Technical Analysis - a hero or Joker?

TracerBullet

Well-Known Member
#11
Then I do not know the point you are making. Where did I say that TA works on random data.
I read 1) Stock market is random 2) TA can work
So it implies TA can work on random data.
Anyway, doesnt not matter, good luck with your trading ..

Edit -
mohan.sic said:
But in general even stock prices are random, but some what predictable. Because, data flow is somewhat sequential and the next data point usually falls within a range from previous data point. So with a reasonable amount of past data, this gives a scope for prediction.
Random is random, it is not predictable.

If you are interested --

You can create random price data as described above. This data will also be somewhat sequential but the % percent changes will be in randomized order ( even though they are real % changes, the order is shuffled )
In this dataset no real edge is possible. Anything you find will be only curve fit to the data and then you can randomize again to confirm that the edge was not real. I have done this, tested a system on random data and the edge goes away.

Again, all of this is just for tp. If you have an edge, enjoy the money ..
 
Last edited:

mohan.sic

Well-Known Member
#12
I read 1) Stock market is random 2) TA can work
So it implies TA can work on random data.
Anyway, doesnt not matter, good luck with your trading ..

Edit -


Random is random, it is not predictable.

If you are interested --

You can create random price data as described above. This data will also be somewhat sequential but the % percent changes will be in randomized order ( even though they are real % changes, the order is shuffled )
In this dataset no real edge is possible. Anything you find will be only curve fit to the data and then you can randomize again to confirm that the edge was not real. I have done this, tested a system on random data and the edge goes away.

Again, all of this is just for tp. If you have an edge, enjoy the money ..


I understand the point you are expressing. Very valid. Edge goes away. I will go once step further and say that even Optimizing a system or method is useless. Because when we are optimizing a system, we fail to realize that what we are actually working on is optimizing a system to a specific set of data. And when the new optimized rules applied on new data series can turn out useless.

Now about "Stock price is Random and TA works on Stocks"
Take a example of a basic pattern like Double bottom which normally signals a Buy. It works sometimes and fails sometimes. Since it works sometimes, I gave a expression that TA works. Now a informed trader like you can point that since the same setup failed many times and hence TA does not work. I have to agree with you without second thought without making a foolish excuse that no method is 100% accurate.

More over in that point I was not upholding TA. I was only saying that at least TA is reasonable prediction method that irrational methods like Astrology or Gann. It is not about success rate. It is just about using a logical method for predicting. To give you another example, I will say that Total buy qty: Total Sell qty ( more tbq suggests buy and tsq suggests sell) is more logical than sophisticated TA indicators if we know how and when to use it.. Again I am not on the point of edge or success rate in using Tbq:Tsq. It is only about rational analysis.
 
Last edited:

TracerBullet

Well-Known Member
#13
I understand the point you are expressing. Very valid. Edge goes away. I will go once step further and say that even Optimizing a system or method is useless. Because when we are optimizing a system, we fail to realize that what we are actually working on is optimizing a system to a specific set of data. And when the new optimized rules applied on new data series can turn out useless.
Yes, need to use some common sense here, there is scope for some optimization but using common sense discretion and probably not by just finding the exact best parameter value.

Now about "Stock price is Random and TA works on Stocks"
Take a example of a basic pattern like Double bottom which normally signals a Buy. It works sometimes and fails sometimes. Since it works sometimes, I gave a expression that TA works. Now a informed trader like you can point that since the same setup failed many times and hence TA does not work. I have to agree with you without second thought without making a foolish excuse that no method is 100% accurate.
This is a very wrong way to think about an edge. It is important to understand what an edge is.
If a setup fails many times, it does not mean TA does not work. EVERY setup will fail sometimes, we are dealing with probabilities and not certainty. You can have everything absolutely perfectly setup and yet the trade can fail.
So what is an edge ? Simple and obvious. Its = Amount You make in winnning trades - Amount you loose in losing trades - costs.
There are other things you can add to refine what an edge is but above is the most basic idea. You probably know this already but above post made me doubt that so i put it here.

Once we have that then any idea/pattern etc has to be tested on your market for a large number of trades to check whether it actually works. For exampke, Does double bottom/top work ? Define it clearly and then take say 100-200 trades or more and then check - did you make money ? If not then it does not work ( atleast in the way you defined it). Anything that any guru says is irrelevant if it does not test out.The worth of any idea is only to check whether it works, although discretionary parts are hard to replicate and define but we must to do that as best we can in a trading plan.

Some posts you may want to read, here and here also check linked posts within it. Or just google Adam H grimes + edge. Even better read his blog in full and also read the first book. ( There is a lot of stuff other than that too, hard to keep up ).

