It is ironic that taking short positions in a bear market can be highly risky.  One point to consider is the exuberance of the bear market rallies.  Most of the strongest upward days in stock market history have occurred during bear markets.  This was true during the Great Depression of 1932 and it is true today.  In fact, just last July we had two days in which the Dow Jones Industrial Average rose about 500 points -- only, as we have seen, to drop again to levels below the previous lows.   Here is a list that I have prepared of the 100 largest "up" days for the DJIA -- on a percentage basis.  The list reads from left to right: 
 

03/15/1933  10/06/1931  10/30/1929  06/22/1931  09/21/1932  10/21/1987  02/11/1932  09/05/1939  08/03/1932  11/14/1929  12/18/1931  05/06/1932  04/19/1933  10/08/1931  08/08/1932  06/10/1932  06/19/1933  01/06/1932  06/03/1931  10/14/1932

05/01/1933  07/24/2002  07/24/1933  10/07/1929  01/13/1932  11/10/1932  08/22/1932  10/20/1937  09/23/1931  10/20/1987  04/20/1933  07/03/1933  10/31/1929  04/10/1933  04/24/1933  11/04/1932  12/06/1932  10/11/1932  07/29/2002  08/15/1932

03/02/1932  06/24/1931  06/02/1932  10/04/1933  04/30/1931  11/15/1929  01/04/1933  08/17/1933  07/13/1932  12/17/1930  05/27/1970  11/07/1932  08/05/1932  10/20/1931  09/08/1998  10/29/1987  03/16/2000  08/17/1982  11/16/1933  01/08/1932

06/20/1938  11/10/1937  06/12/1940  10/09/1974  10/28/1997  05/29/1962  07/27/1932  05/10/1933  06/19/1930  02/09/1931  02/15/1932  10/01/2002  01/17/1991  01/15/1934  11/26/1963  10/01/1930  09/24/2001  11/01/1978  07/25/1932  11/07/1940

08/11/1931  02/16/1932  07/11/1932  11/03/1982  04/05/2001  06/29/1938  01/10/1938  09/02/1932  04/21/1932  10/11/2002  05/29/1933  10/15/1998  11/26/1937  07/14/1930  10/23/1957  10/06/1982  04/22/1980  04/01/1938  10/29/1974  09/11/1933

One might have thought that the largest "up" days would occur during bull markets.  But, the data shown above clearly is weighted toward bear market conditions.  Clearly today's big movement in the DJIA would fall right into the middle of this list. 

The reason for such volatility, I suppose, is American optimism and exuberance.  But, one stock market analyst I follow closely -- Richard Russell (see www.dowtheoryletters.com) -- describes the bear market as if it were a cunning creature, attempting by every means possible to seduce as many investors as it can to putting money into the market.   I have a hard time anthropomorphizing the market like this.  I do not believe it really has a mind of its own with such evil intentions.  Nevertheless, I respect Richard Russell as an analyst -- far older and more experienced than I -- who knows bear markets well.  He was forecasting this one many years ago -- at a time when many optimists thought the market could never drop. 

The S&P futures contract rose over 40 points today.  Even if I remove from consideration the OmniTrader "down" signal that I reported last night (see "apology" below) as a false start, I still must look at the neural network signal.  This is particularly true since, in fact, all 27 component systems of my "MegaSystem" were providing me with "down" signals.  That is very unusual, so it was quite an irony for me to wake up and learn about the enormous surge in overnight trading of the futures contract.  This morning the CNN commentators are attributing this to good earnings reports from GM and other companies.  But, undoubtedly, more is at play here. 

It should go without saying that the stock market is full of surprises.  Short-term market timing, the focus of this Letter, is one of the most difficult tasks imaginable for this very reason.  With all of the tools at my disposal -- and there are many more that I have to share in this letter -- there will always be a ceiling beyond which all forecasting efforts will fail.  The task at hand is to define that ceiling.  It seems to have fuzzy boundaries.  Offhand, I would say that it lies somewhere between a 70% and 80% hit rate.  However, I do not claim to be close to that level yet.  At the moment, I think my systems are about 60% accurate on a daily basis.  And, perhaps, even that is optimistic. 

