Pattern Recognition -- Any Good?

Do patterns repeat?
Is it all random?

If you spend enough time in this thread (https://www.forexfactory.com/showthread.php?t=420159), I think you'll see that patterns DO repeat and it's NOT all random.
That thread & author are very controversial but if you focus on his material I think you'll see enough to agree that markets are more structured than it may appear to the untrained eye.

I didn't join his "commercial" group, just spent time in his thread & studied the charts. I didn't learn enough to trade his entire method but I definitely see certain patterns that I trade and have found others that I use after looking more closely at the "micro" structure of patterns.

Every trader is different so some will see absolutely NOTHING but randomness in a chart and the next person sees patterns that keep repeating.

***BTW, don't plan on spending 1 hour in the thread and understand what he's saying. It's a bit unorganized and takes some "connecting the dots" to even begin to see it. :)
 
Last edited:
I didn't read too much of the other threads but did extensive work on pattern recognition with several different approaches, and, at least from my results, didn't find anything that was worth spending time on.
I suspect that patterns may work subjectively during specific times on specific stocks, so generally they seem to work, but if you utilize the same pattern recognition/signal with stocks you don't know, they'll just fail more often than help.
Even if/when some patterns seem to have potential, they are at the bottom of usability, as comparing to practically anything else, meaning variety of indicators. I was actually getting better results by excluding certain patterns, but even that later turned out to be worth less than just not looking at patterns at all.
I'm still a bit intrigued with this topic every time I see someone tweeting about some pattern they're looking at, but just wasn't able to reproduce such results programmatically.
Of course everyone may have different approach and some people may be successful, especially in the bull market. So pattern recognition may work as an aid to someone's trading, but may not be as useful when fully automated.
Even when a pattern fails for any trader, they'll just say "oh, we've just entered an overall pullback or choppy market so no pattern would've worked". While during periods like the last 3 months every pattern will work and everybody will be running around excited about discovering and confirming every pattern they can think of.
At least that's my experience.
 
Last edited:
guru said:
Interesting, I was just listening to this new video / interview with an NYU professor of AI and Data Science, and he does mention pattern recognition a few times,
More...
It is highly likely that he is using the term in the sense of Fukunaga's classic text "Intro to Statistical Pattern Recognition," not in the sense used by posters here. Similarly, "IEEE Transactions on Pattern Recognition & Machine Intelligence" (bright yellow cover) contain a wealth of market-relevant information and ideas, but none of the papers, to my recollection, ever mentioned candlestick formations.
 
Just buy Thomas Bulkowski's book Encyclopedia of Chart Patterns. He actually, conducted detailed studies on the different chart patterns. There is such a thing as a failure rate. No single pattern works 100% of the time. A lot of them win 50% of the time or even less. What matters in the end is the size of your winning trades vs the size of your losing trades.
 
tommcginnis' Guide To Profitable Pattern Recognition

_______________________Stay away from long wicks.___________________




_____________________________the end.___________________________


:wtf:
:cool:
 
GuerrillaTrading said:
well, patterns are formed from repetition and I think it's wise to pay attention, and a break in the pattern can just as well be as powerful.
More...


If patterns worked than a break in a pattern shouldn’t be a good thing. Or if it would be a good thing then it would’ve been included in the pattern, so you’d still end up with a pattern. Basically you’ve got closer to proving that patterns don’t work, or showing how lunatics use patterns and then find excuses when patterns don’t work but something else works. This basically sums up all those discussions about patterns.
 
There are patterns in everything. Just look at your own life. You spend your entire day doing the same thing you've done the day before, the week before, the month before, and possibly for years. You get up at the same time, eat the same food, speak the same boring lingo, watch the same damn shows, then browse the same webpage like ET, then go to bed at about the same time. Tell me that ain't no pattern. Ggrrr :banghead: :finger:
 
You must log in or register to reply here.
Back
Top