Wow, I'm surprised this thread went as long as it did. To try to narrow the focus a little bit, I want to make a distinction between technical analysis as it is practiced by individual investors and TA/quantitative analysis as done by the pros. As I said earlier in the thread, it may be possible to make money in the market without doing fundamental analysis; however, this requires heavy-duty quant to identify true arbitrage opportunities. Perhaps in-depth analysis of previous price movements might reveal some subtle and intricate patterns which can be exploited, as John W. Henry's fund claims to do (though he has lagged the S&P500).

The technical analysis which is provided as advice on this board and other investing forums generally consists of strategies like, "There's heavy resistance at 40, but if it breaks out, it could go to 50," or, "It's generally a bad idea to buy a stock under its 200-day moving average." This type of analysis seems woefully inaccurate and superstitious.

[ QUOTE ]
The whole idea of predicting accurately poses large problems.

This is part of the thinking behind his comment.

JWH does not predict anything. He simply follows trends.

Most of this thread is about who has the more valid prediction model.

The need to be right (for example being right about a prediction) is a perennial snare that catches many 'traders' and 'investors' and empties their wallets.

It will be interesting to hear from some of the more experienced traders on this topic.

Predicting is not an essential task. What is essential is a willingness to follow a non-discretionary backtested method, long-term, based on expected value.

[/ QUOTE ]
How can you say that predicting is not essential? The entire point of stock-picking is to choose which stocks will go up and which will go down in a given timeframe. I'm not saying that TA systems have to be right 100% of the time. If they were 1% better than chance, they'd be hugely profitable. If you could correctly pick where a roulette ball is going to land 1 in 10 times, you'd make plenty of money.

The real test is whether TA can predict future results. Anyone up for this challenge? Take a look at the entire stock market, and choose 10 or 20 stocks you think have the charts with the clearest indication of movement, either up or down. Post your predictions for whether the stocks will go up or down, and how long it will take for the movement to happen (e.g. 1 day, 1 week, etc.)

Depending on how many people take up this challenge, we'll likely have some better than 50%. We'll use multiple-comparison corrections to evaluate whether anyone was successful at p < 0.01.