TA doesn't work period.
Does TA work better or worse on lower or higher time frames?
- Thread starter iamnewuser911
- Start date
TA doesn't work period.
Some claim that, but only because they don't understand how to do it correctly.
I would advocate a top-down approach.
I'm pretty demanding of my trading platform (once a tick-scalper, always a tick-scalper??) And so,
"TA" is nothing more than running some moving averages over market action (i.e., priceXvolume). There is nothing magic there. AND it needs to earn its keep (its space on my screen)...
How useful that [price;volume] information is diminishes with each tick of the clock. However, with each tick of the clock, *more* information is added! So, there is a battle between how useful the information is, and how much of it you wish to gather.
But "TA" is fundamentally a gathering of sample data from a phenomenon *known* for both an inherit Weiner Process variability, AND a recognizable predisposition (let's call it, "trend") that violates I.I.D. (Independent Identically-Distributed) Gaussian symmetry. (We just described a "bell-shaped curve" that has a bias up or down [or both!]...)
The upshot is, "how useful" TA is depends on how stable the underlying data are. "How stable" is a relative term itself, and begs historical reference (and for trading, "historical" may be the last 90 minutes, or the last 40 years). So, "how useful" and "stable data" thoughts command you to always inspect your TA (because you're NOT going to affect any change on the overall market), and *tune* your TA to the point where it nicely captures the market you see. (This means, adjusting those look-back parameters handed out as popularly *fixed* by every software platform I know of.)
And, when your TA set-ups no longer seem to work as well? Push those parameters around again, and see what works in the new(er) market regime. FWIW, my TA has not changed in many years, although I review/inspect its fit.... pretty much daily.
I'm pretty demanding of my trading platform (once a tick-scalper, always a tick-scalper??) And so,
"TA" is nothing more than running some moving averages over market action (i.e., priceXvolume). There is nothing magic there. AND it needs to earn its keep (its space on my screen)...
How useful that [price;volume] information is diminishes with each tick of the clock. However, with each tick of the clock, *more* information is added! So, there is a battle between how useful the information is, and how much of it you wish to gather.
But "TA" is fundamentally a gathering of sample data from a phenomenon *known* for both an inherit Weiner Process variability, AND a recognizable predisposition (let's call it, "trend") that violates I.I.D. (Independent Identically-Distributed) Gaussian symmetry. (We just described a "bell-shaped curve" that has a bias up or down [or both!]...)
The upshot is, "how useful" TA is depends on how stable the underlying data are. "How stable" is a relative term itself, and begs historical reference (and for trading, "historical" may be the last 90 minutes, or the last 40 years). So, "how useful" and "stable data" thoughts command you to always inspect your TA (because you're NOT going to affect any change on the overall market), and *tune* your TA to the point where it nicely captures the market you see. (This means, adjusting those look-back parameters handed out as popularly *fixed* by every software platform I know of.)
And, when your TA set-ups no longer seem to work as well? Push those parameters around again, and see what works in the new(er) market regime. FWIW, my TA has not changed in many years, although I review/inspect its fit.... pretty much daily.
Last edited:
Geez Tomm... as much as I respect your posts... you apparetnly don't get it either. Yes, it's not "magic". But it's also much more than "running some MAs over the market". (And in my trading, I don't consider volume at all.)
Geez, Tom ... as much as I respect your posts (and that's a lot) ... I'm going to dissent strongly from your opinion, there: all my own trading (both earlier on in my career, as an independent trader, and now as an institutional one) is strongly related to support and resistance - however local - and that's certainly TA, and it's equally certainly a whole lot more than "running averages over market action" (which I don't do at all).
I'll agree with that part.
Okay (he said, picking up the gauntlet..
....), TA itself IS nothing more than moving averages (or 0.0↔1.0 normalizations). "That's a fact." I mean, it's algebra, fergawdsakes. And I'm sure you'd agree with that. (With a pint and a moment to ruminate.)But if you're speaking of the messages it provides -- MACD histograms, ADX crossovers, or, very much so, Japanese Candlesticks -- these visual cues *do* (for me, at least) carry great (if intuitive) import. Especially candlesticks which, in a single slash-of-an-image, carry not 1 price point, but 4. (I mean, that *rocks*.)
But this 'visual cuing' is learned, too, and not something programmable. If you wish to *try* to program it, you're going to be doing nothing but MAs.
And, to bring it back to the OP's post, ....... That learned, visual cue "magic" (intuition) just *so*much* depends on the lessons along the way -- I think that was Xela's point, too.{??}
For myself, I think you're confusing "TA" with "indicators", there.
What you say is more or less true of "indicators" (though even in that context it might be a slight exaggeration?), but it's certainly not true of "TA" in general.
Okay (he said, picking up the gauntlet..
WRONG about it being "nothing more than " ____________". I do NOT agree about it being "nothing more" as you've described. (++ for the pint)
As for "indicators", most aspiring TA'ers don't use them correctly, either.
... all my own trading (both earlier on in my career, as an independent trader, and now as an institutional one) is strongly related to support and resistance - however local - and that's certainly TA, and it's equally certainly a whole lot more than "running averages over market action" (which I don't do at all).
"Program that."
(Which is to say, if you go after programming the visual cues that are "Support" and "Resistance", you will look for statistical turns. To get those statistical turns, you're going to do what? Establish a trend. To establish that *trend*, you're going to..... to...... to.......)
Yes. You're going to establish a moving average.
(In my mind's eye, I'm clinking glasses with you and Scat-man.
)