow us to more fully re-tune our models and portfolios after the inclusion of yet one more day's worth of market data.

2. Optimizing our choice of stocks and ETFs to model: Our Grail System consists of about 100 ETF models and 900 stock models, and these are now chosen in a 2-step process. First, our Preprocessor stage winnows our initial universe of about 9,000 stocks and ETFs into a 'selected' set of 1,600 equities. This selected set of equities is used by the Preprocessor to derive its portfolios (see NEWPICKS), generate the MTI (Market Timing/Temperature Indicator) value), and ultimately to produce intermediate-term (I/T) timing models. In the second stage, we utilize our new Prospector(TM) software to extract the best 1,000 equities for our Grail System short-term (S/T) modeling process. The objective is to maintain about 900 stock models (and 100 ETF models), and to periodically make minor additions and deletions so as to retain only stocks with back-tested modeling accuracies of 80% or greater (and with acceptable annualized gains and Average Maximum Drawdowns). Since we have plenty of stocks to choose from, there is no point in modeling any stocks that do not meet a set of high performance and accuracy standards.

 

*** New (12/18/06): About 60 ETFs have been added to our system, bringing the total to well over 100 in our Grail System. In order to take advantage of this, four new ETF-only portfolios have been added to PORTSTAT (P9E-P12E), now giving us a total of 44 portfolios based on our Grail System signals.

 

*** New (12/11/06). Modeling Accuracy. We have increased the fidelity of some of our Grail System signals by capitalizing on the extremely high back-tested accuracy of the IJS model. Although our principal model is the one for QQQQ, and this forms the basis for the preponderance of our Grail System stock/ETF models, it turns out that there are a number of stocks which can benefit from utilizing the trade signals generated by the IJS model (S&P 600 smallcap value) – as 'adjustments' to the underlying QQQQ signals. The average back-tested Long accuracy of our 1000 Grail System stock/ETF models is now 82.51% (over the past 4.5 years). During this period QQQQ has a Long accuracy (back-tested) of 97.14%, while IJS is our only model which boasts a (back-tested) 100% track record for 41 Long trades.

 

*** New (12/11/06). Bond ETFs -- we have added four bond ETFs to our Grail System, and their buy/sell signals can be seen in the SIGNALS report. These ETFs are TLT, LQD, IEF, and SHY. Note that these buy/sell signals are for informational purposes only; such ETFs may be fine for diversifying long-term portfolios, but offer insufficient annualized gain potential to be useful in our style of trading. We will probably add a few more of these in the near future. These ETFs were added to satisfy subscriber requests.

 

*** New (12/8/06). We have added a suite of 8 ETF-only portfolios to the PORTSTAT Report. These portfolios, named P1E-P8E, will handle a maximum of 1 ETF (P1E) up to a maximum of 8 ETFs (P8E). Since these portfolios follow our short-term (S/T) Grail System signals, they represent a logical step toward diversification over simply holding QQQQ Long. These portfolios do not 'short' – rather, they reduce their risk by dropping positions as the market weakens ... and refusing to add new positions unless the MTI level is > 1.45. The trading frequency of these new portfolios is, naturally, very low.


*** New (12/8/06). We have renamed the 'non-H' (normal) portfolios of PORTSTAT so that they use the suffix “N”. The portfolio names are now P1N, P2N, etc. This gives us the following 4 classes of portfolios, and in increasing order of safety and conservativeness (and steadily diminishing trading frequency) these are: “H”, “N”, “C”, and “E”.


*** New (12/8/06). We have modified the 'weighting factors' that we employ with the QQQQ day-trading 'Poker strategy'. Since the preponderance of our signals are relatively weak (?+,?-,*,S), we need to boost the allocations somewhat so that we are using 100% of available trading equity more frequently. We will thus use a 50% allocation for the '?+' and '?-' signals, a 100% allocation for the '*' and 'S' signals, and allocations in excess of 100% (i.e., using 'margin') for the much stronger **,***,****,SS,SSS,SSSS signals.


*** New (12/7/06). We have now switched from Microsoft Word (for our html editing) to the Star Office Suite by Sun Microsystems. This cures the html formatting problems that we encountered when we switched from Microsoft Outlook to Mozilla's Thunderbird. We are thus now using Firefox 2.0, Thunderbird 1.5, and Star Office as our primary tool suite.


*** New (12/5/06). Two new “C” (conservative) portfolios have been added to PORTSTAT. P20C is designed for a maximum of 20 stocks, while the P25C is designed for a maximum of 25 stocks.


*** New (11/14/06).  We have re-added two Reports to our daily report set.  The first, SECTION2.TXT, is a report that summarizes all of our intermediate-term (I/T) signals.  These signals have hitherto only been found in Section 2 of the Market Status Report (MKTSTATS).  Also, the reports VARPORTA.TXT, VARPORTB.TXT, VARPORTC.TXT, VARPORTD.TXT, VARPORTE.TXT, VARPORTF.TXT and VARPORTG.TXT contain intermediate-term portfolios constructed from these I/T signals.  The maximal sizes of these portfolios ranges from 6 stocks (VARPORTG) to 20 stocks (VARPORTA).  It is recommended that anyone interested in these portfolios paper-trade them for a while to gain familiarity with their performance, and their drawdowns.

 

*** New (11/14/06).  Two more VARPORT portfolios were added, bringing the total up to 7.  These intermediate-term portfolios range from 6 to 20 stocks in size.

 

Recent Changes to the MUTFUNDS (mutual fund) Report:

*** New (11/10/06). Many new funds have been added for AllianceBernstein and AIM Funds.

 

*** New (11/10/06). The Recommendation (REC) field for Short/Bear/Ursa/Inverse funds is now forced to a 'B' (Buy) when the QQQQ Grail System signal is a 'SELL',and to an 'S' (Sell) when the QQQQ Grail System signal is a 'buy'. This is currently our 'best guess' as to the appropriate time to purchase shorting funds.  Note that this is PROBABLY not the best signal to employ for non-tech Bear funds, e.g., funds that short GOLD, OIL, REAL ESTATE, or Japan. This is currently an area that we are trying to improve.

 

*** New (11/13/06). Shorting funds are now tracked according to the status of the QQQQ Grail System timing signal.  In other words, the performance (%gain) shown in Section 1 for each Fund Family assumes that the shorting (BEAR) fund was bought when the QQQQ signal was a 'sell', and sold when the QQQQ signal went back to a 'buy'.  Remember, however, that many shorting funds have rather recent inception dates, and so their Section 1 performance will not be indicative of their long-term performance.  All Shorting (BEAR) funds are now clearly marked with the tag '**BEAR**'.

 

*** New (11/13/06).  We have now reached a total of almost 1,100 funds.

 

*** New (11/3/06).  The current value of the Mutual Fund Timing Indicator (MFTI) is now shown at the beginning of the Report.  This indicator has the states: 'buy', 'hold', and 'SELL', and for each state the report gives a recommendation about actions that should be taken for mutual fund portfolios.

 

*** New (11/6/06).  SunAmerica Funds have been added.

*** New (11/7/06).  Many new funds have been added for GE Funds, Delaware, Vanguard, Fidelity, Oppenheimer,  BlackRock, Rydex and Profunds.

 

*** New (11/7/06).  The REC field is now shown for all "long" funds, but a trailing question mark (?) is applied if the correlation coefficient (R**2) is less than 0.9800.

 

*** New (11/8/06).  More Putnam Funds have been added.

*** New (11/9/06).  Many new funds have been added for American Century, Morgan Stanley, Janus, Dreyfus, T. Rowe Price, USAA, and Van Kampen. We have now reached a total of nearly 1000 mutual funds.

 

*** New (11/9/06).  Special error-checking logic has been added to detect mutual fund data errors and dividends.  Dividends are now fully accounted for in the historical data used to perform ETF 'decompositions'.

 

Recent Changes to the MUTFUNDS (mutual fund) Report:

*** New (11/2/06).  American Funds have been added.

*** New (11/3/06).  ING Funds have been added.

*** New (11/6/06).  SunAmerica Funds have been added.  Also, many more DELAWARE and GE Funds have been added.

 

*** New (11/3/06).  Section 3 has been eliminated.  Mutual fund recommendations are now contained strictly within Section 2, which shows the performance of each mutual fund (%gain) over the past 3 trading days.  The REC field has been added next to the SCORE field, and it will contain a S, H, B, or ? code depending on whether the  mutual fund is considered a 'sell', a 'hold', a 'buy', or 'unknown' (shorting/bear funds and funds with inaccurate 'decompositions').

 

*** New (11/3/06).  The current value of the Mutual Fund Timing Indicator (MFTI) is now shown at the beginning of the Report.  This indicator has the states: 'buy', 'hold', and 'SELL', and for each state the report gives a  recommendation about actions that should be taken for mutual fund portfolios.

