April 2004

Trading Tip:

All of us have bad days in the market.  That’s a fact.  No one ever wins all the time.  What goes around comes around.  I think it is critically important to determine what happened when you do poorly.  It’s always easy to see what you did right when you have a good day.  So what is it when you do poorly?  Either you don’t feel well, felt lethargic and not “with the program”, or there are times you just can’t put your finger on what is wrong.  It is important to be as objective as possible in determining what went wrong.  This is not a witch hunt.  It’s an attempt to discover the errors you committed and install fixes for those errors with procedures and processes that will allow you to be successful.

So how do you do that?  First, never beat yourself up.  You are not a dummy etc.  Allow no negative self talk.  That’s hard to do when things are going badly.  The first thing you should do is stop trading real money as soon as you feel “out of control”.  Traders are bull headed animals usually and we are all trying to demonstrate discipline and that we are “with the system” and will follow it implicitly.  That philosophy sometimes manifests itself as bull headedness where you are losing trades and just will not stop trading and “following the system”.  You aren’t actually following the system because you are losing trades.  You are “out of control” when you lose two or three successive trades and can’t seem to figure out why, immediately stop, take a break, a walk around the block, or at a minimum go to paper trades and take the heat off your back.  Try to determine what it is that you are doing incorrectly.  Be objective, not personal.  Consider yourself an “expediter of the system”.  You are an extension of the system.  Your only function is to run the system correctly.  Make it a game.  Not a personal life or death mission.  Determine if there is something that is distracting you or limiting your ability to do that.  Also look at the methodology.  Are you really following the trading plan and the system’s signals?  Be brutally honest in this regard.  You cannot work with lies and deceptions to yourself.

A trader has four basic missions/responsibilities in trading.  First is to develop a good sound trading plan and relegate it to writing.  Make it easily understandable, as simple as possible, and such that it can be applied ruthlessly on a consistent basis.  In that plan, be sure you have clear, concise, and relevant entry and exit criteria.  What makes a valid entry signal and exit signal?  Exactly what are you looking for?  If you have four indicators, are you looking for three of the four to be firing for the signal to be valid?  Or do you want the stochastic to be turning over from the top or bottom of its range and coming through the 20% or 80% line, thus taking trades only when the stochastic is coming from an extreme condition?  Or take only trades when the Moving Average is in your favor.  That’s all up to you and how your system is designed.  The main thing is to have it all down in writing and be able to consistently follow it when trading.

Second, realize that your job as a trader/operator of the system is to recognize the signal that may be occurring and then validate the signal (does the signal meet your entry criteria as set forth in your trading plan?).  It is not your job to evaluate whether the signal will be successful or not.  Your only job at that time is to validate the signal.  Is it a valid signal in accordance with your trading plan criteria for entry/exit and whether or not the entry fits your risk/reward ratio are the only things with which you should be concerned.

Third, you are to execute the entry/exit in a timely manner.  Let’s say that last part again, “in a timely manner”.  That is critical.  You must have the faith, trust, and confidence to execute the signal that your system generates on time.  Late/hesitant entries are what kill a lot of traders.  Late = lose in most instances.  Early doesn’t guarantee success but it sure slants the odds in your favor.  If you chase trades, you are like the dog that used to chase cars.  Notice I said “who used to chase cars”?  I think you can see my point.

The fourth and last mission of the trader is to manage the trade in accordance with his/her trading plan.  Stay in the trade until your system tells you to exit.  Do not cut winners on pure emotional decisions.  The name of the game is to run winners and cut losers.  The trader who can control his/her losses is the one who will win the war.  You can lose many battles but you must run winners and cut losers to win the war.  It is therefore of some importance to develop and exit strategy that can be consistently applied.  That’s how you lock in profits and stay ahead of the power curve.  Have specific reasons to exit trades and follow them.  If you get out of a trade, there should be a concrete reason.  Look for significant places on the chart where an exit might occur.  Support and resistance lines, recent pivot points, yesterday’s high or low, etc.  These are places where things can happen on the chart.  When these areas are approached have a plan about how to exit.  Maybe a fast stochastic hooking or an Ergodic that is hooking along with the simultaneous building of a down bar.  Just have something that you look for when you near the important areas on the chart.  Consistency in exits is a big key to success.  Once you have an exit system perfected, do it the same way every time.

