An edge is positive expectation. As to your followup question, just test it the same way you would trade it: with multiple instruments. So you're multiplying a low trade count per year by the number of instruments you use. As to thousands of trades, you might downsize to hundreds of trades in your backtesting.
I think all patterns could be profitable.. but you have to know when to take them.. Wouldn't you agree?
Yes, I agree. For me, a trade setup is a pattern + context. The context tells me a) if there's enough "airspace" to a next level likely to be tested and, b) whether my maximum acceptable loss on the trade is feasible from the entry price.
but i'm still waiting for an answer from ET members,why patterns aren't' working anymore or at least-not as good as it was before? (US stocks)any thoughts? http://www.thecrosshairstrader.com/...he-failure-of-popular-stock-trading-patterns/
Who says they're not? Whether or not a pattern "succeeds" or not depends on how it's defined. For example, there was recently a lot of buzz about the "H&S" in the ES which amounted to nothing. But this morning there was a legitimate H&S in the NQ which worked out perfectly.
There are various types of pattern that exist as a visual on a chart. A dead cat bounce for example, is what I would refer to as a low grade category pattern, because of its poor parameter definition and characteristics. Mathematical patterns offer the highest quality in terms of structure and application.
It depends what kind of patterns you are talking about. Old chart patterns (h&S, wedges, triangles, candlesticks, etc.) do not work anymore because everyone can spot them and there are many free tools available for identifying them. Stock screener by Finviz for example. More importantly some or even most of those never worked well because the success rate was not known and it was actually terrible. There are also mathematical chart patterns where precise measurements of parameters are possible for any group of securities. Here is a tool that finds those and determines their statistical parameters. The analysis in that blog is representative of the amount of work one must do to increase chances of success. It is rather ludicrous to expect non-quantifiable visual inspections of charts to generate profitable setups unless one has many years of experience doing this type of thing and can rely on instinct.
Look further into the field of digital signal processing (Nyquist-Shannon sampling theorem), the fourier series is central to the original proof.
Nope, them triangles are way too obvious and everyone, I mean everyone, has at least one eye to see 'em with. (Those trend lines across longer-term swing highs and lows are pretty darn awful, too.)