 

*** New (10/29/06).  (MUTFUNDS Report) A ‘SCORE’ value has been added to each Fund that is shown in Section 2 (Performance over the past 3 Trading Days).  These SCORE values are based on the MUTCOMPS.TXT report (not part of the daily report set), and they represent the aggregate buy/hold/sell (intermediate-term) signals of the 5 different ETFs into which a Fund has been 'decomposed'.  By finding a weighted sum of 5 ETFs which approximates the recent price history of the mutual fund, and then generating an aggregate Score for the combination, we effectively arrive at a buy/hold/sell signal for the mutual fund as a whole.  The SCORE varies from 0 to 10, with a 0 representing a hard 'sell', a 5 representing an average 'hold' value, and a 10 representing a strong 'buy'.  

 

*** New (11/1/06).  SCOREs are now suppressed for all 'shorting' funds: Bear, Short, Ursa, and INVERSE.  Clearly, it is mathematical nonsense to try to find a set of 5 Long ETFs that will mirror the recent price history of shorted equities.  Also, SCOREs are suppressed for any mutual fund that had a very poor 'fit' (low R**2 correlation coefficient) in the ETF Decomposition  process.  If the fit is poor, then it is unlikely that the set of 5 ETFs comes close to the actual composition of the mutual fund, and in this case trying to  generate a 'score' is unjustified.

 

*** New (11/2/06).  We are adding more funds and fund families to our Mutual Fund Report (MUTFUNDS).  Today 15 American Funds have been added, and we will be adding more funds as our subscribers request them.  It is too early to tell how much (if any) the new mutual fund SCOREs will improve bottom-line performance, but they can't 'hurt' ... and this method probably represents the extreme limit as to what can be accomplished with mutual fund predictions.

 

*** New (10/6/06).  The weak '?+' and '?-' signals are now modified so that they correspond to the prevailing Grail System signal for QQQQ.  In other words, if QQQQ is in the 'buy' state, then a '?-' signal is automatically converted and shown as a '?+'.  Similarly, if QQQQ is in the 'sell' state, then a '?+' signal will be converted to a '?-'.  The theory is that in the case of very weak steering signals, the most likely outcome for the day is that QQQQ will move in the direction of its longer-term Grail System signal.

 

*** NEW (9/28/06): We have put our NEWPICKS report back into action, and it is now included with the other reports in our daily archive zip file.  Please read the report description carefully, and do your own technical analysis on the stocks listed before trading any of them.  Remember that many of these ‘picks’ are small-caps, or micro-caps, and may thus be thinly-traded (low volumes and larger spreads).  However, there are many high-performing ‘gems’ to be found here, and very profitable portfolios can be constructed using them.  It is HIGHLY IMPORTANT, however, that these ‘picks’ ONLY be bought “long” if our QQQQ signal (Grail System) is a ‘buy’.  During a ‘sell’ period you can expect these stocks to dive just as fast as they can soar.

 

*** New (7/3/06).  The 'BBBB' signal in Table II of our Prediction Table (our QQQQ Long/Short Day-Trading Signal) was extended to use a 4-character field.  The signals are now: ?, *, **, ***, ****, S, SS, SSS and SSSS, where more asterisks suggest greater 'Long' accuracy, and more S's suggest greater 'Short' accuracy.  We have modified Section C (below) accordingly.  This change affects the way we ‘call’ QQQQ each day, and definitely alters the way in which we ‘score’ our performance – there will now be days in which we make no ‘call’ at all.

 

*** Our recent QQQQ/SPY day-trading activity has achieved an 84.6% accuracy over the last 13 trading days (11 out of 13).  This has produced a net gain of 7.74%.  We are currently refining the format that we will use for our Tactical Plan – see Section B.

 

*** New (6/15/06). QQQQ Day Trading: The Poker Strategy.

We are introducing a new concept which will clarify the day-trading signals for QQQQ, and also give us a more concise way of expressing the ‘forecast’ for the next trading day.  See Section C (below) for details on this QQQQ strategy.  At this time we view this method as highly experimental, and we will be tracking its performance (and refining its rules and parameters) over the next 2 weeks.  Since our Market Timing Indicator (MTI) is primarily a QQQQ indicator, coming up with a highly specific QQQQ signal is the ‘best’ that we can do in the way of predicting overall market behavior – from the

tech’ standpoint at least.  Our MTI DOES NOT directly model the Dow or the S&P 500; it is primarily a QQQQ timing indicator.

 

*** New (6/8/06).  The Mutual Fund (MUTFUNDS) Report has been added back to our daily report set.  This information is readily extended by adding more funds and fund families at the request of subscribers.   In addition to mutual funds, this report can be used to improve the performance of variable annuities.  Heeding the Mutual Fund Timing Indicator (MFTI) is critical to achieving the best possible performance from your stock funds.

 

*** New (6/6/06).  The Prediction Table (PREDTABL) has been enhanced by the addition of a Table for DIA.  This Table, V-A, shows the expected values of H for DIA during the next trading day.  H (Height) is the midpoint of the body of the candle, and thus open prices substantially greater than H suggest that the price must inevitably fall.  Similarly, an Open price for DIA that is substantially below H suggests that the price is likely to rise during the day.

 

*** New (6/5/06):  We have enhanced the SIGNALS report by adding two more Sections.  Section II is the original report: all stocks grouped by Market Industry.  Section I is a summary of just the stocks with signal changes, and Section III is a summary of market sectors that shows the ‘hottest’ sectors.

 

*** New (6/1/06):  We have added a new Report to our daily archive file, and it is named SIGNALS.TXT.  There is a good description of the Report in its prolog, but basically this is a summary of our 1000 short-term (Grail System) signals. The summary is broken down by Market Industry, and the stocks and ETFs are listed alphabetically within their Industry. For  each equity we show the previous day’s recommendation (buy or sell) plus the current recommendation.  The format makes it  easy to quickly locate all stocks that are new ‘buys’ … or new ‘sells’.  Additionally, the PPF (Predicted Performance Factor) is shown which gives a good idea as to which stocks are higher in quality (high gains with lower risk and drawdown).  We invite comments on this Report so that we can quickly optimize it to satisfy our subscriber’s needs.

 

** New (4/27/06).  We now use a fixed 25% allocation for each of the 4 listed Portfolio Equivalent Equities (PEEs).  In addition, there are two alternates for each of these 4 principal stocks.  These alternates do not yield quite the same

performance as the first stock listed, but they are useful in the event that the primary stock has 'obviously' become a buy candidate -- rather than a shorting chandidate.

 

*** New (4/26/06).  The Portfolio Equivalent Equities Report (PORTCORR) now utilizes all Grail stocks (1000 of them) that are currently in a 'sell' state. Obviously, some of these will be much better shorting candidates than others, but the most important thing is to have a wide selection of stocks to choose from so that we can generate an 'equivalent' portfolio of 4 stocks.  In our next release (4/27/06) will be adding two alternate stocks for each of the primary ones so that the odds will be increased that suitable shorts can be found at any time during the trading day.  Often, a 'short' that works well in the morning may not be so good later in the day -- so the new report format will provide the means for selecting alternative 'shorts' that ARE working.

 

*** New (4/11/06).  Information has been added to Section D of this header material.  The concept of a Characteristic

Equity (CE) extends some of the ideas that underlie our new PORTCORR Report (Portfolio Equivalent Equities), and can be an invaluable aid in stabilizing a portfolio so that it doesn’t suffer significant losses during a dip.  These methods are only suitable for margin-enabled accounts.

 

*** New (4/3/06).  We have added a new report to the daily report archive.  This report is contained in the file PORTCORR.TXT, and the report is named the “Portfolio Equivalent Equity Report” … or the PEE Report.  Yes, a terrible acronym.  This report is still highly experimental, and we encourage our subscribers who are following one of our Hedging Portfolios (PORTSTAT.TXT) to paper-trade this information for a while.

 

*** New (03/24/06).  We have added a new indicator to Section 2: the Composite Neural Network Indicator (CNNI).  The CNNI had a value of 3.50% at Thursday’s close, but has advanced to 4.06% for Friday.  The CNNI is an average of 10 neural networks that each try to predict the net percentage change in QQQQ at various times in the future: 1 day (NN1), 2 days (NN2), 3 days (NN3), …, 10 days (NN10). 

 

*** New (3/21/06).  The FORECAST file (this output: what we send out daily in our email) has been changed from a text file (.txt) to an html file in order to improve formatting.  The daily archive file is thus FORECAST.HTM.

 

*** New (3/20/06): We have replaced about 40 stock models in our set of 1000 customized models.  What we did was to eliminate the 40 models with the following characteristics: (1) very low G/d (Annualized gain divided by drawdown) ratio, and (2) price < $10.00 *or* low trading volume.  The newly added stocks are better suited to possible inclusion in our Hedging Portfolio system (PORTSTAT), and thus give us a larger set of candidates from which to choose each day when the portfolios are updated.