When the day is over and you do your homework (what I call an “After Action Review) and look at the day in hindsight (highly recommended) you will usually find that the system made money.  If you lost money, the bottom line is you lost money, and that is the criteria of success or failure in trading whether we like it or not.  If you have a good system, that consistently makes money, and you apply your systems, methods, and procedures correctly, you should also make money.  If you are successful in your playbacks (always replay each day), but unsuccessful in the real time environment then I suggest you evaluate your trading psychology.  The real time, real money pressure is changing the way you trade. Read Mark Douglas’ books The Disciplined Trader and Trading in the Zone.  Commit the principles found there into your memory and your actions.  You have to believe the things Mark talks about to implement them. Reading them is not enough.  You have to internalize them.  That is a step that is essential to getting over the psychological hump.  If you don’t live it you can’t do it.  I also strongly recommend you review the article written by Judy MacKeigan in the January 2004 Trading Tips newsletter.

I hope this helps with your trading.

Every time someone publishes an article about the benefits of a tool or study, a buzz of excitement is created in the investment community.   Many traders are eager to have access to the study, hopeful that it will make a favorable difference in their trading performance.   Just such a buzz was created by John Ehlers' article 'The Inverse Fisher Transform' published in the May 2004 issue of Technical Analysis of Stocks and Commodities magazine.  This trading tip will give a little more insight into the inverse Fisher transform as applied to the Relative Strength Index.

The inverse Fisher transform is :

y =  ( Exp(2 * x) - 1 ) / ( Exp(2 * x) + 1 )

where  x  is a value from the original study, and  y  is the transformed value to be plotted.  The following is a plot of the inverse Fisher transform.

The transform creates boundaries on the result so that it is in the range from -1 to 1.  Input values larger than 2 generate a result that is nearly 1, and input values less than -2 generate a result that is nearly -1.   This boundary characteristic can be put to good use when the Relative Strength Index is the input.

As seen in the plot of the transform, the input data needs to be in the range of approximately -5 to 5.  The RSI study data is in the range of 0 to 100, but this can be converted to a range of -5 to 5 using the following formula:

x = 0.1 * ( RSI value -50 )

The following table illustrates the conversion of RSI data to the output of the inverse Fisher transform:

The output of the inverse Fisher transform was normalized back to the range of 0 to 100 using this formula.   Normalized data is then easier to compare with the original RSI input data.

The table illustrates how original RSI values would be plotted when the inverse Fisher transform is performed on the data.  RSI values above 60 will be squeezed into the top 12 percent of the range and RSI values below 40 will be squeezed into the bottom 12 percent of the range.  The transform is causing the RSI transition from below 40 to above 60 to be plotted as a sharper swing from very low levels to very high levels.    Here is a comparison plot of an RSI in red and its inverse Fisher transform in blue.

The next technique that is typically performed is to do the transform on an average of the RSI instead of the raw RSI values.   Since the input will be smoother, the resulting plot of the transform is also smoother.   The following plot shows the original RSI, but the transform is being performed on a 9 period exponentially smoothed RSI.

Averaging the input data has created a smoother transform plot.   The turning points are still sharply peaked with a rapid transition from one extreme to the other.   The chart shows the last 2 hours of the trading day for April 23rd, 2004.  This is an example of a sideways market with a narrow range, which is very difficult to trade.  

Now, lets investigate the RSI and its inverse Fisher transform in a trending market.   The chart is still for April 23rd, 2004, but shows a period of time earlier in the day.

In this example, the inverse Fisher transform of the averaged RSI is showing smooth swings.  Let's see if we can identify some trading signals from this plot of the inverse Fisher transform.   Though there are an infinite number of possibilities, lets incorporate one of the characteristics of the Fisher transform.   Remember it will make a rapid transition from one extreme to the other.   Thus the crossing of the midpoint or 50% level will typically be a crisp crossing, as is demonstrated in the example shown above.

