Technical analysis is total bunk.

[tripper22]

He's calling down here: https://bitcointalk.org/index.php?topic=144565.msg1533460#msg1533460

And his call might be right. This time. But TA is the same as flipping a coin. Sometimes you guess correctly. It's not because the squiggly lines predicted the future. It's not because you are a TA guru. You just guessed correctly. This time.

[1714869960]

1714869960

[1714869960]

1714869960

[1714869960]

1714869960

[1714869960]

1714869960

[notme]

Quite easily...

and the award for completely sidestepping valid points goes to........

you. not me.

hint: how many algebraic terms do you think one would need to construct a polynomial function that estimates price?

That's because I agree with your points about calculus, volume, and momentum.

It depends on how much information you have.  If you knew everything you could likely reduce it to a very small number of variables (ie big players).  If you are trying to tease it out of a bunch of noisy signals, quite a few.  Your example was just a quadratic shifted linearly.  I've seen those before.

[arepo]

Make some TA predictions in advance so we can be in awe of your future predicting abilities. It's easy, all you have to do is say up, down or trade in a range and there will be a good chance that you will pick the right one. 1 of 3. TA is bunk. If it's not, prove it.

He's calling down here: https://bitcointalk.org/index.php?topic=144565.msg1533460#msg1533460

And his call might be right. This time. But TA is the same as flipping a coin. Sometimes you guess correctly. It's not because the squiggly lines predicted the future. It's not because you are a TA guru. You just guessed correctly. This time.

whoa, are you going to let me try to prove it or are you going to stick stubbornly to your assumptions? if nothing i say will change your mind, i'm not going to waste my time. but since my reputation is on the line...

this was a good one,

and this bearish divergence predicted the flattening of prices (=trading within a range, as opposed to continuing the trend).

and i may be wrong in my new call. but that doesn't disprove TA either. it obviously can't predict the future because markets are anti-inductive and some irrational or large players or big news events can add additional pressures to the underlying market forces. but the fact is, we're overdue for a correction. the indicators all agree.

and it's certainly better than 50/50 though, i can tell you that. you do need to understand stochastic calculus for that point, though.

also, are you going to respond to any of my points? do you understand the basic principle that momentum changes before price, at least?

[arepo]

Quite easily...

and the award for completely sidestepping valid points goes to........

you. not me.

hint: how many algebraic terms do you think one would need to construct a polynomial function that estimates price?

That's because I agree with your points about calculus, volume, and momentum.

It depends on how much information you have.  If you knew everything you could likely reduce it to a very small number of variables (ie big players).  If you are trying to tease it out of a bunch of noisy signals, quite a few.  Your example was just a quadratic shifted linearly.  I've seen those before.

i'm glad we agree on something

im not sure if your formulation would work though. i'm talking about a function that traces the actual path of the price, which could be estimated by a function of only one variable with a really, really large taylor expansion (c1 x^n + c2 x^(n-1) + c3 x^(n-3) ... + cn x + C) rather than one which solves for price given multiple parameters f(n, V, N) where n= number of large players, V=daily volume, n=good news or something like that.

the beauty of taylor approximations is that they're possible. any differentiable function of one variable can be estimated by a polynomial to any arbitrary precision.

but that's besides the point. the original example was a mixed cubic. its derivative was a mixed quadratic. good on you that you know that it has a basic cup-shaped graph but you wouldn't be able to, for instance, immediately tell me what its intercepts and minimum value were without a little bit of math. if you have the graph in front of you, these things are immediately obvious.

but this explanation is mostly for other readers anyway, as it seems we are in agreement as to why representing the data graphically is advantageous.

[notme]

Quite easily...

and the award for completely sidestepping valid points goes to........

you. not me.

hint: how many algebraic terms do you think one would need to construct a polynomial function that estimates price?

That's because I agree with your points about calculus, volume, and momentum.

