Technical analysis is a way of studying markets through their own trading data: price, volume and momentum. It looks for statistical patterns in that data to estimate how the price may move next. The CMT Association awards the Chartered Market Technician designation. It describes technical analysis as a methodology that "evaluates financial markets by analyzing statistical patterns in market data such as price movement, volume, and momentum" to help forecast market behaviour. It differs from fundamental analysis, which estimates value from earnings, cash flows, token economics or network use. Technical analysis is not a branch of fundamental analysis, though many analysts use the two together.
The idea behind it
Technical analysis rests on three working assumptions:
- Price already reflects available information. The analyst studies the market's reaction to news, not the news itself.
- Prices move in trends. A move under way is more likely to continue than to reverse at random.
- Crowd behaviour repeats. People respond to fear, greed and round numbers in similar ways, so similar chart shapes can recur.
The CMT Association says the discipline draws on behavioural economics. It also covers quantitative research and the design and testing of rules-based trading systems, not just reading chart patterns by eye.
The main academic objection comes from the efficient market hypothesis. Eugene Fama reviewed it in the Journal of Finance in 1970. Its weak form holds that current prices already contain the information in past prices. If that is true, rules built only on price history should not earn consistent excess returns. Most of the research summarised below tests that claim.
Core tools
Charts and timeframes
Candlestick, bar and line charts plot price over a chosen period, such as one minute, one hour, one day or one week. Most analysts check several timeframes. The longer one sets the context and the shorter one is used to time decisions.
Trend and moving averages
A simple moving average is the average closing price over the last n periods. It smooths out noise so the direction of the trend is easier to see. Analysts watch whether price sits above or below the average, and where a short average crosses a long one.
Support and resistance
Support is a price area where buying has tended to halt declines. Resistance is an area where selling has tended to cap rallies. These are zones drawn from past trading, not exact levels.
Bollinger Bands
John Bollinger defines them as "an n period moving average with bands drawn above and below at intervals determined by a multiple of standard deviation." His defaults are a 20-period moving average with bands two standard deviations above and below it, calculated with the population standard deviation. When the bands widen, volatility is rising. When they narrow, it is falling.
Relative Strength Index (RSI)
J. Welles Wilder introduced the RSI in his 1978 book New Concepts in Technical Trading Systems. It compares recent gains with recent losses and moves on a scale from 0 to 100. Wilder based it on 14 periods, and most charting software still uses 14 as the default. Readings above 70 are usually called overbought and readings below 30 oversold. These levels are conventions, not fixed rules, and some analysts use 80 and 20 instead.
Volume
Volume shows how much traded during a move. Analysts generally give more weight to a breakout on rising volume than to one on thin volume.
Two ways to apply it
Top-down
Start broad and narrow the view. For stocks, that can mean a major index such as the S&P 500, then sector charts, then a single company on daily and then hourly charts. For crypto, the same order runs from the overall market, through the largest assets, to a single token.
Bottom-up
Start with one asset that interests you, often one that is moving against the wider market. Check it with fundamental research, then use charts to look for entry and exit levels.
What the studies on technical analysis found
Compare six studies by the data they tested, what they found and the main caveat, sorted by year if you like.
| Study | Year | Data tested | What it found | Main caveat | Source |
|---|---|---|---|---|---|
| Brock, Lakonishok and LeBaron (Journal of Finance) | 1992 | Moving-average and trading-range-break rules on Dow Jones data, 1897 to 1986 | Strong support: returns after buy signals were higher and less volatile than after sell signals | The original tests did not account for transaction costs | |
| Sullivan, Timmermann and White (LSE discussion paper) | 1998 | 26 trading rules on about 100 years of Dow Jones data | Re-tested the earlier work with a bootstrap that corrects for data snooping | The article describes only the method, not the results | |
| Lo, Mamaysky and Wang (Journal of Finance) | 2000 | Statistical pattern recognition on roughly 31 years of US stock data | Several indicators provide incremental information and may have some practical value | "May have some practical value" is not a claim of reliable profits | |
| Park and Irwin (Journal of Economic Surveys) | 2007 | Review of 95 modern studies | 56 positive, 20 negative, 19 mixed; profits in a range of speculative markets at least until the early 1990s | Most studies had testing problems: data snooping, rules chosen after the fact, risk and cost estimates | |
| Detzel, Liu, Strauss, Zhou and Zhu (Financial Management) | 2021 | Daily Bitcoin returns | The ratio of price to its moving average forecast returns in and out of sample, with alpha and Sharpe ratio gains over buy-and-hold | Does not show that charts reliably predict prices | |
| Wei, Kyriakou, Sermpinis and Stasinakis (International Journal of Finance and Economics) | 2024 | Thousands of technical rules and dozens of fundamental factors on 12 cryptocurrencies | Only a short-term log moving-average ratio and a hashrate index were profitable in and out of sample | The authors question the value of both technical and fundamental analysis on cryptocurrencies |
This table summarises only the six studies named in the article; it does not rank them, cover later research or show whether any rule will work in today's markets.
Does technical analysis work? What the studies say
The evidence is mixed. Results depend heavily on the market, the period and how carefully the testing is done.
- Brock, Lakonishok and LeBaron (1992, Journal of Finance) tested simple moving-average and trading-range-break rules on Dow Jones data from 1897 to 1986. They found strong support for the rules. Returns after buy signals were higher and less volatile than returns after sell signals. The original tests did not account for transaction costs.
- Sullivan, Timmermann and White (1998, LSE Financial Markets Group discussion paper) went back over that work with a bootstrap that corrects for data snooping, the bias that comes from testing many rules and reporting the winners. They examined 26 trading rules on about 100 years of Dow Jones data.
- Lo, Mamaysky and Wang (2000, Journal of Finance) used statistical pattern recognition on roughly 31 years of US stock data. They concluded that "several technical indicators do provide incremental information and may have some practical value."
- Park and Irwin (2007, Journal of Economic Surveys) reviewed 95 modern studies: 56 found positive results, 20 negative and 19 mixed. They found that technical strategies made economic profits in a range of speculative markets "at least until the early 1990s." They also found that most studies had testing problems, including data snooping, choosing rules after the fact, and difficulty estimating risk and transaction costs.
Technical analysis in crypto
Crypto assets often have fundamentals that are hard to value, which is one reason charts are so widely used in this market. The research is split here too:
- Detzel, Liu, Strauss, Zhou and Zhu (2021, Financial Management) built a model in which technical analysis comes out of rational learning about assets that are hard to value. Their data showed that the ratio of price to its moving average forecast daily Bitcoin returns both in and out of sample. Strategies built on that ratio showed alpha and Sharpe ratio gains over buy-and-hold.
- Wei, Kyriakou, Sermpinis and Stasinakis (2024, International Journal of Finance and Economics) tested thousands of technical rules and dozens of fundamental factors on 12 cryptocurrencies. Only a short-term log moving-average ratio and a hashrate index showed genuine profitability both in and out of sample. The authors concluded that their findings "question the value of both technical and fundamental analysis on cryptocurrencies."
The two papers agree on one point: simple moving-average signals held up better than most other rules. Neither paper shows that charts reliably predict prices.
Using it with care
- Test a rule on data it was not built on before trusting it. A rule fitted to the past can fail on new data.
- Include fees, slippage and funding costs. Several classic results get weaker once costs are counted.
- Treat signals as probabilities, not certainties, and decide your risk limits before you open a position.
- Combine technical analysis with fundamental and news analysis rather than relying on one method.
Limitations: the academic findings cover specific markets and periods, and past patterns may not repeat. This article is general information, not trading or investment advice.