For the investigation, lets work with the following rules to define the buy, sell and exit signals.

Buy - When the Fisher crosses 50 going up.
Sell - When the Fisher crosses 50 going down.
Exit Longs - When the Fisher is above 80 and crosses below 80.
Exit Shorts - When the Fisher is below 20 and crosses above 20.

These rules are simple enough.   The following plot shows where the signals would occur.

The chart shows a 7 point move over a 3 hour period.   And the hope is that the sharper swings from the inverse Fisher transform will help us profitably pocket a good chunk of that move.  Here are the results of the 5 trades from our trading signals defined above and marked on the chart with thick purple lines.

Unfortunately, the results are not as favorable as the initial impression had been.  For the trading results, the execution price used was the Close of the bar that gives the signal.   No commission is factored into the results.  After 5 trades over a 3 hour period, in a fairly typical trend in the ES market, 2 small winners pay for 1 small loser.  One trade was a wash.  And the 5th trade represents the total profit for our exercise.   The Net for the 5 trades is + 2.25 points, before commissions and slippage are factored in.

A quick examination of the signals in the choppy market shown in the 2nd chart result in 13 losing trades in a row, and no winners.   The 13 losses are all small, but they do add up and the fact there are 13 trades without a winner is painful.   

In summary, our trading rules seemed so promising and the rapid transition of the Fisher plot from one extreme to the other seemed so clear and useful.  Yet, in a typical trending market our system had marginal results and in a choppy market the results were pathetically painful.  Some may argue the fault with our trading system was in the selection of the rules, that our entries are too late, or that our exits were either too soon or too late or based on the wrong criteria altogether.  Research on variations of the ideas presented in this article should generate better system results.

Finding a trading system based on the inverse Fisher transform of an averaged RSI was not the intent of the article.  John Ehlers article in Stocks and Commodities started out with this thesis, "How often have you been indecisive about entering or exiting a trade?  Here's one way to get a clear indication."  While the inverse Fisher transform may indeed make the signal clearer, it does NOTHING to make the signal better.  Pointing out that fact is the purpose of writing this Trading Tip article.  A signal that is late, or wrong, is still going to make for a bad trade.

Fisher's article stated that "using the inverse Fisher transform [will] alter the probability distribution function of your indicators."   I guess I misread what that meant upon first reading.  I had hoped it meant that it would improve on the probability of success in using a study for a trade signal.  Now that I have gone through the exercise of programming the Fisher transform into Ensign Windows for the RSI study, and experimented with this study to write this article, I think it is safer to make no claim that using a Fisher transform is going to improve study signals.

The Fisher transform is basically just stretching the middle of the RSI plot outward.  Compressing the top 40% of the RSI chart into a 12% band at the top, and compressing the bottom 40% of the RSI chart into a 12% band at the bottom, is the essence of what is happening.  This effect is shown by the values in the table for the RSI and its normalized transform.

In this article we used 80% and 20% as trigger levels for the transform signals.  These transform levels are achieved when the RSI input is around 57% and 43%.   Whether you use the inverse Fisher transform levels of 80% and 20%, or the original RSI input levels of 57% and 43%, it is the same signal trigger!  The Fisher transform does nothing to change the quality or probability of the signal or the success of your system rules.   Success in designing a trading system that incorporates the inverse Fisher transform of an averaged RSI will have to factor in the characteristics that have been pointed out in this Trading Tip article.

When an article is published, sometimes our desire for finding a 'holier grail' lets our expectations get carried away.  We want to attribute sacredness to a new study that we do not yet understand, and proceed with blind hope that somehow yet another mathematical massage of the underlying price data will give us that elusive trading advantage.  Having the inverse Fisher transform in your trading tool arsenal is not going to magically change your trading success.  It is just another level of mathematical crunching of the price data, but one worthy of additional research.