| |
The Power of Patterns and Using the Neural Net Selections in Trading: SCVL (Shoe Carnival) Part I – What the Nets Are Seeing
When I first decided to write this article, I thought that I could show an entire trade in one chart. When I began to think about it, I realized that the article would be way to busy to finish that way so let’s talk about what the nets look for and then analyze the set-up.
What do the nets do?
The nets are basically looking for Bullish Gartley (long trade) patterns that have a high probability of repeating, as has been demonstrated in the past price history of the stock. The nets also include a momentum component which measures how likely the stock is going to reverse and move up in price.
The nets refer to neural nets, which are nothing more than programs mimic the human brain’s ability to recognize patterns or relationships, but at much higher speeds via the use of computers. All these programs do are identify patterns that look a lot like Gartley patterns, or, at minimum, like major Fibonacci retracements, and then apply a heuristic (overall) assessment of pattern repetition based on the buy points that the nets see.
By the way, if you are doing your back-testing as Todd describes, you are creating a heuristic http://en.wikipedia.org/wiki/Heuristic Impress your significant other or friends with that term at a dinner party.
Does that mean it is some kind of “holy grail” methodology that is right every time? HARDLY! All these programs can really do is to optimize what patterns it sees over the course of years of daily data, and then decide if the patterns it sees measure up to a reasonable trading format from a long-only perspective. Even if a trade set up were 80% probable to be correct, that next one, just like with all trading systems or methodology could be WRONG. Even so, there are ways to manage the risk in the trades and that will be discussed in a later article.
The set-up for SCVL (Shoe Carnival):
Prior to the open on 11/29/2006, SCVL was screened for momentum, and the neural nets liked what they saw. The statistics for the long set up on the model, including the set up for 11/29/2006 was 1.89/1 58.1%. That is, the trades it saw on this model should (based on data it had not seen before but tested in prior history and then extended into untrained data) should provide $1.89 in profits for every dollar lost, with each trade 58.1% likely to be profitable.
Now, that is just what the model states for statistics with the overall set ups in the model which are based on roughly 5 years of end-of-day data. Real risk to reward in terms of what price action allows will be discussed in detail in an upcoming article.
But what did the nets really see? Let’s take a look.
Chart 1:
 |
| Momentum Reversals shown |
Chart 1 shows the momentum crossovers that the neural nets are looking for to define a pattern. It just so happens that the reversal shown on October 4 and the reversal that completed on November 28 separate what looks like an X-A-B-C pattern that ends up with the A-B-C-D leg at a 0.618 retracement of the X to A leg. Though it is not perfect, it is very close to a bullish Gartley pattern. See Chart 1A below:
 |
| How the nets see the pattern |
Take a look at Chart 3:
 |
| How the nets see the pattern with momentum added |
That is basically the set up.
Though Chart 4 is extremely busy, note how finding points of confluence show that the retracement of the previous down leg with the retracement of the X-A leg that started on October 4 shows a confluence around $27. The nets recognize this and will score that set up a little higher because of the price support. That is the reason this particular set up on Shoe Carnival (SCVL) was so well liked by the neural nets.
 |
| The price conflucence points |
What you have seen is how the nets use the power of patterns. The nets do it in much the way that your reticular activating system would find them. It just looks a little faster and harder than the average person. The next question is how do we use this information to trade this stock profitably?
That is the subject of the next article. Stay tuned.
|
|
|
|
|
Products
I believe you are probably the best out of the literal hundreds of traders that I have witnessed.
