After using this trading simulator I'm convinced technical analysis simply doesn't work

[nxt7]

http://chartgame.com/

(Edit: The game in the link above loads up historical data of real stocks. You trade at a particular snapshot in time of the stock's performance.)

I don't mean to cause any offence to those who frequently use it here, but things I've learnt after having a play with this for several hours:

- Short/long term crossovers don't work unless you just use it to buy/hold for the long run. Even then, plenty of whiplashes undo your gains....
- Support/Resistance works sometimes but not most of the time..
- Chart patterns rarely work...
- MACD, RSI and ADX just gives too many false signals...
- Buy/hold wins everytime if you're "lucky" to hang on to a good stock....

Mind you this is just a simulator. If it were actual trading I would have probably lost my account several times over + broker fees.

Whether its years of experience staring at charts and eventually "seeing the light" or just being lucky every time, it leads me to wonder how some people find success using TA at all....

Now I still want to believe TA works and that its probably me not using it properly. But after this, kind of wondering what strategies there are left that give one the "edge" after months/years of experience...

 

[Spectre2007]

with random data none of the above will work, real world exchange data has inefficiencies based on human triggers. Human triggers are the inefficiencies in trading.

 

[nxt7]

with random data none of the above will work, real world exchange data has inefficiencies based on human triggers. Human triggers are the inefficiencies in trading.

Its not random data.

The game in the link in the OP loads up historical data of real stocks. You trade at a particular snapshot in time of the stock's performance.

 

[Spectre2007]

if I explained it, I would be giving it away. The change in price at the bid-ask level is a fractal of the larger timeframes extrapolated out.

 

[speedo]

http://chartgame.com/

(Edit: The game in the link above loads up historical data of real stocks. You trade at a particular snapshot in time of the stock's performance.)

I don't mean to cause any offence to those who frequently use it here, but things I've learnt after having a play with this for several hours:

- Short/long term crossovers don't work unless you just use it to buy/hold for the long run. Even then, plenty of whiplashes undo your gains....
- Support/Resistance works sometimes but not most of the time..
- Chart patterns rarely work...
- MACD, RSI and ADX just gives too many false signals...
- Buy/hold wins everytime if you're "lucky" to hang on to a good stock....

Mind you this is just a simulator. If it were actual trading I would have probably lost my account several times over + broker fees.

Whether its years of experience staring at charts and eventually "seeing the light" or just being lucky every time, it leads me to wonder how some people find success using TA at all....

Now I still want to believe TA works and that its probably me not using it properly. But after this, kind of wondering what strategies there are left that give one the "edge" after months/years of experience...

It takes a lot of experience watching charts before they make sense. Reading books, attending seminars or "playing" with simulations will not get you there. It is a common tendency of the novice to look for the flaw everywhere but within the novice.

 

[nxt7]

if I explained it, I would be giving it away.

I'm not interested in your strategy. It most likely won't even work for me (or others) because I would see it very differently than you do.

The change in price at the bid-ask level is a fractal of the larger timeframes extrapolated out

I've tried extrapolating trends in different time frames. Honestly, don't see any fractals.

 

[botpro]

Having TA is better than having nothing at all.

There are about 200 indicators and oscillators.
So, it highly depends on which of them you have used in your test, and of course also the parameters passed to them.

For some people for example MACD is the whole trading system...

You can generate TA signals and let it additionally confirm by your own methods for signal generation...

Or make like this: collect signals (ie. 0/1 signals) from many signal sources and take it only if the count makes at least x%... for example 80% or higher

You need to do many tests to find those which work for you for the tested instruments.

Important: the bar-size plays a big role, ie. the timeframe of the bars like 1-minute bars, 5-minute bars, EOD bars, etc...

And: it is also possible that the said web backtester can have bugs.
I would suggest to convince yourself by comparing its results with the result of another such backtester... As everybody knows: most web apps are just buggy cr*p...

 

[Spectre2007]

I'll try to explain some of it to you, but its at the core of all trading.

- the current bid/ask when nonmotile and static over time represents 100% efficiency. Meaning the participants have agreed the current price differential of buyers to sellers is at equilibrium. So price is consistently the same second to second.

- as equilibrium is disrupted there is disagreement in the current differential, so the bids and asks get hit with varying quantities, if the quantities are grossly the same both sides, the price might stutter or shake at the micro level but stay the same. As greater quantities are weighted on the bid or ask, the price becomes unstable and starts to move in the direction where matching occurs in supply/demand at new price level.

- bid/ask - tick level
- bid/ask - highest lowest over 5 minutes
- bid/ask - highest lowest over 15 minutes

etc... to the daily..monthly...

its essentially a breakout from the smallest to the largest timeframe. So price will try to discover itself by bouncing off these levels, and breaking out to find a new level. The direction of the breakouts at different timeframes.. creates a trend. You want to be in the direction of the breakouts but demand the low end of the range of the timeframe being traded.

 

[botpro]

Now I still want to believe TA works and that its probably me not using it properly. But after this, kind of wondering what strategies there are left that give one the "edge" after months/years of experience...

Diversification is one of the ingredients for success...

 

[botpro]

I'll try to explain some of it to you, but its at the core of all trading.

- the current bid/ask when nonmotile and static over time represents 100% efficiency. Meaning the participants have agreed the current price differential of buyers to sellers is at equilibrium. So price is consistently the same second to second.

- as equilibrium is disrupted there is disagreement in the current differential, so the bids and asks get hit with varying quantities, if the quantities are grossly the same both sides, the price might stutter or shake at the micro level but stay the same. As greater quantities are weighted on the bid or ask, the price becomes unstable and starts to move in the direction where matching occurs in supply/demand at new price level.

- bid/ask - tick level
- bid/ask - highest lowest over 5 minutes
- bid/ask - highest lowest over 15 minutes

etc... to the daily..monthly...

its essentially a breakout from the smallest to the largest timeframe. So price will try to discover itself by bouncing off these levels, and breaking out to find a new level. The direction of the breakouts at different timeframes.. creates a trend. You want to be in the direction of the breakouts but demand the low end of the range of the timeframe being traded.

Good, but isn't something missing? Like how that is related to the topic of this thread?...