Discussion of PRISM from the Fasttrack Listserv

[What follows is a long post to the Fasttrack listserv – in fact a combination of posts on PRISM.  This was posted by John Isensee, and the backtest was done by Sid Kaiser.  Sid used Metastock to do the backtest, and I do not know exactly how to interpret his inputs.  I think this is very informative, so I am putting it here on my site.  I have re-formatted and edited a bit, but I can claim absolutely no credit for the fine work done here. There is some contention as to whether or not Roy Merwin or someone else invented PRISM, but Roy led the discussions on the Fasttrack listserv. --Joel Williams]

Lee - here is the PRISM method directly from this bulletin board.

All the Best - John Isensee

"At 03:07 PM 3/21/99 -0500, Ken Close wrote:

Ken Read: can you say something more about the Merwin Method?  I have never heard of it.  How can I/we learn more about it?

Ken Close

Ken Close and Harris Schiller asked about the details of the Merwin Method which is also known as the PRISM Method.  This method was developed by Roy Merwin who was a regular and valuable contributor to this ListServ.  We haven't heard from Roy for several months.  I trust he is in good health!

The message that follows was previously posted on 11 Nov '98.

Here is the PRISM Method developed by Roy Merwin with a detailed quantitative analysis and evaluation by Sidney Kaiser.  Note in particular that Sidney's assessment is, "I feel the best use of this system/technique is as an early warning for a possible trade."

I trust this will yield heaps of money for all.

Best Regards, Ken Read Roy Merwin posted, in real time, accurate calls for the stock market corrections of Oct '97 and July '98.  Moreover, Richard Wells back tested the Merwin parameters and found that they were nearly 100% accurate going back as far as 1990.  According to Richard, the Merwin parameters predicted market declines as small as 1.5%.

There have been just a few detailed questions regarding Roy's postings about Exiting the Market, and these questions were quickly answered.  Nevertheless, a number of postings such as the posting earlier today by Harry Larson indicate that there are apparently quite a few people on this ListServ who did not fully understand Roy's work.  I confess that I was among those who did not appreciate the significance of his work until after I lost a lot of money.  Now that I have done my homework, I believe I understand just what Roy's contribution is and I trust his method will be just as successful in the future.

By piecing together a number of Roy's comments and running my own tests using Fast Track I have been converted into a Believer of the Merwin Exiting Method.  Let me forewarn the reader that this message may become lengthy, and it will be necessary to actually look at some Fast Track plots to fully grasp the significance of Roy's work.  If this message is unduly long and not needed by many of the readers, I apologize.  I hope it will be useful to many of you - it certainly helped clarify my own understanding of Roy's
work.

All of the quotes that follow are attributed to Roy.  The quotes are not in sequential order and they come from quite a few different postings.  I trust I am not using them out of context.  In the interest of brevity, I have not cited the specific references.

The basic premise of the Roy Merwin Method for Exiting the Market is,

You do not need 3rd party software to get adequate warning of a sell.  All you need is Fasttrack.

...let Fasttrack (FT) show you the indicators when the broad market is leading to an unfavorable correction (10-12 to 10-24-97).  Then you look at individual funds for your decisions.

I think you should exit the market when you see these unfavorable indicators in FT".

Roy is emphatic about what indicators are essential!

In real estate it is location, location, location. In mutual funds it is rank, rank, rank and then relative strength (RS), RS, RS and then momentum.

You should then look at RS compared with VFINX and RUT-I, and MACD and Stochastics.

When you put this all together with the PRISM screen data..., he contends we have enough data to exit the market.  An essential point of Roy's Method is that you do not rely on a single chart.  On the contrary, You need to look at the pattern of the 5 charts.

Roy's parameters for Moving Average Convergence/Divergence, Stochastics Chart, the Welles Wilder RSI, and the Days Between Points are the same as the Daytona Parameters (FT Monitor - summer '95) and the Donoghue Parameters (FT 6398.cmt of 11/1/95), viz.

MACD Long Average 26 Macd Short Average 12 Macd Trigger 9

RSI 14

Stochastic Range 14 Stochastic Smoothing 4 Days Between Points 1

Roy contends that "...relative strength is a main consideration". 