 

*** New (3/16/06): 2 new Portfolios have been added: P4C and P15C.  This gives us a total of 30 portfolios, including the 6 "conservative" portfolios.

 

*** Hedging Portfolios (PORTSTAT)

*** New (3/7/06): 4 new Portfolios have been added: P6C, P8C, P10C and P12C. These portfolios are designed to have much higher trading accuracies, lower drawdowns, and lower trading frequencies.  They are specifically designed for trading in accounts in which trades must be kept to an absolute minimum, and as a result they do not offer annualized gains that are nearly as high as the standard portfolios.  The "C" suffix stands for "Conservative".

 

We have been in a QQQQ 'sell' since 2/21/06, and actually (as seen by back-testing) this 'sell' began at the end of January.  When any kind of portfolio is held during such periods, there will inevitably be drawdown.  Just because drawdown is inevitable, however, doesn't make it enjoyable.  There are a number of things that can be done to reduce drawdown to an absolute minimum:

   (1) Never use Margin when QQQQ is in a 'sell' state.  The leveraging effect of margin is more likely to hurt than help.  Instead, preserve the margin so that it can be used for a judicious "short" of QQQQ so as to provide a hedge against deep losses on the bad days.

   (2) In a non-marginable account, gradually increase or decrease the amount of money dedicated to each "slot" depending on whether we are in a 'buy' or a 'sell' state.  We are not trying to match the LONG% value suggested by the Hedging Vectors, but we are trying to minimize our "bets" when the House has the advantage ... and increase our stakes when we know that our changes of winning are greater.  The Hedging Portfolios (PORTSTAT) do this sort of thing automatically, but it never hurts to add even more insurance to this process.

   (3) Switch to a more conservative portfolio when QQQQ is in a 'sell' state, e.g., switch from P10H to P10 -- or P10C.

 

The 4 new "conservative" portfolios (P6C, P8C, P10C, P12C) should be a valuable addition to our system.  These portfolios are currently all in Cash, so they are not very exciting at the present time.  This is, in fact, both their virtue and their biggest disadvantage.  It is often the best policy to stay in cash at times when the market is weak.  At the same time, however, it is difficult to follow a portfolio that doesn't seem to get excited about buying anything for up to several weeks at a time.  We are currently switching investment funds to trade these new portfolios, and we will soon be officially tracking the progress of these investments.

 

*** New (3/1/06).  We have improved the adaptive response of our 24 Hedging Portfolios by adding logic to make them more conservative (1) when a new High-Water Mark has been reached, or (2) when the portfolio has been losing money for several days in a row.  Also, the Penalty Functions used to optimize the portfolio parameters have been modified to further reduce portfolio drawdown. We will not have completed our massive retuning procedure, however, until Monday ... and signal accuracies should be gradually increasing during this period.

 

*** New (2/27/06).  We have added two different formats of Transaction Summaries at the end of the Hedging Portfolio (PORTSTAT) report.  These summaries are intended to facilitate the automatic processing of transactions, e.g., via TradeStation(TM).

 

We have reduced the content of our daily Report archive file substantially. There are now just 3 files in our daily .ZIP download: MKTSTATS, PORTSTAT, and PREDTABL.  We are trying to consolidate our system so as to include only the most "profitable" reports, and also to facilitate the occasional transmittal of our reports via email attachments rather than downloads.  The following reports are still available if our subscribers request them: MUTFUNDS, NEWPICKS, MTLOGGER, VOTEQQQQ, SECTION2, VARPORTA/D, FORECAST.  Remember that ALL of our signals are summarized in the various sections of the MKTSTATS report.  We are considering adding the PREDTABL and PORTSTAT files as appendices to the MKTSTATS file, thus giving us a single file which contains 'everything'.  Furthermore, we will reduce the size of the "textual" prologs for each report section by simply providing links to html files on our web site.

 

*** New (02/23/06).  The reengineering of our 1000 custom models to employ A.M.D. (Average Maximum Drawdown) instead of a global Maximum Drawdown is still underway, and the retuning process will take a few more days.  Whenever drawdowns are quoted for our Grail System signals (our short-term signals), they should be understood to be Average Maximum Drawdowns, i.e., "typical" drawdowns.  For any given stock, and at any given time, a much larger drawdown can be expected to occur "occasionally".  Thus the A.M.D. for QQQQ is about -1.62%, whereas our true Maximum Drawdown (over the past 4 years) is -5.08% -- about 3 times greater.  The Portfolio drawdowns quoted in our Hedging Portfolio report (PORTSTAT), however, are true maximum drawdowns ... and are, of course, calculated by back-testing.

 

*** New (02/23/06).  Our "official" QQQQ signal has now been totally decoupled from our 1-day predictions contained in the Prediction Table (PREDTABL).  What this means is that our official QQQQ signal is now totally free to do its thing -- without any interference on our part, e.g., attempts to minimize the number of transitions, or to prevent exits that would represent "long" losses, etc. This now gives us 3 distinct signals for QQQQ based on the 'time frame' involved:

                          *** 3 QQQQ Signals ***

(1) Intermediate-Term Signal (see SECTION2 Report).  Period = weeks to months

(2) Grail System signal.  This is our "official" signal.  Period = 11.09 days

(3) Short-term signal (see Table II of the Prediction Table).  Period = 1 day

 

Signal (2) is our primary signal, and is thus the one which we utilize in generating QQQQ trading performance and accuracy.  Our (relatively) new Signal (3) is a 1-day signal and is contained in the new Prediction Table (PREDTABL). Remember that if you plan to try day trades for QQQQ using Signal (3) that you must examine Table II at the market open, and "use" the actual QQQQ Open price for the day to determine whether there is a QQQQ Long signal (* or **), or a QQQQ Short signal (S or SS), or a questionable signal (?).  The inherent accuracy of Signal (3) is, at best, 70-75% ... such is the difficulty of

predicting the market even for the course of a single day.

 

*** New (2/21/06).  Our Hedging Portfolios are rapidly converging to new solutions as our new AMD metric percolates throughout our System (see the next paragraph).  This has caused an "apparent" reduction in the annualized gains of the non-H portfolios.  This reduction is only "apparent", however. In fact, the new annualized gains are much more realistic ... and hence more indicative of what might actually be achievable.  A further improvement to the non-H portfolios has been made which moves the focus to lower-risk stocks. This has also reduced the annualized gains somewhat (since "safer" stocks can not be expected to perform quite as well as the riskier ones), but has had a beneficial effect in that the average number of Trades per Day has decreased. The major objective behind our recent work on the non-H portfolios (see Portfolios #13-24 of the PORTSTAT file) has been to maintain superb annualized gains while continually reducing the average number of required daily trades.

 

*** New (2/21/06).  We are in the process of retuning all of our customized stock/ETF models (1000 of them) to employ Penalty Functions that take account of Average Maximum Drawdown (AMD), rather than Maximum Drawdown itself.  It will take several more days before this conversion process is completed, and for all textual descriptions to be revised.

 

What we used before, Maximum Drawdown, was a very simple concept.  We simply looked at the largest drawdown that had EVER occurred for a stock during our entire optimization period of about 4 years.  This meant, however, that stocks like IMCLONE (IMCL) that suffered a 1-day massive drawdown (in the Martha Stewart time period) would forever bear the stigma of being a 'bad' stock. Worse, our models would tune themselves to hell and back trying somehow to become prescient enough to foresee (and thus avoid) such precipitous drawdown. We have now decided that the 'optimal' metric is the Average Maximum Drawdown. We calculate this as follows:  (1) we examine all Long periods for each stock or ETF, and (2) within the Long period we record the maximum drawdown that occurred during that particular Long period, and (3) finally, we average all of the maximum drawdowns together.  This gives us, in effect, a "typical" drawdown that can be expected to occur.  Of course, at any time a large drawdown of 2-3 times (or more) of this typical drawdown can be expected to occur.  This is why we like our Portfolios (PORTSTAT); they provide protection against such large, unexpected drawdowns.  The confirmation that this new AMD measurement (Average Maximum Drawdown) is a worthwhile idea was provided by the fact that the retuning has increased the annualized gains of our 1000 models by a significant amount -- and models are no longer trying to avoid (or predict) the massive 1-time drawdowns that have occurred in some of our high-performance stocks.  As a result of this we have more high-performance stocks that are now beginning to participate in our Hedging Portfolios.

 

*** New (2/8/06):

We have modified the strategy parameters for P1-P20 (PORTSTAT) to further reduce their average trading frequency.  These non-shorting Portfolios are specifically designed for retirement accounts, or for other situations in which trading frequency must be minimized (e.g., CASH accounts).  This reduces the annualized gain somewhat, but of what use are outsized gains if you can't make the trades?  The potential gains remain impressive.  This is an ongoing activity: to make the non-shorting portfolios more practical for CASH accounts.

