Feb

7

 Do markets learn from each other? For example, is the S&P market this year, following a similar path to bonds last year, with every trepidatious move down being requited with a rise? Are such "learnings" graduated to the point of regularities. And is it a domino effect or a path of least resistance or consilience or convergent evolution or what have you? What do you think? Can it be quantified? Should it be quantified?

Gary Phillips writes: 

Price discovery has become as anachronistic a term as capital formation. The Fed is supposed to be listening to the market to give it guidance in it's policy decisions, not dictating to the market, what the market should do. If investors feel there are no surprises left, as the Fed is concerned, they will once again lever up, and inflate asset prices…QE1 - QE whatever. Rinse and repeat.

Bill Rafter adds: 

Here’s a related (perhaps derivative) question: Do stocks learn from each other?

Let’s say you take a list of ~6,000 stocks and look at them over a 10+ year period encompassing both up and down markets. And you come up with a trading plan that buys a stock if it exhibits pattern ‘A’ and sell it if it exhibits pattern ‘B’. It is not unreasonable to then have a universe of perhaps 150,000 transactions.

On the first pattern you pick you will find positive average results for certain stocks, while other stocks on average will be negative. Some of these winning stocks will knock the cover off the ball by having say 35 of 50 trades positive, and vice-versa.

Now let’s say you pick different patterns, and again you find a collection of stocks that outperform. You think this is going to be somewhat of a random process where some of the winning stocks from pattern A/B become losers when you try patterns C/D or Y/Z. And that does occur. But you also find that some stocks are consistent winners throughout the various patterns. And of course, some are repeat losers (perhaps hoo-do stocks).

That leads to further inquiry to find what constitutes winning qualities or hoo-do qualities. Note that this is not a study of profitable patterns since the behavior is exhibited across different patterns, some mutually exclusive like trend-following and mean-reversion. Nor is it a study of good management styles, as the same behavior is exhibited with ETFs, which typically have no management.

You then try to identify specific characteristics (or group membership) of the winners. You might think sectors, because the behavior also occurs in ETFs, but not all of the constituents of a winning ETF are consistent winners, and ETFs behave differently than their constituents. You trot through the various possibilities: volume, volatility, beta, name recognition, size, sales, earnings, debt, short interest, institutional holdings, etc.

Continuously you come back to the possibility that some stocks are simply winners and others hoo-dos, until you can prove otherwise. It turns out (tongue firmly in cheek) that this is a good thing to know.
 

Alex Castaldo writes: 

Probably I misunderstand or oversimplify the issue, but I think what is happening is like this.

Among stocks in your database there are some that have considerable price runups and declines, and others that have fewer such features. It is not exactly a question of volatility as conventionally defined, but it is somewhat related to it.

When you examine trading rules, selecting ones that are more successful with some stocks, you are necessarily picking up more stocks in the first category and fewer in the second.

There is a kind of oversampling at work that concentrates the successful population with stocks from the first category.

To take an absurd extreme to make things clear, if there is a stock that spent the entire data period at 10.00 (the ultimate quiet stock), then no method will make money from this stock, not trend-following, not mean-reversion, not seasonality, etc. This stock has zero chance of being among the 'winner' stocks, not because of its industry, or who is the CEO but simply because of the time-pattern of prices.

 


Comments

WordPress database error: [Table './dailyspeculations_com_@002d_dailywordpress/wp_comments' is marked as crashed and last (automatic?) repair failed]
SELECT * FROM wp_comments WHERE comment_post_ID = '7166' AND comment_approved = '1' ORDER BY comment_date

Name

Email

Website

Speak your mind

Archives

Resources & Links

Search