Why does TA not work (for you)?

[logic_man]

Quote from marketsurfer:

"read the markets". That sure sounds intuitive or skill based. I have no issue with intuitive TA. It can't be tested, I would never use it, but if it works for you, I have no reason to argue. But why you , logic , and the rest of the TA brigade does not admit this and insists its objective and programmable. Is what I don't understand. NoDoji admits to intuition.

Why do I insist on this? Because for me TA "can be" (not that everyone would actually implement it this way, but I implement it this way) a form of algorithmic trading. Algorithms are, by definition, objective and programmable.

What I don't understand is how you can, on the one hand, defend and praise HFT, which is also (primarily?) algorithmic trading and deride TA, which definitely has overlaps with algorithmic trading, even if not everyone's variation of TA is algorithmic.

Then, there is the contradiction of you saying that "fixed systems" are obsolete, yet HFT, which you seem to like, is almost all "fixed systems", since you don't have time to be deciding among various options when your execution times are that fast. There is probably a good deal of "branching logic" in HFT algorithms, but even with branching, they are "fixed systems".

 

[cornix]

Quote from logic_man:

Why do I insist on this? Because for me TA "can be" (not that everyone would actually implement it this way, but I implement it this way) a form of algorithmic trading. Algorithms are, by definition, objective and programmable.

What I don't understand is how you can, on the one hand, defend and praise HFT, which is also (primarily?) algorithmic trading and deride TA, which definitely has overlaps with algorithmic trading, even if not everyone's variation of TA is algorithmic.

Then, there is the contradiction of you saying that "fixed systems" are obsolete, yet HFT, which you seem to like, is almost all "fixed systems", since you don't have time to be deciding among various options when your execution times are that fast. There is probably a good deal of "branching logic" in HFT algorithms, but even with branching, they are "fixed systems".

Indeed... HFT is based purely on information about trades/orders, without any fundamental factors considered... But HFT is "cool" and TA is "not".

 

[kut2k2]

Quote from marketsurfer:

Why does is a chart generated with a random number generator ( within parameters obviously) look just like a stock chart then?

As a rule, it doesn't. There are any number of longtime traders who can tell the difference between a random chart and a real stock chart. I'm not one of them but my experience is with data analysis algorithms, not charts.

 

[cornix]

Quote from marketsurfer:


Why does is a chart generated with a random number generator ( within parameters obviously) look just like a stock chart then?

Are you sure it also acts just like a stock chart?

Many things look similarly, which doesn't make them equal.

 

[jcl]

Quote from kut2k2:

As a rule, it doesn't. There are any number of longtime traders who can tell the difference between a random chart and a real stock chart. I'm not one of them but my experience is with data analysis algorithms, not charts.

From what I've read, veteran traders can normally NOT tell the difference between a random chart and a real chart. Aronson mentions a couple of studies about this in his book.

A computer can easily tell the difference. The chart below shows the distribution of price movements in the same direction. The x axis is the number of same-direction bars. Red is a real price curve, blue is random data.

.

You'll get such a distribution with all assets and all time frames, except for low-volume stocks or very short time frames, where the prices behave random.

 

[logic_man]

Quote from jcl:

From what I've read, veteran traders can normally NOT tell the difference between a random chart and a real chart. Aronson mentions a couple of studies about this in his book.

A computer can easily tell the difference. The chart below shows the distribution of price movements in the same direction. The x axis is the number of same-direction bars. Red is a real price curve, blue is random data.

.

You'll get such a distribution with all assets and all time frames, except for low-volume stocks or very short time frames, where the prices behave random.

So, if I am interpreting your chart correctly, there are more multi-step movements in the same direction in actual prices than would be predicted by random prices? If prices were truly random, we would see more sequences of 1 move in one direction followed by 1 move in the opposite direction, correct?

 

[wrbtrader]

There was a thread here at ET were someone posted several charts and ask ET members to determine which charts were random and which were real stock charts.