 

*** New (2/6/06).  There have been 3 major innovations in the PORTSTAT Report:

  (1) There is now a Summary Section at the end of the Report which lists each Portfolio and shows the key performance statistics:  Maximum Drawdown%, Average Trading Accuracy, Average Number of Trades per Day, and the Annualized Gain.  Notice that the "H" (shorting) Portfolios demand a greater number of trades than the non-H Portfolios.

  (2) Our new Generalized Candlestick (GC) Neural Network is now used to narrow the choice when determining which stock to buy.  Stocks with a large negative shift in the H (Height) value for the candlestick pattern are rejected.  In addition, the "H" Portfolios utilize the GC Neural Network to perform "cautionary" sells -- despite the fact that the Grail System signal still shows a 'buy' for the stock.  Such "sells" are marked with the tag (GC) to show that they are cautionary sells.

  (3) We have enhanced our daily procedures so that the Timing Models for all stocks which appear in this Report are fully tuned before the Report is finalized.  This increases the amount of time necessary to generate our reports each evening, but the additional time is well worth it.  It is flatly impossible to tune all 1,000 models each evening, but we can certainly accomplish this for the 100 or so that are involved in this Report.

 

*** New (2/2/06):

We have made some improvements to the sector balancing logic in our VS Portfolios (VARPORTA/B/C/D) to ensure that they don't get so overloaded in any one sector, e.g., energy stocks.  We will soon be adding a 15-stock portfolio so that we will then have portfolios with 6, 9, 12, 15, and 20 stocks.  These portfolios utilize our intermediate-term signals, trade much less frequently, and are thus better suited to retirement accounts which do not allow frequent trading.  Of course, they also don't produce the kinds of gains that we see in our Hedging Portfolios (PORTSTAT).

 

*** New (2/1/06):

We have revamped the Hedging Portfolios report (PORTSTAT) so that it now contains a spectrum of 24 different portfolios.  These divide equally into two sets of 12:  P1H,P2H,P3H,P4H,P5H,P6H,P7H,P8H,P10H,P12H,P15H,P20H ... and P1, P2, P3, P4, P5, P6, P7, P8, P10, P12, P15, P20.  By our naming system a portfolio with the name PxH means that it can have a maximum of 'x' stocks, and the 'H' signifies that it will short QQQQ with some of the available money as Long positions are dropped.  The non-H portfolios (designed for retirement accounts and other non-marginable accounts) simply protect themselves by dropping

positions and going increasingly to cash. *** Notice that the portfolios have been reordered in the Report, so be sure to look for the specific portfolio "name" that you have been dealing with, e.g., P12H.

 

*** New (1/30/06):

The Hedging Portfolios (PORTSTAT.TXT) have just been significantly enhanced to further improve overall trading accuracy and reduce drawdown.  The technique used is to incorporate an "H" predictor (based on 5 years analysis of QQQQ, SMH, and SPY) which estimates the candlestick pattern Height (H) for the next day based on a neural network which analyzes the various patterns which have occurred during the past 11 days, and calculates the most likely candlestick pattern for the next day.  This allows stocks to be excluded from consideration that "might" have losing candlestick patterns on the day that they are bought. The average trading accuracy of these 16 portfolios is now approximately 75% (back-tested, of course).

 

*** New (1/25/06). We have added 4 more Hedging Portfolios (P15/P15H/P20/P20H) to our PORTSTAT report.  These are now the largest portfolios in this report. Remember that the PORTSTAT portfolios utilize our short-term (Grail System) signals ... unlike the VARPORTA/B/C/D portfolios which utilize our intermediate-term signals.

 

*** New (1/24/06). We have added 2 new Hedging Portfolios (P12 and P12H) to our PORTSTAT report.

 

*** New (1/23/06).  We have modified the logic for our Variable-Sized (VS) portfolios to reduce "future" over-concentration in energy stocks.  This does not, however, yet affect the current holdings -- which are very heavily into oil stocks.  In the future the holdings will be limited to no more than 50% energy stocks, although this may still be an excessively high number.

 

*** New (1/19/06).  We have modified the Variable-Sized (VS) Portfolios (see VARPORTA/B/C/D files) so that they add or subtract equity positions in response to fluctuations in the MTI value exactly like our Hedging Portfolio P10 (see PORTSTAT file).  This change should eliminate a lot of the churning that has occurred of late, and will allow a direct comparison of these two classes of portfolios.  Remember that the VARPORT portfolios utilize our intermediate-term signals, while the PORTSTAT portfolios utilize our short-term signals.

 

*** New (1/19/06).  We have retuned Hedging Portfolios P9 and P10 to improve their overall performance:  (1) restricted stocks on those with prices >= $ 10, (2) restricted stocks to those with better volume and dollar movement, and (3) restricted stocks to those with slightly higher trading accuracy (back-tested) and lower drawdown.

 

*** New (1/17/06).  We have now added Tables III-A and IV-A to our Prediction Table (PREDTABL.TXT).  These tables support SMH and SPY, respectively, and they give us the ammunition that we need to more accurately predict the MCS (Market Close Signature).  We will eventually add Tables III and IV as well (analogous to Table II for QQQQ).

 

*** (1/4/06-1/9/06):

Recent Signals, Drawdown, etc.

------------------------------

Despite the fact that our principal signal (QQQQ) has failed to "see" the recent market surge, our LONG% allocation level has seemed reasonable (prudent), and our Hedging Portfolios (PORTSTAT.TXT) have stayed almost 100% LOng and have consequently done very well indeed.  Note that P10 just sold SCSS and MNST, making a gain of 40.15% and 46.10%, respectively ... since the rally's start on 10/19/05.

 

Although drawdown is an inescapable part of investing, it is nonetheless always important for us to examine it closely to see if our timing model is perhaps overlooking some vital clue that might reduce such drawdown in the future.  As we mentioned yesterday, our new model (MB2006-1A) would "now" eliminate some of the recent drawdown as follows:

 

               Prev     New

 Date   QQQQ   Model    Model    Reason

------ ------  -----    -----   --------

123005  40.41  buy      buy

010306  41.31  buy      buy

010406  41.74  SELL --> buy     Incorporation of H Prediction (Table II-A)

010506  41.92  SELL --> buy     Incorporation of H Prediction (Table II-A)

010606  42.68  SELL --> buy     Incorporation of H Prediction (Table II-A)

010906  42.85  SELL --> buy     Incorporation of H Prediction (Table II-A)

011006  42.88  SELL     SELL

011106  43.21  SELL     SELL

 

The signal changes for 1/4/06-1/9/06 resulted from the use of the analytical approach shown in Table II-A of the Prediction Table (PREDTABL.TXT).  When the "consensus" prediction for H (mid-point of candlestick pattern for QQQQ) was fed into our QQQQ timing model, it shifted these 4 days of signals from a "sell" state to a "buy" state ... and without any real shifts in the previous 4 years worth of signals.  This suggests that this model enhancement is in a sense "proven".  Our new MB2006-1A QQQQ timing model now incorporates these new H predictions, and will thus deal with such anomalously strong market behavior in the future.

 

In any event, our earlier signal change to a "sell" on 1/4/06 translates into a 3.5% loss of gain ... or a 3.5% drawdown, if we had been foolish enough to short this market.  Since we had never "confirmed" this particular sell signal, it should be viewed as an unfortunate lost opportunity.  However, there are an awful lot of days and signals remaining in 2006, and it is far too early in the game to begin second-guessing our model.  Also, even if we had confirmed the "sell" and thus implicitly signaled a "short" for QQQQ, a drawdown of -3.5% is well below our back-tested maximum drawdown of -7.04%.  So, although it is nothing to celebrate, our model is (so far) performing within its specifications.

 

*** IMPORTANT NOTE (1/10/06):

We have taken the Prediction Table (PREDTABL) one step further with the introduction of the new Table II-A.  In turn, we have found that use of these new H predictions (generated by our subnetworks SN1-SN3) in our QQQQ model results in clarification of the recent 4-day "unconfirmed" sell.  We will publish more explanatory material about the new Table II-A (and how to use it), and our new QQQQ model MB2006-1A (which takes advantage of this data) over the next few days.  The fascinating thing is that the new numbers essentially ONLY alter the past 4 days worth of signals .... flipping the QQQQ "sell" signals to "buys" ... without altering the previous 4 years worth of history in any significant way.  What makes this model enhancement "feel" even more correct is that the back-tested 100% accuracy for Long signals is maintained, and the overall annualized gain figures improve slightly as well.  All of this suggests that our model enhancement (MB2006-1A) is legitimate ... and represents a fundamental improvement in our model.