Of course, a 60% hit rate can be profitable.  In fact, many traders would prefer a 30-40% hit rate.  The difference can simply be a matter of how tight the stop loss is set.  If losses are kept very small, one can expect more of them.  The strategy is to depend upon some big winners to more than balance out the many smaller losses. 

My practice has been to set stops at about 15 points, per day, when trading the S&P contract.  So, on a day like today, had I been in the futures market (which I was not, incidentally), my position would have been stopped out overnight.  Even if the market had dropped before the trading session ended, today would have been a loser for me.  However, 15 points per day is not what other would consider a "tight" stop. 

One of the best strategies I know of for setting stops is the MAE/MFE (i.e., mean adverse excursion and mean favorable excursion) developed by John Sweeney.  For information about this, see http://63.215.121.72/Web/TeHelp/usingTEMFEMAE.html or http://www.tradefutures.com/active.htm.  The system is simple and elegant.  I will provide more details about it in a future edition of this Letter. 

Another way to keep from getting badly clobbered when taking an end-of-day trading signal is to pay close attention to the intraday signals.  Had one been watching the market carefully last night, it would have been possible to reverse direction and thereby take advantage of today's strong movement.  One system I work with, for this purpose, is known as KWIK*POP.  See (www.e-tradingsolutions.com).  I will be describing this system more extensively in the future.  It is an extremely short-term trading system designed to take advantage of trading opportunities that last for about five minutes at a time.  For now, before I move on to another topic, let me at least tantalize you by showing you the KWIK*POP two minute chart of the S&P e-mini contract.  This is from today's session. 

Suffice it to say for now that the use of this system requires some skill at chart reading of real-time data.  The system is highly intuitive in that it does not issue precise buy and sell signals.  Instead, one must learn to combine many indicators that appear on the screen.  This chart was created using a special KWIK*POP template in a program called MetaStock.  I'll have more to say about both of these in future editions of this Letter. 

In yesterday's discussion about the Commitment of Traders Report -- as a leading indicator of stock market movement -- I did not provide you with details regarding the ratio of long to short positions.  The diagram I posted simply registered +1 or -1 for a long or short signal.  Then, I added as an afterthought that the gap is closing.  More specific information is presented in the chart below which shows the actual ratio of long over short positions among the large, commercial hedgers of the S&P 500 contract.  As you can see, since June 1 that ratio has moved from .74 to over .95. 
 
 

It is important to recognize one major reason why the stock market is so hard to predict: it is a learning system!  Once a trend or finding becomes recognized by a large number of participants, they then change their behavior to incorporate this new knowledge.  This may be precisely what is happening with the Commitment of Traders Report -- as the data that I presented yesterday has now become rather well known.  As a result, many people are now acting in anticipation and therefore, perhaps, sooner along the upward curve than had been the case in the past. 

I am a creature of impulse, and certainly the establishment of this daily newsletter and the expansion of this website -- for the first time in about three years -- was an act of impulse.  In my eagerness to post data yesterday about the Nirvana OmniTrader system, I overlooked the simple -- and obvious -- fact that the futures contract I was testing was being read by the system as a stock.  Of course, it dawned on me this morning that OmniTrader is not such a clumsy program as that.  The fact is that it uses different algorithms for futures contracts than it does for stocks.  It also has different algorithms that can be set for long, short and medium term trading -- and different trading styles within each of these.  Whether today's signal would have been long or short is, of course, dependent upon one's trading style.  So, today, I have redone the article on OmniTrader using the appropriate algorithms for futures trading.  And, in all fairness to OmniTrader's versatility and complexity, I have reposted the trading signals, which turned out to be up.  Yesterday's OmniTrader signal was, to put it bluntly, a false start. 

Jeffrey 

BioComp Profit Neural Network S&P 500 Futures Contract "MegaSystem" Forecast: 
Wednesday's close: down from Tuesday's close 
Thursday's close: down from Wednesday's close 

Nirvana OmniTrader Composite Technical Forecast for the S&P 500 Futures Contract: 
Short-term, aggresssive strategy: Wednesday's close up from Tuesday's close 

View the Results of Previous Forecasts 

Return to ForecastingSystems.com