It depends on how much information you have.  If you knew everything you could likely reduce it to a very small number of variables (ie big players).  If you are trying to tease it out of a bunch of noisy signals, quite a few.  Your example was just a quadratic shifted linearly.  I've seen those before.

i'm glad we agree on something

im not sure if your formulation would work though. i'm talking about a function that traces the actual path of the price, which could be estimated by a function of only one variable with a really, really large taylor expansion (c1 x^n + c2 x^(n-1) + c3 x^(n-3) ... + cn x + C) rather than one which solves for price given multiple parameters f(n, V, N) where n= number of large players, V=daily volume, n=good news or something like that.

the beauty of taylor approximations is that they're possible. any differentiable function of one variable can be estimated by a polynomial to any arbitrary precision.

but that's besides the point. the original example was a mixed cubic. its derivative was a mixed quadratic. good on you that you know that it has a basic cup-shaped graph but you wouldn't be able to, for instance, immediately tell me what its intercepts and minimum value were without a little bit of math. if you have the graph in front of you, these things are immediately obvious.

but this explanation is mostly for other readers anyway, as it seems we are in agreement as to why representing the data graphically is advantageous.

If you're going past x^4 and you don't have a damn good reason you are overfitting.  Why would you try to model price like that?

[arepo]

If you're going past x^4 and you don't have a damn good reason you are overfitting.  Why would you try to model price like that?

i wouldn't. i was contrasting a taylor expansion-style price estimator with a function that takes multiple parameters.

[notme]

If you're going past x^4 and you don't have a damn good reason you are overfitting.  Why would you try to model price like that?

i wouldn't. i was contrasting a taylor expansion-style price estimator with a function that takes multiple parameters.

Can you give me an example of what a taylor expansion-style price estimator is?  I understand taylor expansions, but I have no clue how to use that to "estimate price".  I also don't see how building an equation to guess the price is not modeling the price, but maybe your example can clear that up too.

[arepo]

i dont want to derail the thread. but let me clarify:

you suggested a formulation that might take a parameter like "number of large players". i am more in favor of a (mental) model of the price that looks like a taylor expansion and which acts like a polynomial with local minima and maxima that can be anticipated by tracking the derivative, or rate of change (price 'momentum' in trader terms). it wouldn't actually be a good method to try to model price as far as fitting of a curve.

in other words, it was merely an example to show that the idea of tracking the derivative of a much simpler polynomial can be applied to tracking the momentum of the 'price function'. a thought experiment, if you will.

[ciphermonk]

Technical analysis is based on one assumption:  that there exists time-correlations in market prices.

If someone manages to formally prove the existence of these correlations, that would settle it for me.

I've seen very complex attempts at extracting these correlations, through artificial intelligence algorithms such as high-dimensional support vector methods. These algorithms can find extremely complex correlations in the data that would be very hard for us humans to grasp, or completely unintuitive. If these methods fail at detecting correlations, I have a hard time with the credibility of "toy functions" used in classical TA.

My 2 bitcents. I could be completely wrong and thats fine.

(note: finding correlations amounts to predicting the price better than random. Of course, random can be right sometimes, but it's predictive power is useless.)

[Piper67]

Even Proudhon is right twice a day... TA is just so much snake oil, as the number of unknown variables is greater than 1.

[piramida]

Yeah, TA is the same as a coin toss - but only if the coin would generate similar results for everybody tossing at the same time. So what Puppet said - instead of being 50% random, TA gives the edge of knowing when to enter / exit the market because of herd mentality, markets are all about herds actually.

Not always right, but even more than 50% is good enough.

[KTE]

PRICE and only PRICE should dictate your actions as a trader.

I'm sorry but that's a ridiculous statement and unfortunately does show inexperience in financial analysis. Price is a completely arbitrary metric, and has zero effect on trading decisions. Market cap is what is relevant, and as a trader even that is only relevant if you can estimate the future perceived market cap of what you're trading.

[xxjs]

TA promoters should be aware of the fact that the value of a bitcoin was zero from the time of genesis, so the appreciation of one bitcoin (in percent or fraction) is always infinite. You can therefore not go too far back in time with your formulas. Personally I think the crash of 2011 is not relevant any more, too much have changed.

[arepo]

Yeah, TA is the same as a coin toss - but only if the coin would generate similar results for everybody tossing at the same time. So what Puppet said - instead of being 50% random, TA gives the edge of knowing when to enter / exit the market because of herd mentality, markets are all about herds actually.

Not always right, but even more than 50% is good enough.

technical analysis works in ways other than simple herd behavior. this point has been addressed in this thread already. in fact, the most objective comment is by far this one:

Technical analysis is based on one assumption:  that there exists time-correlations in market prices.