 

However, resolving the past 4 days so that the "unconfirmed" sell for QQQQ reverts to a "buy" is now a moot point.  We now have a real, "hard" sell signal for QQQQ ... and this time we will not temporize by calling it "unconfirmed". We think that this 'sell' is real, and that it is going to cause some damage ... although we have been wrong before ...

 

*** New (12/30/05).  We have added a new feature to the Prediction Table (PREDTABL.TXT).  Table II shows an array of "open" prices for QQQQ, and tries to predict the "average" price that will occur on the next trading day for the open price that matches the actual QQQQ open price.  This Table is still experimental, and we will be focusing some major attention on it for the next few weeks.  The advantage of this approach is that we use all available price and volume data (plus the MTI value) for QQQQ, in addition to the "open" price, in an attempt to predict the H (HEIGHT) value for the day.  For reasons that will be apparent later when we release our new Tech Note ("Japanese Candlesticks and RaDiSH Numbers"), we have defined H (HEIGHT) as the "average" price for the day, i.e., (h+l)/2 where "h" is the High for the day and "l" is the Low. H is thus the center value of the candlestick "wick".  It turns out that H is a lot easier to estimate than the actual "close" price.  Once we have H, and the "open" price, it is then possible to "estimate" whether QQQQ will close "up" or not -- and what sort of percentage gain might be realized.  There are a lot of IFs involved here, but it is at least an additional step forward in our continuing endeavor to develop intra-day trading strategies to supplement our normal "swing" trading signals ... and "intermediate-term" signals.

 

*** New (12/12/05):

*** Note:  We will begin a follow-up on the idea of a Market Open "Signature" introduced in Tech Note 2005-7 (www.schulenberg.com/download/tn200507.txt). Each morning we will make a prediction as to the trend that QQQQ, SMH, and SPY will take during the day -- and then report on our "score" in the Forecast that evening.  What are our objectives?

   (1) Determining if we have ANY predictive power whatsoever; the odds are certainly stacked against us, and the only "edge" that we have is our unique Market Timing Indicator (MTI).

   (2) Determining when the "best" entry time is likely to be: at the market open, or later in the day.

   (3) Determining if "hedging" is advisable in the event that our portfolios begin to sag -- by shorting QQQQ or SMH with any remaining funds.

We will add our Prediction Table to our daily report set with the file name PREDTABL.TXT. This Table is currently based on about 40 months of market data (since 8/16/02), and will gradually change as new market data is incorporated.

 

*** New (12/5/05):  Following up on the advancement that made the MB2005-5B model possible, we are continuing to improve the overall accuracy of our short-term (Grail System) models.  In particular, note that SPY and IBB are now showing "buy" signals.

 

As of 12/2/05 we have switched to the new MB2005-5B timing model for QQQQ.  This new model now gives our QQQQ signal a back-tested Long trading accuracy of 100% for 29 long intervals during the past 40 months, and yields about a 94% Short trading accuracy for 29 short intervals during the same period.  The heightened accuracy of this new model is also reflected in our total model set of 940 short-term stock- and ETF-specific timing models (the Grail System).

 

As of 12/1/05 we have started a new investment experiment using a real ($) Ameritrade trading portfolio.  This experiment will track the actual gains of a P10 Hedging Portfolio (see PORTSTAT.TXT report), and the purpose is to (hopefully) demonstrate that a steady income of $ 4,000 per month can be produced from a $ 50,000 equity.  We shall see ...  The day-by-day performance of this new portfolio will be tracked in our usual PORT06HV.TXT report.  The name of this report is no longer applicable, but we are keeping it for historical purposes.

 

*** New (9/13/05):  The transition to our new signal system (STAR Report signals with corroboration from our older Grail System signals) is now complete.  The MKTSTATS (Market Status) Report now uses these new signals for QQQQ, IBB, SMH, and IGN, so that we now have a "buy"/"hold"/"sell" signal.  The MKTSTATS report remains our master report; it contains all of our stock (SECTION 2) and ETF signals (SECTION 3).  The PORTSTAT report (Hedging Portfolios) has likewise been enhanced to capitalize on the increased accuracy of the new signal system. Although this transition has been difficult to achieve (it is like changing piston rings on a running engine), we believe that we now have a much simpler, and more elegant system.

 

A key benefit is that the tendency for signals to "whipsaw" during choppy market conditions has been greatly reduced.  when our only choices were between "buy" and "sell" (and traditionally the "sell" signal has meant a license to "short"), it was inevitable that the signal would be forced to track the choppiness of the market.  With a 3-way signal, the "hold" signal can be used during these directionless intervals, and the "hold" appellation allows for lower MTI values to be attained before a real "sell" is triggered.

 

*** New (9/12/05):  The following reports are now obsolete: THEGRAIL.TXT, STARREPT.TXT, STARREPT.CSV, STRATEG3.TXT, and STRATEGY.TXT.  For the time being, however, the Grail Report (THEGRAIL.TXT) will continue to be included in the daily archive for the benefit of those investors who have been using its signals.

 

We now have only 5 principal Reports:

 

Report 1:  NEWPICKS.  The Preprocessor "PICKS" Report

Report 2:  MKTSTATS.  The Market Timing Report (our "master" report)

Report 3:  SECTION2.  The SECTION2 Report (formerly the STAR Report)

Report 4:  PORTSTAT.  The Hedging Portfolios Report (based on the Grail System)

Report 5:  MUTFUNDS.  The Mutual Fund Performance Report

 

The Grail System signals have now been integrated with our STAR Report signals to produce a new and more accurate buy/sell/hold signal for both stocks and ETFs.  For the time being the PORTSTAT Report continues to rely on the Grail signals, but this too will be changed in the near future as the STAR Report signal algorithms continue to percolate through our report system.

 

The Market Status Report (MKTSTATS) has been heavily revised (simplified) and now consists of just three Sections.  Section 1 is the same as before – a complete history of optimized signals for QQQQ, SMH, IBB, and IGN for the period from 8/16/02 to the present.  At the moment these are "still" the familiar Grail signals (with roughly a 12 trading day "long" retention period).

 

Section 2 of the Market Status Report now contains the essential signal information extracted from the SECTION2.TXT report (which is itself the logical successor to the STAR Report).  Section 2 shows all stocks and ETFs that are in a current BUY (B) state, or a HOLD (H) state during an ongoing "buy" period.

 

Section 3 of the Market Status Report is now a consolidated report on ETFs.  In this brand-new report format, ETFs are shown in 4 signal states:  BUY (or strong buy), HOLD, Guarded HOLD, and SELL.  The Guarded HOLD state is, of course, new. Please consult the Market Status Report for full details about our new ETF signal scheme.

 

*** New (9/8/05): Over 40 ETFs have now been added to the SECTION2 Report. Although these newly added ETFs (and signals) are in a "probationary" period, they should provide a higher level of accuracy than has been hitherto obtainable using our Grail System signals.  Also, the average retention period will be much longer for these signals ... good news for longer-term investors.

 

*** New (8/31/05): Enhancements to the new SECTION2 Report. On 8/31/05 the '?' column was modified to show whether the corresponding Grail System stock was currently generating a BUY ('b') or a SELL ('s') signal.  Thus, instead of simply showing a question mark to indicate that a stock was, in fact, a Grail System stock (and one that had been moved from Section 3 to Section 2 of the STAR Report), we show the actual signal.  This field should be used only as an "aid" in making up your mind between alternative investment candidates. Obviously, a 'b' (a Grail System "buy") or a '*' (non-Grail), is better than a stock marked with an 's' (a Grail System "sell").

 

The Grail System signals are sometimes followed by one or two exclamations (!). These are intended to provide a rough estimate of the reliability of the Grail buy/sell signal ... and of the investment suitability of the stock.  See the description of the 'G!!' column below (in Subreport 1).  Remember, however, that the Grail signals are generally NOT AS ACCURATE as the primary signals of SECTION2.  They may be useful for helping "finalize" a pick, but they should not be used as the final arbiters of whether to buy or sell a stock.

 

*** New (8/29/05). 327 new stocks were added to the SECTION2 report.  These stocks had formerly been found in Section 3 of the STAR Report (STARREPT), but since the buy/sell algorithm of Section 2 has been proven to be superior to that of Section 3, we have moved a large number of stocks out of Section 3 and into Section 2.  A new column (?) has been added to mark these stocks, and they are being viewed as in a "probationary" state.  Their past trading accuracy is currently below that of the Section 2 stocks, and this has dragged down our performance figures, but we feel that over the long term these newly added stocks should perform as well as our previous set.  We recommend that you do not use more than 1 or 2 of these newly added stocks (marked with a '?') in your portfolio until they have had time to prove themselves.

 

Also, two new columns were added to Subreport 1: "days" and "%Gain".

