Mark Crisp's 1-2-3 system
Author: Defendi
Creation Date: 4/22/2013 8:03 PM
profile picture

Defendi

#1
Mark Crisp The system 123 is very, very profitable! Will give to the code in WealthLab? grateful

The explanations of the system are the link:

http://www.webtrading.com/downloads/123system.pdf
profile picture

Eugene

#2
While I don't mean this system, but in a sales pitch about any system in general, have you ever noticed that it's NOT being described as "very, very profitable"? ;)

What the exaplanations don't tell you is how exactly to define a "bottom" (or "top") mechanically, and how deep a valid "correction" may go without invalidating this ZigZag setup. There are many ways to program this strategy ranging by complexity. From our downloadable strategies, the closest options available right now are:

* The ActiveTrader 2009-01 | Double-RePo system. It's based on a similar principle: strong move up, reaction, secondary move up - short when the price breaks down.

* Wyckoff spring setup / Wyckoff Upthrust pattern. A percentage-based pattern: bottom (top), rally (reaction), second low (high).

Hope that hepls.
profile picture

Defendi

#3
The basic idea is as follows:

When I have 2 locks below I have the first movement, while there lowest closing movement is one
When you do a lock up ... starts to move 2 .... lasting as long as there highest closing
Shutting down starts the motion 3

Fund 3 can not be less than 1 background
Breaking 2 - starts buying

Target is the distance from 1 to 2 designed up .... there are those who design in 161% up
Stop background 3

As the hit rate is around 60 to 70% and as the stop is always less than the gain, the system becomes very profitable.
Many traders in Brazil operate this system, but would like to test it in WealthLab

.PS: works well for 60 minutes daily and weekly
profile picture

Eugene

#4
That Google Translate output sometimes gets really odd. The "locks" and "lowest closing movement", and how "fund can not be less than background", are quite amusing!

Nevertheless, my understanding of this kind of swing- and zigzag-based systems tell that the long setup might be coded as follows: a new Lowest Low (1) followed by a Highest High (2), then an intermediate new Low (3). Breaking above the High of (2). Or vice versa for shorts.

There are many ways to skin a cat, i.e. to program this system: using alternate CrossOvers/CrossUnders, or zigzags (Peak/Trough), or swings, or highest highs and lowest lows as above. The Wyckoff Spring/Upthrust strategies are undoubtedly the closest to the setup from your PDF book. I'd expect that you have a careful look at the strategies above, as the underlying principle is the same.
profile picture

Eugene

#5
With a minor modification to exits, here's Crisp's 1-2-3 System Short:

CODE:
Please log in to see this code.
profile picture

Eugene

#6
And a companion Crisp's 1-2-3 System Long:

CODE:
Please log in to see this code.


Feel free to tweak the code to code a different exit, for example. But the entry detection is working precisely as described in the PDF. You can also define proximity between swings #1 and #3.
profile picture

Cone

#7
I just spent a good amount of time on a flight looking at this. The PDF isn't very precise, and I haven't run Eugene's code yet, but to program this pattern for backtesting (without peak/trough constraints) is definitely a lot more complex than it appears at first glance.

For example, the pattern can contain any number of bars, so a lookback period can't be used. You essentially have to peek ahead and store the bar on which a pattern is identified and then make sure your code doesn't make use of an identified pattern before that bar. I'm interested in this one for one of our TSL articles in ActiveTrader magazine, so I'll take this one on.
profile picture

Defendi

#8
My ignorance is absurd in English ... sorry

I appreciate the willingness of you, do these codes should be too much work

Whenever the bottom one is broken by the bottom 3, the system is canceled and a new count is established

And a new countdown starts own ....... the bottom 3 will turn a new fund 1

I believe this is the problem that the system is making so few trades, and in fact, he gives many signs

thank you very much
profile picture

Eugene

#9
Ricardo, please copy/paste the code for both strategies above as I made an important edit.

Robert, the version above assumes that the pattern can contain any number of bars within reasonable proximity set by user. Looking forward to seeing your rendition, although I don't quite understand your point re: not using an identified pattern.

Here's simplified 3-bar version just for lulz:

CODE:
Please log in to see this code.
profile picture

Cone

#10
The peak/trough solution with one series is certainly a valid pattern, but it's not what Mark Crisp presented - even though it actually does identify many of the proper patterns w.r.t. high/low. In other cases, it gives false positives w.r.t. high/low. The 3-bar pattern code looks good though.

Re: not using an identified pattern.
There are two ways to do it. Identifying a pattern in the future is probably the hard way. I usually pick the hard way at first. Now I think it's better to search for the pattern in reverse, i.e., find 3-2-1, looking back from the current bar. I think that's the tack I'll take now.
profile picture

Defendi

#11
I could not make it work

Anyway thanks