Where the problem lies
The core idea of technical analysis is that past price patterns repeat, so you can guess the next move from a chart. The problem is that the human brain finds patterns even in randomness. Telling how much of a trend or support level on a chart is real and how much is coincidence is very hard.
On top of that, markets themselves change. If a rule really works, many people start using it, and the advantage shrinks. So “it worked in the past” and “it will work in future” are not the same thing.
Toss a coin 100 times. For heads, raise an imaginary price by Tk 1; for tails, lower it by Tk 1; and plot each price on a graph. The result is pure luck, yet the graph will often show “uptrends”, “downtrends” and “support” where it keeps pausing. So patterns on real charts cannot be assumed to mean something without evidence.
Why it matters to an investor
- To test claims: social media and messaging groups are full of “the chart says it will definitely rise”. Knowing what research says helps you weigh such claims.
- To understand costs: frequent buying and selling on chart signals means brokerage commission and other costs every time. Even a small advantage can be wiped out by them.
- To catch your own biases: we remember the calls that worked and forget the ones that didn’t, which makes a method seem better than it is.
- To count the cost in time and attention: watching charts every day takes a lot of time and stress. It is worth asking what you get in return.
What the research says
In 1970 the economist Eugene Fama set out the idea of “efficient markets”: new information is quickly reflected in prices, so it is hard to earn steady extra profits from past prices. In 2013 he shared the Nobel prize in economics with Robert Shiller and Lars Peter Hansen. In the Nobel committee’s words, there is no way to predict the price of stocks and bonds over the next few days or weeks, though it is quite possible to foresee their broad course over longer periods such as three to five years.
Research on technical rules is mixed. A major 2007 review (Park and Irwin) found that of 95 modern studies, 56 reported profits, 20 reported none and 19 were mixed. But the authors warned that many studies suffered from data snooping (testing many rules and keeping the one that looks best), choosing rules after the fact, and not measuring risk and costs properly.
- Good in the past, not later: some simple chart rules looked good on the US Dow Jones index from 1897 to 1986. But Sullivan, Timmermann and White (1999) showed that the best rule no longer gave a statistically meaningful gain from 1987 to 1996.
- Very few succeed: a study of day traders in Taiwan from 1992 to 2006 (Barber and co-authors, 2013) found that less than 1% of them could predictably and reliably earn profits after costs.
Source: Park & Irwin, 2007
See the numbers as a table
| Found profits | Mixed | No profits | |
|---|---|---|---|
| 95 studies | 56 | 19 | 20 |
Why it is so hard
- Data snooping: test enough rules and a few will look brilliant on past data by luck alone. “Winning” strategies shown in screenshots are often like this.
- One chart, many readings: analysts often disagree on where to draw a trendline or which level is support.
- News outside the chart: a sudden announcement, policy change or global shock does not show up on a chart in advance.
- Small-market reality: in thinly traded shares, a handful of orders can change how the chart looks, so a “pattern” there is not always the combined view of many people.
Worked example
- Say you trade Tk 1,00,000 on chart signals, buying and then selling 20 times a year (40 trades in all).
- Say each buy-and-sell round costs 0.5% in commission and other charges. Then the yearly cost is 20 × 0.5% = 10%, about Tk 10,000. Real costs differ by broker and by the rules in force; the numbers are only an example.
- You would need roughly 10% extra gain a year just to cover these costs, and any profit may also be taxed.
Result: where research finds an advantage from chart rules, it is usually small. The cost of frequent trading can easily be bigger than that advantage.
Common mistakes
- Treating a backtest or a screenshot of profits as proof, without asking how many rules were tried before this one was picked, and where the losses are.
- Counting profits before trading costs and tax.
- Remembering the predictions that came true and forgetting those that didn’t.
- Believing “the chart guarantees a profit”. In the share market, a promise of certain profit is itself a warning sign.
Charts are good at describing the past; when someone claims to predict the future with them, ask about the evidence, the costs and the risks.