What strategy actually works? - page 14

New comment
 
AxelQuant #:
Ideally, if the market is inherently dualistic, the strategy itself must be dualistic as well.

Exactly.

Somewhat late in my trading career, I actually find myself wanting to abandon the entire trending versus ranging distinction. I'm slowly coming to the realization that in the oldest instruments, all price action patterns have already occurred in the past to an 80% to 97% certainty. In this way, both trend and range are irrelevant. It's only the pattern projection, and the length of the "future run" therein, that matters at any given time.

(Chaos theorists will obviously despise this post so, by all means, let the diverse views fly).
 
Ryan L Johnson #:
all price action patterns have already occurred in the past to an 80% to 97% certainty.

Ryan, that's an intriguing point. It makes me curious: what kind of patterns are we talking about? Are these classic price action setups (like triangles or flags), or are you looking at something more mathematical, like specific volatility clusters or cycles?

It resonates with the idea that price action rarely invents anything truly new. While candlestick patterns might appear chaotic on the surface, the underlying mechanics of the movement remain remarkably consistent.

Take AUDNZD, for example:

  Fire on Water

If a market lives in a state of heavy mean reversion, it retains that return-to-average DNA even within a year-long trend. Market memory exists, the challenge is simply extracting it from the noise.

I'm currently finalizing an article on this subject, where I'll be sharing my indicator and the statistical findings behind it.

 
AxelQuant #:
[W]hat kind of patterns are we talking about? Are these classic price action setups (like triangles or flags), or are you looking at something more mathematical, like specific volatility clusters or cycles?

Code Base

Price prediction by Nearest Neighbor found by a weighted correlation coefficient

Vladimir, 2010.07.12 11:50

This indicator finds the nearest neighbor by using a weighted correlation coefficient, in which more recent prices have larger weights. The weight decays linearly from newer to older prices within a price pattern.
Note: To use this indicator to its maximum standalone potential, you should set your max bars in chart to unlimited in MT5. Of course, your pc must support that.
 
Ryan L Johnson #:
Interesting approach!
 
Ryan L Johnson #:
Price prediction by Nearest Neighbor found by a weighted correlation coefficient
We could add a calculation for the Root Mean Square Error between the current pattern and its scaled neighbor from the past. If the correlation is high but the error is large, it means the overall shapes match, but there are significant outliers or noise within the pattern.
 
AxelQuant #:
We could add a calculation for the Root Mean Square Error between the current pattern and its scaled neighbor from the past. If the correlation is high but the error is large, it means the overall shapes match, but there are significant outliers or noise within the pattern.

Sounds interesting. I seem to recall having read about that somewhere. I can see... your Article growing lengthier in the future.🔮

For my purposes, I'm content to work with the k-NN theory, the correlation coefficient, and the blue line study that overlays the applied historic pattern over open prices. There's also a straight symbol open price open price indicator posted in the NNWC indicator Discussion thread─as a pattern/price comparison tool for manual traders:

Forum on trading, automated trading systems and testing trading strategies

Indicators: Price prediction by Nearest Neighbor found by a weighted correlation coefficient

Ryan L Johnson, 2026.04.28 01:20

Here's a "stupid simple" bar open price indicator to go apples-to-apples, if you will, with the historic blue line.

 
Ryan L Johnson #:
I can see...
I feel like I'm never going to finish this. )) Thanks for the link, I'll definitely check it out.
 
AxelQuant #:
I feel like I'm never going to finish this.
On the one hand, that is humorous. On the other hand, I've experienced that too. I started writing an Article years ago and it spiraled out of control at less than halfway through. It was like a duel between the scope/subject of the Article and the targeted audience. One minute, I was saying to myself, "Inexperienced traders aren't going to have a clue what I'm talking about." The next minute... "This is going to be several dozen subjects in one Article." I eventually concluded that short form Articles simply ain't for me.
 
Exactly! Any research topic runs deep, and without context and nuance, the conclusions just don't land. I went a bit overboard myself and ended up with 48 pages of tables and charts. The moderator rightly pointed out that no one would ever finish reading it. So now I'm trying to kill my darlings. In the end, the only format that seems to work is: "Neural Networks Made Simple. Part 98" :)

What was the subject of the article you were planning to write?
 
AxelQuant #:
Exactly! Any research topic runs deep, and without context and nuance, the conclusions just don't land. I went a bit overboard myself and ended up with 48 pages of tables and charts. The moderator rightly pointed out that no one would ever finish reading it. So now I'm trying to kill my darlings. In the end, the only format that seems to work is: "Neural Networks Made Simple. Part 98" :)

What was the subject of the article you were planning to write?

Yeah, there are so many neural networks Articles, creating something new is probably a high bar to clear as well.

I was writing how to adapt OTC FX code to a centralized futures market. I guess that you could say that I ended up "self-moderating."

1...78910111213141516
New comment