  The "date" field shows the total number of trading days that the stock has   been held "long" -- including tomorrow, since that is the next possible trade date.  This field may be very useful in selecting "recent" buys in the event that an insufficient number of brand-new buys is available.

 

  The "%Gain" field shows the total percentage gain/loss since the stock was purchased.  The base price is assumed to be the Open price on the morning following the generation of the "buy" signal.  The percentage gain is then calculated by using Today's closing price.

 

*** New (8/23/05).  Based on a just-completed study of the performance of our Hedging Vectors, we have added a new one: the Prudent Hedging Vector (P.H.V.). We currently view this as our best single estimate of the optimal allocation of funds between Long and Cash positions.

 

*** New (8/22/05). The IWM buy/sell signal was incorporated into our QQQQ model as a "corroborating" signal.  This has resulted in a new model, MB2005-4B, which has several nice characteristics: (1) it preserves our current "buy" interval, (2) it maintains a back-tested accuracy of 100.0% for QQQQ "long" trades over the past 36 months, and (3) it decreases the QQQQ trading frequency.

 

*** New (8/15/05).  The STAR Report was further enhanced by adding a new file (SECTION2.TXT) which documents the actual trades that underlie the spectacular gains made by the Section 2 stocks since 12/28/04.  The link to this report is: www.schulenberg.com/download/SECTION2.TXT

 

*** New (8/02/05).  The STAR Report was enhanced by adding a companion file which contains the Section 2 and Section 3 material organized as a .csv (comma-separated value) file.  This file is named STARREPT.CSV, and it may be loaded directly into a spreadsheet program such as Microsoft's Excel.

 

*** New (8/02/05).  Section 1 of the STAR Report (ETFs) was modified to improve the accuracy of the signals.  The star ratings now vary from 0 to 4 stars, with 0 or 1 being a "sell", 2 being a "hold", 3 being a "buy", and 4 being a "strong buy".  Modifications to Section 3 are still being investigated; on the other hand, we are quite pleased with the performance of the Section 2 stocks.

 

*** New (6/14/05).  A minor enhancement was made to Hedging Vector #0.  The new vector now takes full account of the recent history of the TOT parameter.

 

*** New (6/13/05).  A minor enhancement was made to the QQQQ model.  Version MB2005-4A now takes full account of the recent history of the TOT parameter.

 

*** New (6/03/05).  A significantly enhanced QQQQ model has been developed and put into action.  MB2005-4 boasts perfect (back-tested) 100% trading accuracies for both QQQQ "long" and "short" trades -- within 50 buy/sell intervals that stretch out over almost 34 months.

 

*** New (3/29/05).  Some of our "actual" performance figures are grossly incorrect due to the recent ticker change of LGND (to LGNDE).  This makes several of the Strategy 3 and 4 portfolios look very bad.  This problem should be fixed shortly, so that we can continue to accurately track actual results.

 

*** New (4/7/05).  We have terminated the running experiment described in the file PORT06HV.TXT.  This experiment, which was initiated on 2/4/05, was intended to test trading parameters for a "hedged" portfolio.  During this two month period we actually tried three different approaches:

 

   (1) Using L/10 stocks (NEWPICKS.TXT) for "longs" and QQQQ shorts for "hedging".

   (2) Using STARREPT Longs (Section 4) and QQQQ shorts for "hedging".

   (3) Using IWO (and then adding XOM) FOR "longs" and QQQQ shorts for "hedging".

 

On Friday (4/8/05) we stabilized the Portfolio by increasing the QQQQ "short". This resulted in a slight gain and left us being 5.80% down after 2 solid months of trading.  Our best result was a gain of 3.85% on 3/1/05, and our dip into negative territory was exacerbated by one bad day (3/16/05) when we made the considerable error of dropping our hedge altogether.  Had this not occurred, then we would be down about 3.50% as of Friday.  This would give us a total excursion of (+3.85%,-3.50%) for a 2 month period (a protracted "sell") in which margin was heavily used.  This is not bad, and is well within our theoretical limits for drawdown.  Nonetheless, it shows that it is almost an exercise in futility to try to make money (safely) during a down market.  Note that we have had an almost continuous "sell" period (for QQQQ at least) since 2/1/05 (see MKTSTATS.TXT and MTLOGGER.TXT).  This has been a very difficult test period, even for a hedging algorithm.  Note that we have not had a GREEN market color code since 1/21/05.

 

We will initiate a new test when QQQQ enters a "buy" period. We will also do a post-mortem on the results over the past 2 months to see how we would have fared if we had scrupulously followed Hedging Vector #2.  This Hedging Vector is tuned for IWM "long" and QQQQ "short".  It appears at first glance that we would have fared much better if we had simply followed this simple strategy "to the letter": buying IWM "long" and shorting QQQQ in accordance with Hedging Vector #2.  We will issue a report on this shortly.

 

*** NEW (3/16/05) ***

Improvements are being made to the Grail System to produce a slightly higher trading accuracy, and slightly lower drawdowns.

 

*** NEW (3/9/05) ***

Section 4 of the STAR Report has been improved by replacing the Average Volume column with a new one which shows the most recent percentage gain/loss in each stock's closing price.

 

*** NEW (3/8/05) ***

We have "further" revised the star rating system for Section 3 (STAR Report) stocks to address the fact that these stocks have been underperforming.  This revision to the rating system results in a lot of status changes in Section 3 and Section 4.  Section 3 stocks are especially tuned to provide optimal results when used in a "hedged" fashion, "a la" our 6HV Portfolio.  If you are NOT hedging, then Section 1 ETFs and Section 2 stocks will be safer investment candidates.

 

21 additional stocks have been added to our set of "select" stocks.  Twenty of these were necessary to ensure that all 100 of the NASDAQ 100 stocks are included.  Also, XOM was added since this major stock has been such a top performer of late.  We will be happy to add some additional stocks if we receive requests from subscribers.  Please let us know what you would like to see included.

 

We have added a new column to the STAR Report which indicates if a stock is a component of the NASDAQ 100, that is, part of QQQQ itself.  The Portfolio Rules discussed at the beginning of our tracking file (PORT06HV.TXT) have been augmented by two new guiding principles (7 and 8).  The upshot of this is that a reasonable percentage of our "long" positions should be kept in NASDAQ 100 stocks whenever a high degree of shorting is being performed.

 

*** NEW (3/7/05) ***

We have renamed our 6HV Hedging Portfolio tracking file to PORT06HV.TXT in order to be more descriptive; its former name was PF1L10HV.TXT.

 

We have revised the star rating system for Section 3 (STAR Report) stocks to address the fact that these stocks have been underperforming.  For the moment our 6HV portfolio is focusing on Section 2 stocks.  This revision to the rating system results in a lot of status changes in Section 3.

 

we have now installed a new QQQQ model (M2005-2) which has been optimizing for the past 3 weeks.  It confirms the current "sell" period, but gives a slightly more upbeat tone to the recent weeks.  In particular, this model suggests that mutual funds have never dipped below the "hold" state -- we have been compensating for this manually, but it's nice to have this corroborated by our Mutual Fund Timing Indicator.

 

*** NEW (3/3/05) ***

Tech Note 2005-5 was issued.  This suggests some "drills" that new investors might possibly consider, and lays some of the groundwork for the methods that we are employing in our Reference Portfolio (see PORT06HV.TXT).

 

*** NEW (3/1/05) ***

On 3/1/05 we added daily reports on the new 6-ETF Portfolio described in Tech Note 2005-4 (www.schulenberg.com/download/TN200504.TXT).  This is a Portfolio that is designed for non-margin accounts like retirement accounts, and strives for high-performance while minimizing drawdown and trading

frequency.

 

*** NEW (3/1/05) ***

Tech Note 2005-4 describes a new 6-ETF portfolio that is intended for non-margin accounts where asset protection is crucial -- and where trading frequency must be kept moderate.  We are now reporting the contents of this new Portfolio at the beginning of the Forecast Report (FORECAST.TXT) where the Long Allocation Vector (LAV) is presented.

 

*** NEW (2/28/05) ***

Section 4 of the STAR Report now shows the "change" in rating from the previous day.  This makes it easy, for example, to spot stocks that have just switched from the HOLD to the BUY state.

 

*** NEW (2/24/05) ***

We have now finalized our rules for Investment Strategy 6HV and made it "official".  See the daily file PORT06HV.TXT for complete details on this strategy.

 

*** NEW (2/22/05) ***

A useful new Section has been added to the Star Report: Section 4.  We will be using this in conjunction with our "real" Ameritrade portfolio, which will henceforth be a Strategy 6HV (Star Report with QQQQ Hedging).

 

In Section 4 we list only the "B" (BUY) stocks that were shown in Sections 2 and 3.  Moreover, only those stocks are shown that are relatively high-volume (and not too cheap), i.e., have a "Star Rating" that ends with the "1/2" flag. These stocks are sorted in order of decreasing PPF value so that the "best" stocks will tend to appear at the top of the list.

