I know there are a whole group of technical analysts who make thier careers by attempting to use previous/past stock trends and extrapolate future performance.
But exactly how accurate are these guys? If thier predictions are correct is it more about how lucky they are than good? How do we ever know?
And in general, is it really possible to predict future stock market trends based on past data?
Are there any other ways of predicting the (future) performance of a given stock? I’m thinking about the analysis of company cost structure, breakdown of assets etc. But you tell me, I’m no specialist.
And what kind of literate do you need to be when analysing these stocks? I mean for example, if you were looking at trend graphs, perhaps you would need mathematical knowledge to dissect future patterns given a certain set of previous data.
Or maybe when analysing the costing structure of a prospective investment, you would need some basic accountancy skills.
Ok look I’m no good at this.
nope, no more possible then it is to make 100% accurate weather predictions. Simply too many unknown variables to factor in, in a situation sensitive to initial conditions. Read “Fooled by randomness” by Nicholas Taleb, he covers all this and more, excellent read.
One of the major issues at hand is the fact that computers can only work with a finite number of decimal places. No matter how many places you have, the error you get by truncating causes your predictions to eventually be completely different from what will happen.
If you graph the digits of pi, you will find instances of the “head and shoulders” pattern that technical analysts swear is so significant.
Basically, they assume that a close reading of a stock’s chart for patterns will give them a clue as to where it’s going. If it works, then they say the “head and shoulders” pattern was a predictor. If it doesn’t work, they say it really wasn’t a “head and shoulders” pattern. So they’re pretty good at predicting the past, but not the future.
They make their careers by selling the recommendations they come up with. They don’t make their careers (i.e. MONEY) from actually doing that. It’s a bogus “science”.
Not very accurate. Not more accurate than an investor who buys an index fund and just holds onto it. . .arguably the “gold standard” that a technical analyst should be measured against.
It is more about how lucky than good they are. And if you look at enough of them basically picking at random, some of them are going to be lucky for an extended period of time. It is no indication that they will continue to be lucky in the future, but that’s how its marketed.
You can “know” by getting their predictions and comparing them against the S&P 500 or some other broad index.
If there were “sure fire” ways of predicting former performance, the price of the stock would reflect it so fast that it would be useless.
People have applied many mathematical models to the stock market. Some have helped for sure, but there is also so much psychology at play in stock prices that mathematical models are often inadequate.
Well, you need to know how much cash is on hand, how much debt they have, inventory, revenues, expenditures, how much the dividend is, the earnings, etc. etc. etc… . .of course all of that is usually pretty freely available, but you’d want some experience to wade through it all.
Now, if you could just plug all those things into a formula that then says, “the stock of the price should therefore be $XXX” then, the stock price would pretty much reflect that. If it got above $XXX people would sell it. If it got below $XXX, people would buy it.
This is simplified, but where it gets tricky is that built into the price of a stock, beyond all that “hard data”, is an estimate of “future growth”. Now, if everybody was able to estimate a company’s growth perfectly, that would be automatically built into the cost of a stock and the price would therefore hover around $XXX, too.
Now all that keeps it confusing enough. When you start trying to figure in speculation and fear – call it “mass psychology” or something – you’re adding something that really can’t be modelled at all.
Does any of that help? I’ve been trying to educate myself on investing for a while now. If you have any mathematics in your background at all, you might enjoy the book “A Random Walk Down Wall Street” by (???) Malkiel, a great book. Most of what I wrote here is a regurgitation of some of his ideas.
I agree. Most (all ?) of these predictions are not accurate. But let’s look at it another way. Suppose they are somewhat accurate. Suppose for simplicity that the market (or a particular stock) is up or down in a gven month with a 50% probability. Suppose our guy can predict movements corectly with a p% probability. How many months would you have to watch him before you were convinced?
The variance of N of his predicitons (assuming he’s guessing) is N0.5(1-0.5) = N/4, and the standard deviation is Sqrt(N)/2. For 95% confidence (a standard level) we need to be 2 standard deviations above the mean. So for a given p we want
If he an pick it right 51% of the time we’d need to watch for 2500 months or a little over 208 years. If he’s right 55% of the time we’d need to watch for 100 months or 8 1/3 years. If he’s right 60% of the time we’d need 25 months.
So even if the guy is good, we may never be able to catch it . So how good might we expect someone to be? You might think that 60% wouldn’t be too hard to get, but you’d be way off. (I’ll have to dig up a cite sorry). Managers would could guess 52% right would soon become very wealthy just investing their money not to mention all the clients they’d attract. It’s very unlikely that eve were there 52% right market investors, yo’d ever be able to confirm it.
I’m not a big fan of “technical analysis”. It’s kind of like looking at clouds. People see what they expect to see.
The problem with trying to predict an individual stock is that the stock price generally reflects all the data that is currently available. You cannot gleen any new information about what the stock will do based on past history. At least not 100%. You can generally rely on the face that a GE or IBM will continue to stay in business in the future, but they could just as easily become another Enron or Worldcom and there would be no way to predict it.
There are companies that do ‘valuations’. Basically, they can attempt to figure out what is a reasonible stock price based on the NPV of the companies expected future cash flows (as well as other techniques). Once again, it’s imperfect because those cash flows are not guaranteed.