If someone manages to formally prove the existence of these correlations, that would settle it for me.

I've seen very complex attempts at extracting these correlations, through artificial intelligence algorithms such as high-dimensional support vector methods. These algorithms can find extremely complex correlations in the data that would be very hard for us humans to grasp, or completely unintuitive. If these methods fail at detecting correlations, I have a hard time with the credibility of "toy functions" used in classical TA.

My 2 bitcents. I could be completely wrong and thats fine.

(note: finding correlations amounts to predicting the price better than random. Of course, random can be right sometimes, but it's predictive power is useless.)

i also added my own 2 cents about the relationship between technical analysis and calculus, which i believe is why it is effective.

[Zangelbert Bingledack]

TA ain't bunk, but there is also the maxim: "Just because something is true doesn't means it's always useful."

Provided you believe in Bitcoin's potential, trying to trade on volatility in this market, even with the advantage TA can provide, strikes me as incredibly piggish. 1000% growth every 18 months not good enough for ya? There's such a thing as being too smart for your own good.

[arepo]

@OP

[snip]

Make some TA predictions in advance so we can be in awe of your future predicting abilities. It's easy, all you have to do is say up, down or trade in a range and there will be a good chance that you will pick the right one. 1 of 3. TA is bunk. If it's not, prove it.

[snip]

since my reputation is on the line...

this was a good one,

and this bearish divergence predicted the flattening of prices (=trading within a range, as opposed to continuing the trend).

and i may be wrong in my new call. but that doesn't disprove TA either. it obviously can't predict the future because markets are anti-inductive and some irrational or large players or big news events can add additional pressures to the underlying market forces. but the fact is, we're overdue for a correction. the indicators all agree.

and it's certainly better than 50/50 though, i can tell you that. you do need to understand stochastic calculus for that point, though.

also, are you going to respond to any of my points? do you understand the basic principle that momentum changes before price, at least?

i just want to thank you for the spectacular timing of this post. i can now add my most recent call to the above list. even if technical analysis is bullshit -- which it is not -- you really put your foot in your mouth with this one.

[arepo]

TA ain't bunk, but there is also the maxim: "Just because something is true doesn't means it's always useful."

Provided you believe in Bitcoin's potential, trying to trade on volatility in this market, even with the advantage TA can provide, strikes me as incredibly piggish. 1000% growth every 18 months not good enough for ya? There's such a thing as being too smart for your own good.

holy hindsight bias, batman!

-facepalm-

[Zangelbert Bingledack]

TA ain't bunk, but there is also the maxim: "Just because something is true doesn't means it's always useful."

Provided you believe in Bitcoin's potential, trying to trade on volatility in this market, even with the advantage TA can provide, strikes me as incredibly piggish. 1000% growth every 18 months not good enough for ya? There's such a thing as being too smart for your own good.

holy hindsight bias, batman!

-facepalm-

Not hindsight: note the bolded text. If you're not investing because you think you'll be able to buy a house with a single bitcoin in the future, the statement doesn't apply to you.

[tripper22]

It's nice to see a lot of back and forth on this topic. I enjoy the reading all the varied opinions.

[tripper22]

@OP

[snip]

Make some TA predictions in advance so we can be in awe of your future predicting abilities. It's easy, all you have to do is say up, down or trade in a range and there will be a good chance that you will pick the right one. 1 of 3. TA is bunk. If it's not, prove it.

[snip]

since my reputation is on the line...

this was a good one,

and this bearish divergence predicted the flattening of prices (=trading within a range, as opposed to continuing the trend).

and i may be wrong in my new call. but that doesn't disprove TA either. it obviously can't predict the future because markets are anti-inductive and some irrational or large players or big news events can add additional pressures to the underlying market forces. but the fact is, we're overdue for a correction. the indicators all agree.

and it's certainly better than 50/50 though, i can tell you that. you do need to understand stochastic calculus for that point, though.

also, are you going to respond to any of my points? do you understand the basic principle that momentum changes before price, at least?

i just want to thank you for the spectacular timing of this post. i can now add my most recent call to the above list. even if technical analysis is bullshit -- which it is not -- you really put your foot in your mouth with this one.

You have to do better than 1 correct call. In a previous post I said that there was a chance you could be correct. Let's see how your future calls pan out. I'm not holding my breath.