 

*** NEW (2/9/05)

As of 2/9/05 we have begun adding "FF" files to our download directory.  These are time-lagged archive files and require no password.  They are intended to provide potential subscribers with a way to familiarize themselves with the Strategist(TM) system, and to paper-trade various portfolio ideas before making a financial commitment.  The link to the latest available FF file is shown below.

 

*** NEW (2/4/05) -- an Ameritrade Portfolio with REAL dollars ***

A new reporting file has been added to our daily archive.  This new file, PORT06HV.TXT details the daily performance of an actual Ameritrade portfolio that is being traded strictly in accordance with the rules summarized in the header of the file.

 

This portfolio was initiated with $ 43,095.00 on Friday, 2/4/05.  It will be a faithful application of Investment Strategy 1 (Preprocessor portfolios), using the L/10 portfolio (up to 10 stocks "long"), and with QQQQ "shorting" in accordance with the daily Hedging Vector.  This will be a running experiment, and operational rules may be adjusted in the future.  In addition, we may eventually switch to one of the higher-performing Preprocessor portfolios -- for example L/7.  However, L/10 enjoys one of the highest back-tested trading accuracies (75.7%) of all Preprocessor portfolios, and has a back-tested drawdown of only -6%.

 

*** NEW (2/9/05) ***

A new section (Section 0) has been added to the beginning of the STAR Report (STARREPT) which shows the recent history of ticker changes and stock splits.

 

*** NEW (1/25/05) *** (Part 2)   Strategy 2HV: The new Hedging Vector is very similar to the one generated by the previous algorithm, but it responds more quickly to changing market conditions.  The accuracy of this Hedging Vector appears to suggest what we may consider a new "investment strategy": Strategy 2HV.

 

In this new strategy we use the Hedging Vector to construct a portfolio that is partly Long, partly Short, and partly in cash.  The "short" component will be generally QQQQ, although XLK is a perfectly good substitute (XLK has a smaller "float", however, so there is more of a spread ... and more difficulty

in getting quick fills).

 

The "LONG" component in this strategy will be comprised of (for example) the following ETFs: SPY, DIA, IWO, IWR, IWM, and MDY.  We are restricting our focus to ETFs that have large volumes, low drawdowns, and (historically) good annualized yields.

 

If we "back-test" the Hedging Vector using each of these ETFs, then we obtain the following results:

 

1.  (SPY Long, QQQQ Short, Cash):  Total Gain: 330.74%  Max Drawdown:  -7.94%

2.  (DIA Long, QQQQ Short, Cash):  Total Gain: 292.86%  Max Drawdown:  -7.94%

3.  (IWO Long, QQQQ Short, Cash):  Total Gain: 556.66%  Max Drawdown:  -7.94%

4.  (IWR Long, QQQQ Short, Cash):  Total Gain: 399.24%  Max Drawdown:  -7.94%

5.  (IWM Long, QQQQ Short, Cash):  Total Gain: 474.30%  Max Drawdown:  -7.97%

6.  (MDY Long, QQQQ Short, Cash):  Total Gain: 380.85%  Max Drawdown:  -7.94%

 

Total Gain is the total (back-tested) percentage gain that results when Strategy 2HV is executed from 8/16/02 to the present.

 

The Drawdown figures are primarily due to QQQQ "shorts".  The percentage gain can actually be boosted by an alternate strategy in which cash positions are "never" kept, but are rather used to increase the "short" component.  This, however, substantially increases the drawdown figures.  In the examples above the drawdown has been "capped" at -8.0% and this forces a varying percentage of investment funds to stay in the "cash" pile.

 

Strategy "2HV".  In this strategy we therefore examine the daily Hedging Vector (shown at the top of this Report and in the MKTSTATS.TXT report) and use it to establish 3 separate piles of equity.  The first "pile" will be used to invest LONG in one or more of the ETFs suggested above.  Our favorite is IWO, but IWM, IWR, and MDY make good additional choices.

 

The second "pile" of money will be used to short QQQQ (or XLK), again to the degree suggested by the Hedging Vector.

 

The third "pile" of money will be left in cash.

 

In actual day-to-day trading it shouldn't be necessary to buy and sell ETF positions if the change in the Hedging Vector is "small".  For example, changing from (30%,50%,20%) to (25%,50%,25%) is not really necessary.  The idea is to maintain an approximation of the Hedging Vector while not making an unnecessarily large number of trades.

 

*** Notice that despite the benefits of "hedging" drawdowns are still on the order of -8% -- this is an unavoidable side-effect of trading in the market.

 

Strategy 6HV:

-------------

Since the new Star Report (STARREPT) appears to be functioning very well, the natural extension to Strategy 6 would be to utilize the Hedging Vector to maintain a hedged portfolio here as well.  The idea is similar to that discussed above for Strategy 2HV.  Again, the "short" component in this strategy would be a QQQQ "short", and the "cash" component would of course be cash.  However, the LONG component would be formed from BUY/HOLD stocks in Sections 2 and 3 of the Star Report.  We would not use Section 1 (ETFs) because that would essentially duplicate Strategy 2HV.  Instead we would populate our LONG

part of the portfolio with the best stocks chosen from Sections 2 and 3. Whenever a stock went into the "sell" state then it would be replaced with a new "buy" stock -- provided that the approximate percentages dictated by the Hedging Vector are adhered to.  As in all investments using the Star Report, it is best to select high PPF (Predicted Performance Factor) stocks that are in the "buy" state.

 

We are currently investigating Strategy 6HV and will likely evolve this idea so that several "official" portfolios are generated -- and so that "actual" performance results can be tracked.

 

*** NEW (1/25/05) *** (Part 1)

After prolonged analysis of the market behaviour in early 2005 it was clear that the Hedging Vector and Market Color Code were "off the mark".  Clearly, the market has not been 100% Long (GREEN) for the past 3 weeks.  We have now discovered that a more accurate estimate of the current Hedging Vector can be obtained by working directly with Preprocessor outputs, rather than deducing this information from the "downstream" Hedging Portfolio generator.  The Hedging Portfolio generator works well, but there is too much "lag" due to the intermediate algorithmic layers, and this results in a less sensitive

prediction.

 

By using this new method to generate the Hedging Vector, our new QQQQ model now gives us the "sell" signals that have been apparent in actual market behavior over the past few weeks.  However, as of 1/21/05 even this revised model has switched to a "buy".  Insofar as market timing predictions can be trusted, we have to propose that the market is, in fact, "Long" as of 1/21/05.

 

Although the market may be "Long", it is undeniable that it is still very weak. One thing that the Hedging Vector makes clear is that buying or shorting QQQQ should be based on considerations above and beyond the simple "buy" or "sell" that is signaled in the Market Timing Report; the size of the "short" component of the Hedging Vector should be kept in mind in deciding whether to go Long on QQQQ, or to use QQQQ as a hedge ("shorted") and go Long with something else. We will discuss this more in following sections.

 

*** Note that this modification to the Hedging Vector only affects Strategy 2: Buying/Shorting QQQQ.  Strategies 1, 3, 4, 5, and 6 are all unaffected by this switch in the QQQQ signal.  Generally, these strategies already incorporate their own forms of "hedging".

 

*** NEW (1/24/05) ***

We have developed 2 improved QQQQ models that take advantage of some additional Preprocessor outputs.  The first of these is now in use, and it still shows that QQQQ is in a "buy" state at the present time.  This model also has carved out two new "sell" periods during the first two weeks of 2005, so this model appears to be a bit smarter about selloffs than our previous model.  It also meets our key criteria:

 

  (1) It must be logically defensible.

  (2) It must be mathematically elegant (simple, straightforward)

  (3) It must improve overall gains over our "entire" 29-month tuning interval.

  (4) It must not radically alter our past trading history.

  (5) It must use ONLY our existing signals and indicators.

 

Furthermore, this model was selected because it generates the same current signal ("buy") as our previous model.  The second model is similar, slightly more accurate, and in the testing stage.  Again, it appears to be better able to detect selloff signals.

 

*** NEW (1/12/05) ***

The tracking of "actual performance" for Strategy 2 (QQQQ, IBB, SMH, IGN) has been added to the summary report (STRATEGY.TXT).  The performance is shown for the 4 ETFs from 9/16/04 onwards, and both Long and Short figures are presented.  Also, a new file (MTLOGGER.TXT) has been added to the daily archive.  This logging file is a history of all key signals since 9/16/04 and is derived from previously published reports -- the data is unaltered and neither backtested nor optimized. This new file is used to calculate actual performance data for QQQQ, SMH, IBB, and IGN.