The charts were line charts. Members responded to say that most traders do not use line charts for their trade decisions. Therefore, the question had an intentional bias input by the author (a random walk theorist) via using line charts that the typical trader does not use...similar to http://www.teamten.com/lawrence/writings/are_stocks_a_random_walk.html

Any random walk theorist that uses line charts, charts missing a high, low, close or open input, charts missing time input...that's someone intentionally trying to manipulate the result towards randomness when asking real traders to identify random charts from amongst real market charts via charts not typically used by real traders. This particular random theorist has lost credibility and should be ignored.

Going forward, a new thread was started (different thread starter) here at ET involving bar charts with just High, Low and Close (missing the Open input for each interval). Most of the veteran chart users spotted the random charts easily....mainly due to the funny looking gaps and the odd looking contraction on many intervals.

Later in another thread (different thread starter)...same thing again but this time the charts were bar charts with High, Low, Close and Open (missing time input). Similar result but a little more difficulty. Yet, the funny looking gaps, strange looking very long shadows on top/bottom of the same interval helped identify the random charts.

Continuing, a few years later, there was another thread (different thread starter) asking the same thing but this time the charts were candlestick charts (not bar charts nor line charts with all the inputs - H, L, C, O and Time). Similar like results again with a little more difficulty in identifying the random data generated charts...similar to http://www.smbtraining.com/blog/another-look-at-randomness

Therefore, just because random data charts visually look "similar" to real market data charts...many veteran traders (chart users) can determine the difference based upon their "real trading experience" and experience with data. Thus, just because random charts and real charts "look similar" does not imply that real markets are random especially when there are those that can determine the difference.

Regardless, the issue for some isn't if markets are random...

The issue is why are veteran traders able to visually see the difference between random charts versus real charts when most other traders can not. Answer...they know what time of day when markets tend to be active (that "time input"), they understand volatility, they recognize odd (abnormal) looking intervals that looks like "bad quotes", they understand the relationship of data to other nearby data points and a few other things that most other traders wouldn't notice as being important.

There's a lot more to profitable trading than just TA.

That's why I say that I've never met a consistently profitable trader using TA and nothing else (no money management, no discipline, not properly capitalized, no trading experience, no understanding of market context, no proper equipment, no trading plan to organize everything)...nothing...just TA all by itself.

 

[jcl]

Quote from logic_man:

So, if I am interpreting your chart correctly, there are more multi-step movements in the same direction in actual prices than would be predicted by random prices? If prices were truly random, we would see more sequences of 1 move in one direction followed by 1 move in the opposite direction, correct?

That's correct.

Quote from wrbtrader:

That's why I say that I've never met a profitable trader using TA and nothing else (no money management, no discipline, not properly capitalized, no trading experience, no understanding of market context, no proper equipment, no trading plan to organize everything)...nothing...just TA all by itself.

That's a lot of restrictions and probably applies to no real trader.

When you define TA as using only the price and volume data for trading, and not knowing or caring about any market context or fundamental info, then there are many profitable TA traders.

 

[wrbtrader]

Quote from jcl:

That's correct.

That's a lot of restrictions and probably applies to no real trader.

When you define TA as using only the price and volume data for trading, and not knowing or caring about any market context or fundamental info, then there are many profitable TA traders.

Those other variables I mentioned are obviously important. For example, if your discipline sucks and/or you ignore your trading plan that tested profitable when you had followed your trading plan...market context or fundamentals info isn't the issue for those particular losing trading days. Instead, crappy discipline and a trader that ignores the trading plan is the issue for failure.

That's my point, on any given trading day, one or a few of those variables I've mentioned will have impact on your trading results.

Therefore, doesn't it make sense to have as many variables (e.g. good money management) working for you and not against you while you're using "technical analysis"...of course it makes sense.

That's why there are NO consistently profitable traders using TA and nothing else because they are obviously relying on other tools with their TA.

 

[Redneck]

Quote from cornixforex:

Indeed... HFT is based purely on information about trades/orders, without any fundamental factors considered... But HFT is "cool" and TA is "not".

HTFs can be beat

TA - I leave it to you to decide

(not meaning you CF)

RN