Roy's parameters for Relative Strength are distinctly different from values I have seen elsewhere.  Noting that the Relative Strength parameters are three times the AccuTrack parameters, enter AccuTrack Smoothing 17 AccuTrack Average 17 This input yields Relative Strength values of 51 and 51.  Apparently Roy is telling us that the Long Avg is not important, and we can keep it from distracting us by superimposing it on the Short Avg.

Roy's input parameters for the Price Chart moving averages are also distinctly short, viz.

Price Long Average 15 Price Short Average 5 At times he also used values of 10, 5 and even 15, 15. 

This fine tuning is not essential for an understanding and utilization of the Roy Merwin Method.  Roy kept experimenting with the values, while soliciting our input and comments.  The few indiduals who responded to Roy's requests for input were Richard Wells and Dean Bailey.  Dean also used values of 15, 5.

While the input will not show up on the essential 5 PRISM charts, it may be convenient to enter Roy's values for the V Chart, viz.

Moving Average 15 Moving Average Smoothing .0004

These values are of negligible difference when compared to the FT Manual Tutorial values of 30, .0005.  This sixth chart, the V Chart or Moving Average Chart, will be useful in a few instances where the PRISM charts might need a little more confirmation.

If you load all of the above parameters and save them as a Parameter Set you will be able to see the elegant simplicity and accuracy of the Roy Merwin Exiting Method. I gave his Parameter Set the name Merwin, which is how I will refer to it in the following comments.

Set the Fund to VFINX and the Index to RYURX.  Subsequently, it will be of interest to change both the Fund and the Index to see if the results are significantly different.

If you set the chart time span window to a year or more, or perhaps to just three months during a choppy or sideways moving market you will see an overwhelming abundance of Buy and Sell Ticks.  I found it useful to look at a window from May '96 up to the present.  Such charts simultaneously reveal the power and accuracy of Roy's Method as well as a point of confusion. Before focusing on the charts, remember that according to Roy

The warning signal to sell in Fasttrack was loud and clear provided that you ...look at the pattern of the 5 charts. If you focus unduly on the Price Chart you may very well miss the significance and simplicity of Roy's Method.

Before we focus in on the results of utilizing Roy's Method, let us first use a well known and well respected parameter set - the Daytona Parameters.  Over the same period of a little more than two years we see the Daytona Parameter Set indicates four corrections that are nominally in July '96, April '97, October '97, and July '98.  The Daytona Parameter Set requires just a few switches per year.  This appears to be quite desirable, until you focus in on the values and realize that the Sell Signals were sometimes uncomfortably late.

Going back to using the Merwin Parameter Set and setting the windows to a three month period centered about the corrections noted in the last paragraph yields the following results.  For July '98 we see a Sell Tick on the Price Chart and the confirmation on the Macd Chart with Stochastics crossing 50 and going down as well as rsI crossing 50 and going down, all at essentially the same day somewhere between 22 and 24 July.  The High was on or about 17 July.  Similarly, Roy's Method gave loud and clear  Sell Signals on 17 Oct '97 (the High was 7 Oct), and 5 July '96 (the High was 1 July).  The Sell Signal for 14 March '97 was not as clear as one might desire since the MACD came very close but did not quite cross the line.  However, if we look at the V Chart we can see that the Moving Average crossover confirmed the Sell Signal.

Whether we use RYURX or VMFXX as the Index makes little difference on the Sell Tick, although in some instances the crossovers are more sharply defined using RYURX as the Index.  Using VFINX or RUT-I makes no difference on the date of the Sell Tick.  Using OTC-C as the Fund yields a Sell Tick that is late, i.e.  28 July, and the Chart patterns are not as sharply defined as the other Funds.  The data seem to indicate that for most practical purposes, Roy is justified in claiming that It does not matter what I is.

There is little doubt that much of the 3rd party software that complements Fast Track is indeed excellent and there is ample documentation that the Sell Signals given by these products was quite good for the July Sell Signal.  Nevertheless, the data support Roy Merwins's premise that You do not need 3rd party software to get adequate warning of a sell. All you need is Fasttrack. Roy's Method may require a little more attention in analyzing the data, in order to get those loud and clear signals. Recent history fully supports Roy's statement that You simply have to analyze the Rank, Rel.  Strength, and momentum of the fund you own. That will tell you in no uncertain terms when to get out just like it told you when to get in. It is probably a good idea to check with some of the broad market indicators too.... The big question was raised by Roy himself when he pondered ...will this kind of history repeat. Who knows?