In random data, you will not be able to find anything that makes money over the long term after many trades, but we do have edges in real market data. The pdf linked by Dan is a nice read that speculates why edges might exist in the market and how it can make markets more efficient.

More over in that point I was not upholding TA. I was only saying that at least TA is reasonable prediction method that irrational methods like Astrology or Gann. It is not about success rate. It is just about using a logical method for predicting. To give you another example, I will say that Total buy qty: Total Sell qty ( more tbq suggests buy and tsq suggests sell) is more logical than sophisticated TA indicators if we know how and when to use it.. Again I am not on the point of edge or success rate in using Tbq:Tsq. It is only about rational analysis.
Again this is very wrong. Logical does not mean its real or useful. Anything can sound logical and yet it can be against how market works. The question to ask is whether something works and how well it works and perhaps speculate on why - and not how logical or rational it sounds.
I am not trying to pick on you, just trying to say how i understand things.
 

mohan.sic

Well-Known Member
#14
Yes, need to use some common sense here, there is scope for some optimization but using common sense discretion and probably not by just finding the exact best parameter value.



This is a very wrong way to think about an edge. It is important to understand what an edge is.
If a setup fails many times, it does not mean TA does not work. EVERY setup will fail sometimes, we are dealing with probabilities and not certainty. You can have everything absolutely perfectly setup and yet the trade can fail.
So what is an edge ? Simple and obvious. Its = Amount You make in winnning trades - Amount you loose in losing trades - costs.
There are other things you can add to refine what an edge is but above is the most basic idea. You probably know this already but above post made me doubt that so i put it here.

Once we have that then any idea/pattern etc has to be tested on your market for a large number of trades to check whether it actually works. For exampke, Does double bottom/top work ? Define it clearly and then take say 100-200 trades or more and then check - did you make money ? If not then it does not work ( atleast in the way you defined it). Anything that any guru says is irrelevant if it does not test out.The worth of any idea is only to check whether it works, although discretionary parts are hard to replicate and define but we must to do that as best we can in a trading plan.

Some posts you may want to read, here and here also check linked posts within it. Or just google Adam H grimes + edge. Even better read his blog in full and also read the first book. ( There is a lot of stuff other than that too, hard to keep up ).

In random data, you will not be able to find anything that makes money over the long term after many trades, but we do have edges in real market data. The pdf linked by Dan is a nice read that speculates why edges might exist in the market and how it can make markets more efficient.



Again this is very wrong. Logical does not mean its real or useful. Anything can sound logical and yet it can be against how market works. The question to ask is whether something works and how well it works and perhaps speculate on why - and not how logical or rational it sounds.
I am not trying to pick on you, just trying to say how i understand things.
So what is an edge ? Simple and obvious. Its = Amount You make in winnning trades - Amount you loose in losing trades - costs.
Amount You make in winnning trades - Amount you loose in losing trades - costs.

This is not edge as I see. It is simply profit or loss. ;)

I understand what you mean in that. A trading system with low risk : high reward ratio can generate positive returns even with lower number of winning trades than losers. Eod we may make money with this system even with lower hit rate in signals and you see this as a edge. Agreed. At the same time, not every trader follows same set of rules. I look for edge in the technique I trade. I don't rely on risk reward or money management. I want my base signal generating technique to have an edge. A set of 100 signals should have 75%+ hit ratio and I am happy to trade on 1:1 RR and I see a edge that way.

Again this is very wrong. Logical does not mean its real or useful. Anything can sound logical and yet it can be against how market works. The question to ask is whether something works and how well it works and perhaps speculate on why - and not how logical or rational it sounds.
I am not trying to pick on you, just trying to say how i understand things.
Again I am not saying you are right or wrong as you did, but I cannot agree with you.
For me the method I follow should be sound and logical. It may have an edge or may not have, that's a different thing and can be dealt with. Say If I am a lucky hand and I have a 85% hit ratio in taking a trade by flipping a coin heads or tails...here it works but I cannot follow this illogical method. ( not a fair example I know. but just to set the context). Another similar example, traders rely on certain moving avg like sma50 or ema100. I don't follow those, because I don't see a logic in that. We follow many rules and methods blindly.
I believe candle sticks pattern is not even a predictive tool. I don't see a logic in that. But it is favorite tool for most of traders. I don't prefer to pick illogical ideas and fine tune them for accuracy. My choice is to opt for more rational models and test them, understand where they work, where they fail and trade accordingly.
 