 

*** NEW (1/11/05) ***

The PPF (Predicted Performance Factor) has been improved, both in the Grail Report, and in the Star Report.  Morever, the PPF is now available in Section 2 of the Star Report (STARREPT).

 

*** NEW (1/10/05) ***

Section 6 (Actual Performance Figures for STARREPT) has been added to the STRATEGY summary report.  We are now tracking "actual" performance of trades made using the buy/sell signals shown in the Star Report.  This tracking was initiated on 12/28/04, and thus shows the situation from the most dire viewpoint since it reflects the losses caused by the selloff at the beginning of 2005.

 

*** NEW (1/5/05) ***

The new STARREPT ("star rating" report) is now ready for critical evaluation. We are currently developing "tracking" software which will monitor actual performance of the listed stocks when traded according to the new B (BUY), H (HOLD), and S (SELL) recommendations.

 

*** NEW (1/4/05) ***

The official Preprocessor portfolios (L/1-L/10) shown in the NEWPICKS report have now been made fully tradeable.  We now make 2 separate runs of the Preprocessor each evening.  The first run uses invariant parameters that are not subject to back-optimization (although newly optimized parameters may be introduced periodically so as to provide better "future" performance).  The second and final run uses fully optimized parameters and it is this run that is used to construct the Market Timing Indicator.  Since it is the first (non-optimized) run which produces the NEWPICKS report, the portfolios shown should portray a continuous and stable trading history.

 

In addition to decoupling the NEWPICKS report from the optimized Preprocessor run, "actual" performance figures have been added to the STRATEGY Report which will accurately track the real performance of L/1-L/10 from 1/3/05 onwards.

 

*** NEW (1/3/05) ***

(1) Today's set of reports includes the new prototype report STARREPT.  It is recommended that subscribers monitor this report for a week or two and paper-trade before entering into actual trades.  We are working on a "tracking system" for this report that will determine the optimal "star ratings" for buys and sells by extensive backtesting.

 

(2) We have switching to our new algorithmic system as of the 1/3/05 set of reports.  These new algorithms are in close agreement with our previous system (including the timing of the recent LONG period), but seem to run a bit "hotter".  You will notice that the percentage of Buy signals has jumped substantially using this new system.

 

(3) We are now maintaining our own database of about 8000 stocks and ETFs. This database includes all available "fundamental data" for each stock, in addition to the normal price and volume data.  We now perform our own "splits" and ticker changes on a daily basis.

 

*** NEW (11/20/04).  Our analysis of the recent 29-day steady rise in QQQ has been completed.  We have now developed and deployed a modified MTI algorithm which indeed confirms that the past 29 days have been a "buy" for QQQ.  This new algorithm meets our stringent criteria for making a change to our standard model:

 

  (1) It must be logically defensible.

  (2) It must be mathematically elegant (simple, straightforward)

  (3) It must improve overall gains over our "entire" 28-month tuning interval.

  (4) It must not radically alter our past trading history.

  (5) It must use ONLY our existing signals and indicators.

 

This algorithm capitalizes on the fact that our L/10 portfolio has consistently made money over the past 2 months.  This is confirmation that our Preprocessor still provides accurate signals for timing the market.

 

Using this new MTI algorithm our back-tested performance figures for QQQ are:

 

Annualized Gain   Average gain/     Maximum      Trading

   (Long)         trade (Long)      Drawdown     Accuracy

---------------   -------------     ---------   ---------

    94.95%          5.87%           -6.09%       100.0%

 

Annualized Gain   Average gain/     Maximum      Trading

   (Short)        trade (Short)     Drawdown     Accuracy

---------------   -------------     ---------   ---------

    58.33%          3.98%           -7.07%        92.59%

 

 

*** NEW (11/18/04) The non-hedging version of our "Master" Portfolio has just been added to our Hedging Portfolios report (PORTSTAT).  This is a 10-stock portfolio that ONLY invests "Long".  Like the P10H it utilizes all available stocks (Grail and non-Grail), thus investing in a spectrum of about 430 stocks and ETFs.  Like P10H, this portfolio is too new to allow us to suggest it as an actual trading model.

 

*** NEW (11/17/04): The "Master" Portfolio, P10H, has now been added to our Hedging Portfolios report (PORTSTAT).  This is a 10-stock portfolio that employs hedging by shorting QQQ when appropriate.  This is the portfolio that actually generates the Hedging vector, and the Market Color Code.  Unlike all of the other portfolios, however, it employs non-Grail stocks in addition to the Grail stocks.  This gives it almost 430 stocks and ETFs to utilize, and this gives us a more accurate picture of the hedging ratios.  However, this revised model is too new to allow us to suggest it as an actual trading model. If you are interested in this portfolio, then we suggest paper-trading it for a while.

 

*** NEW (11/17/04): The use of the non-Grail stocks in P10H has improved our Hedging Vector and Market Color Code.  The reason for this is that the portfolio generator now has a much larger collection of stocks from which to choose, and as a result doesn't have to switch to "cash" positions due to a lack of viable candidates.  This permits finer transitions from one state to another.  You can see in the Market Timing Report that this new algorithm gives a continuous series of GREEN color codes from 10/11/04 to 11/11/04 (an entire month).

 

*** NEW (11/16/04):  We have just added additional mutual funds for the Profunds and Rydex fund families.  Aside from bond and money market funds, all equity funds (including "bear" and "shorting" funds) are now represented.  As is discussed in Section 5 of the STRATEGY Report, our current recommendation is to balance mutual fund portfolios at least "monthly" (if possible), and to always select at least 4 funds -- picking those that have logged the highest gains over the previous 3 trading days.  The performance gains shown in the STRATEGY Report are "lower" than their true values due to the inclusion of "bear/shorting" funds, and to the inclusion of newer funds that have much shorter histories. We are now tracking a total of 411 mutual funds.

 

*** NEW (11/15/04): We have capitalized on the reliable performance logged by the L/10 preprocessor portfolio (see NEWPICKS report) to make an improvement to the (Market Timing Indicator) calculation.  This change gives us a better "fit" to the recent trend in QQQ, but more importantly, it increases our overall accuracy and performance for the entire 28 month period that forms our current backtesting interval.  We are still in a QQQ "sell" period, but now we have a couple of new short-term "buy" periods that are clearly visible in the Market Timing Report.  As a result of this change our current "sell"

would have been signaled on 11/9/04.

 

*** NEW:  We have COMPLETED Section III of the STRATEGY report.  Section IIIa now summarizes the "backtested" performance of our 24 Grail portfolios (since 8/16/02) and Section IIIb shows the "actual" performance of the same portfolios starting on 10/1/04.  The holdings for all portfolios are to be found in the new report STRATEG3.TXT, and this report also shows the current buy/sell signals for all stocks in all of these portfolios.  The Grail portfolios are intended to demonstrate the feasibility of constructing custom portfolios, and are not intended to be used for actual trades.  Please glance through these new sections of the STRATEGY report, and also the new STRATEG3 Report.  Also, there is more information on these 24 portfolios in Section 3 below.

 

*** NEW:  We have fleshed out Section III of the STRATEGY report to present the performance figures for our 135 Grail stocks in a meaningful manner. Please examine this new section, and also look at the brand-new report STRATEG3.  There is a lot more development to come in this area so please take a look at this material when you have time.  Also read the material about Strategy 3 in Section 3 below.

 

*** NEW:  The STRATEGY report has been augmented to include %drawdown and %GAIN and %LOSS statistics for strategies 1, 2, and 4.  The %GAIN figure shows the average "gain" for all trades that made a profit, while the %LOSS figure shows the average "loss" for all trades that caused a loss.  Note in particular the L/10 portfolio of strategy 1 (NEWPICKS report).  It is up over 10% since the 1st of October with a drawdown of -5.8%, a (back-tested) trading accuracy of 73.5%, %GAIN of 9.5%, and a %LOSS of -3.7%.  Although there is no guarantee that this level of performance will continue, it shows that the Preprocessor portfolios should be examined closely for possible trades.

 

*** NEW:  A "tracking" program has been developed to automatically keep track of "actual" performance data for the Hedging Portfolios (PORTSTAT.TXT).  The actual gains/losses for the 8 portfolios are now calculated since a starting date of 10/11/04, and using our previously published signal data.  As usual, we assume that all stocks can be bought or sold at the "open" price of the next trading day following signal generation.  The actual performance data is contained in Section IVb of the STRATEGY.TXT report.  The data for October only shows performance since 10/11/04.  The data for November is, of course, up through the date of the report.  The %DRAW (drawdown) figures are "actual" values and are based on end-of-day valuations of each portfolio. A new report (STRATEGY.TXT) has been created.  This report displays the performance statistics for each of our 5 major Investment Strategies.  At this point, Strategies 1, 2, 4, and 5 are treated in detail.  Work on the remaining strategy (Grail Report) will be accomplished over the next few days.