I have a few questions as well:

A. Herr Professor Merwin, did I misinterpret or misreprsent any of your outstanding lessons?  If I did err, don't hesitate to chastise me publicly. I further hope I did not overstep my bounds by answering the question that Harry asked of you.

B.  Did Roy's Method of Exiting the Market work for the specific funds that were held by you Fast Trackers just as well as it appears to work for some Index Funds?

C. Does any further analysis reveal that fine tuning of Roy's Parameter Set appears to be useful?

D. Does Roy's Method work just as well for Entering the Market? I am going to do my homework early this time.  If there any surprise, please don't keep them a secret.

In summary; thank you, thank you, thank you Roy! I just wish I had taken the time, last fall, to thoroughly assess your contribution.  I will try hard to make sure that I do not overlook your next gem.

If this message was much longer than was needed by most of the readers, let me again offer my apology.

From Sidney Kaiser we have the following:

I promised some further results on Roy's system, Here is an update.

I used Metastock to create and test a simplified version of the system.  This includes Stoch 14, RSI 14, 2 MA crossover, and MACD.  MS does not handle two security indicators like relative strength very easily so I left that one out.

The test was to evaluate Roy's technique as an exit signal.  I threw together an entry scheme to get long and let the exits fall where they may.

Merwin SIGNAL FORMULAS
---------------
Enter Long:

Cross(Mov(C, 5, E), Mov(C, 11, E)) AND Stoch(14, 4) > 50

Close Long:

Cross(Mov(C, 13, E), Mov(C, 3, E)) AND (Stoch(14, 4) 50.

Sell on a MA xover and either stoch or RSI

From Sidney Kaiser we have the following:

I promised some further results on Roy's system, Here is an update.

I used Metastock to create and test a simplified version of the system.  This includes Stoch 14, RSI 14, 2 MA crossover, and MACD.  MS does not handle two security indicators like relative strength very easily so I left that one out. The test was to evaluate Roy's technique as an exit signal.  I threw together an entry scheme to get long and let the exits fall where they may.

Merwin SIGNAL FORMULAS
---------------
Enter Long:

Cross(Mov(C, 5, E), Mov(C, 11, E)) AND Stoch(14, 4) > 50 Close Long:
Cross(Mov(C, 13, E), Mov(C, 3, E)) AND (Stoch(14, 4) 50.

Sell on a MA xover and either stoch or RSI < 56 or 44 respectively, AND a
standard MACD histogram crossing below zero sometime in the last 5 days.

Results: remember we are evaluating only the sells here.  Start with $10k.

There are the 7/23/98 and 10/16/97 sells Roy spoke of, along with a few others.

[Note1:  This involves trading in and out of VFINX, the Vanguard S&P 500 mutual fund.  The results for buy and hold over the same period are a profit of about 31,000, so there is an improvement over buy and hold.  In addition, one might have done well by using this as a timing system to trade in and out of good actively managed funds.  – Joel Williams]

[Note2:  Irt is a proff of the religious nature of the buy-and-hold types that when I posted this on the Motley Fool Mechanical Investing Board, one of the leading proponents of buy-and-hold scoffed that the difference was small - only averaging a few % per year.  Of course this was back when we were still in the bull market, or at least maybe he thought that.  Of course, if you went short during the sells, you might find that you did better than those few %.  Personally, I would take the few % per year.  There is a decrease in the risk, as well as an increase of the reward.  And being out of the market during the bear period (it is June 4, 2003 as I write this) has had great rewards.  – Joel Williams]
 
I am sure improvements could probably be made to this system if someone had the time to really dig into it....not me. I feel the best use of this
system/technique is as an early warning for a possible trade. There may be some errors here, but the results appear to be plausible.

For all you newbies out there, this kind of analysis is just the starting off point for evaluating a trading system.

Good Trading, Sid