TracerBullet

Well-Known Member
#15
Again I am not saying you are right or wrong as you did, but I cannot agree with you.
For me the method I follow should be sound and logical. It may have an edge or may not have, that's a different thing and can be dealt with. Say If I am a lucky hand and I have a 85% hit ratio in taking a trade by flipping a coin heads or tails...here it works but I cannot follow this illogical method. ( not a fair example I know. but just to set the context). Another similar example, traders rely on certain moving avg like sma50 or ema100. I don't follow those, because I don't see a logic in that. We follow many rules and methods blindly.
I believe candle sticks pattern is not even a predictive tool. I don't see a logic in that. But it is favorite tool for most of traders. I don't prefer to pick illogical ideas and fine tune them for accuracy. My choice is to opt for more rational models and test them, understand where they work, where they fail and trade accordingly.
Yes wanting ideas to be logical is alright as long as we test them out too. I just the meant that only logical/rationality in itself is useless. Usually my process has been to have an idea to test then test it . Usually it does not give what i expected but then i will see something else which turns out to be useful. Just testing and discretion feedback loop works well. But should be open to ideas too. Ma + channels can work well as a basic trend filter.
 

SwamiNathan

Well-Known Member
#16
Random is random. TA will not work on Random data. I sarcastically asked that question.

Prices keep on moving due to three factors. (Only the three factors and not any other)
1) Inflation
2) Past and expected performance of company, industry and country.
3) Sentiments of market participants.

Number One is a long term trend and Almost known to all mkt participants. Number two comes in the form of news and 99% of that is not forecastable. And number three forms the foundation for Technical analysis. During 1950 to 1990s the third factor's contribution were more. So technical analysis worked. After the transformation of IT, news reach us soon and also there are many trading systems and methodologies which act logically than emotionally, thereby reducing the emotional factor. Therefore TA will work very lesser than it worked before 20 years. The extent it works is not enough to meet the trading cost, like impact cost, STT, brokerage etc...
 

TracerBullet

Well-Known Member
#17
Random is random. TA will not work on Random data. I sarcastically asked that question.

Prices keep on moving due to three factors. (Only the three factors and not any other)
1) Inflation
2) Past and expected performance of company, industry and country.
3) Sentiments of market participants.

Number One is a long term trend and Almost known to all mkt participants. Number two comes in the form of news and 99% of that is not forecastable. And number three forms the foundation for Technical analysis. During 1950 to 1990s the third factor's contribution were more. So technical analysis worked. After the transformation of IT, news reach us soon and also there are many trading systems and methodologies which act logically than emotionally, thereby reducing the emotional factor. Therefore TA will work very lesser than it worked before 20 years. The extent it works is not enough to meet the trading cost, like impact cost, STT, brokerage etc...
2) News - Price does not factor in news immediately. Otherwise we would see a jump and stop on every big/small news. Maybe that could happen if everyone was a robot and all news was available and processed instantaneously and in the same manner - ie if everyone agrees on price instantly.
So this can show up in patterns created by supply-demand. Pullbacks and overextensions, Momentum / Mean reversion. Inital move out of random sideways is probably impossible to pick but once a move has been made, there is scope for TA.

"The extent it works is not enough to meet the trading cost, like impact cost, STT, brokerage etc"
This is not true, our markets, atleast in stocks, do have persistent mechanical behaviors. STT is a lot ofc in equity but its still feasible. If mech is possible then so are discretionary methods. Descretion does not have to be blind and it can use stats as a base. And a lot of things in TA books probably never worked - 50 years back or now.
But probably no easy edges and i find it hard to believe that anyone can be consistently profitable week after week ..
 
Last edited:

DanPSup

Hedge Strategy Trader in Options and Futures
#19
My thought: Consider the technical analyst is a detective...
Yep, you nailed it for certain traders. Traders, which for example on Sunday evening sit down, do analyze all there scripts on daily, weekly and monthly TF for specific TA moments and circumstances. Those traders then go for trades which are specific kind of trades which mostly are not based on intrad day stress ideas, instead on bigger time frame strategies in which TA is part of the analyzes beside other, very important tools. So: Using the word "TA Detectives" is surely not a wrong word in this cases, even thought it will fit also in other cases. But trading only on TA by itself is, at least in my view, very difficult, this specific on intra day basis. I prefer the mentioned way to have less stress in trading.
 
Last edited:
#20
Yep, you nailed it for certain traders. Traders, which for example on Sunday evening sit down, do analyze all there scripts on daily, weekly and monthly TF for specific TA moments and circumstances. Those traders then go for trades which are specific kind of trades which mostly are not based on intrad day stress ideas, instead on bigger time frame strategies in which TA is part of the analyzes beside other, very important tools. So: Using the word "TA Detectives" is surely not a wrong word in this cases, even thought it will fit also in other cases. But trading only on TA by itself is, at least in my view, very difficult, this specific on intra day basis. I prefer the mentioned way to have less stress in trading.
I meant, a detective collects the traces left behind by a thief to catch him. An analyst does the same with the price. But please don't listen to these blah blahs of mine... I'm just a capital-less loser trader.

For the time being...I'm the joker..
 

Similar threads