Technical Analysis in Crypto: Charts, Indicators & Patterns

Introduction

Take 22,018 asset-days of bitcoin, ether and six other major tokens. An RSI reading above 50 and a close above the 20-day moving average pointed the same way on 89.8% of them. Those are two of the most widely used indicators in crypto, and most of the time they are one measurement wearing two names. That single number reframes what a technical analysis panel is doing. This guide is a framework rather than a catalogue. What technical analysis assumes, what every indicator is computed from, how the four tool families overlap, and where the method stops being able to help. The per-indicator and per-pattern deep dives already exist and are named throughout. What follows is the structure they hang on, measured on 23,602 daily bars rather than asserted.

What Is Technical Analysis and What Does It Assume?

Technical analysis is the practice of forming a view about price from price itself, plus the volume that accompanied it. Everything else is excluded by design. That exclusion is a choice with consequences, and the consequences follow from three assumptions the method rarely states out loud.

The three assumptions

The first assumption is that price already reflects the information available. Studying the price series is therefore not a poor substitute for studying the asset. The second is that price moves in identifiable structures rather than pure noise, which is what makes trend and pattern language meaningful. The third is that those structures recur often enough to be worth naming. Each assumption is defensible and none is proven. A reader who accepts all three is doing technical analysis in crypto whether or not they use the phrase. A reader who rejects the second has no reason to look at a chart at all.

What technical analysis is not

Technical analysis is not a prediction engine and does not claim to know the future. It is a way of describing the present state of a price series in a compressed vocabulary. It is also not a competitor to token research. A chart says nothing about a protocol's revenue, its unlock schedule or its regulatory exposure. CoinPaprika's guide Fundamental Analysis vs. Technical Analysis: Which One Is Better? sets that boundary out in full. Most disappointment with crypto charts traces back to expecting the method to answer a question it was never built to take.

The vocabulary is broad, but the raw material is remarkably narrow.

Where Does Every Technical Indicator Actually Come From?

Every indicator on a standard crypto chart is a function of five numbers per bar: open, high, low, close and volume. Of ten commonly used tools, five read exactly one of those five columns. The panel is far less independent than its variety suggests.

Five columns, every tool

An OHLCV record is the entire input set, and a companion article covers how those records are built and where they go wrong ↗. Every calculation on the chart begins by reading one or more of those five numbers. Moving averages, RSI, MACD, Bollinger Bands and rate of change all read the close and nothing else. The stochastic oscillator and average true range add the high and low. Candlestick patterns use all four price columns. On-balance volume and VWAP are the only common tools that touch the volume column at all. Nothing in the standard toolkit reads anything that is not in the table below, because nothing else is in the feed.

Data current as of September 2026.

Half the panel reads one column

The consequence is structural rather than a criticism. Five tools drawing on one column cannot be independent evidence about that column. Bollinger Bands make the point unavoidable. The middle band is by definition the 20-period simple moving average, so a chart showing both is drawing one line twice. Other overlaps are looser. RSI over 14 periods correlates with a plain 14-day rate of change at 0.647 across 23,498 asset-days (Binance Spot API, 2026-09-10) . That is a real relationship rather than a rescaling. Knowing which tools share inputs is the first defence against mistaking arithmetic for confirmation. It also explains why adding a fifth close-based overlay rarely changes a conclusion.

Shared inputs still leave room for genuinely different questions.

What Are the Four Families of Technical Tools?

Technical indicators sort cleanly into four families by the question they ask: where is price going, how fast, how wildly, and with how much participation. Choosing a tool starts with choosing a question. Most confusion in practice comes from reaching for the wrong family entirely.

Trend, momentum, volatility, volume

Trend tools smooth the series to say which direction it has been running. Momentum tools measure the speed of recent change and bound it into a range so extremes are visible. Volatility tools describe how widely price is dispersing around its own average. Volume tools ask whether the move carried participation or drifted on thin trade. The four are not ranked, and none of them subsumes another. A momentum reading in a trendless market and a trend reading in a violent one are both weak, and neither is the tool's fault. Each family also fails in a characteristic way. Trend tools lag by construction, since smoothing cannot react before the data arrives. Momentum tools pin to their extremes whenever a move persists, which is exactly when a trader wants a reading. Volatility tools widen after the move rather than before it. Volume tools depend on a column whose reliability varies by venue. Knowing a family's failure mode beats knowing any member's formula.

Which family answers which question

The table below maps each family to its question and to where CoinPaprika covers it in depth. This article deliberately does not restate those catalogues, because they already exist in full and the framework around them is what tends to be missing. The mapping matters more than it looks. A reader who cares about participation has narrowed ten candidate tools to two. A reader who has decided nothing keeps adding overlays without ever eliminating one.

Data current as of September 2026.

Before any of them, there is the chart itself.

How Do You Read a Chart Before Adding a Single Indicator?

A bare price chart carries most of what a beginner needs. Every indicator added on top is a compression of something already visible in the bars. Reading the raw series first is the habit that separates interpretation from decoration, and it is the step most charting tutorials skip.

Price structure before indicators

Start with where price has been rejected and where it has been accepted, which is the substance of support and resistance. CoinPaprika's Technical Analysis 101: How to Find Support and Resistance Zones? covers the method for marking them. Then look at the shape of individual bars. A long wick and a wide body describe different sessions even when they close at the same price. What Are Japanese Candlestick Patterns? holds that vocabulary. None of this requires a single overlay, and all of it survives when the overlays disagree.

What volume adds to price

Volume is the one column that is not a price, which makes it the only genuinely independent axis on a standard chart. A move on heavy volume and the same move on thin volume are different events. The candle bodies match anyway. Crypto complicates this, because reported volume varies enormously in quality across venues. A companion article on real versus wash volume ↗ covers why. The practical version is narrow but useful. Treat volume as a check on participation rather than a signal in its own right, and prefer a venue whose volume you have reason to trust.

Once the overlays go on, they start echoing each other.

Why Do Two Indicators Often Say the Same Thing?

Across 22,018 asset-days, an RSI above 50 and a close above the 20-day moving average pointed the same way on 89.8% of them. That is not two tools confirming each other in any meaningful sense. It is closer to one measurement arriving twice under two different names.

Two stances, one answer

Taking four standard stances daily across eight assets gives a direct reading of how much they overlap. RSI above 50 and price above SMA20 agreed on 89.8% of days. MACD above its signal line and price above SMA20 agreed on 77.5%. RSI above 50 and MACD above signal agreed on 67.8% (Binance Spot API, 2026-09-10) . The pattern is consistent with what the previous section showed: tools reading the same column over similar lookbacks land in the same place most of the time. The ordering of the pairs is the useful part. Agreement falls as the lookbacks diverge, not as the formulas do, which points at the horizon as the thing doing the work.

Data current as of September 2026.

Confluence is often arithmetic

At least three of the four stances pointed the same way on 86.2% of days, and all four aligned on 32.1% (Binance Spot API, 2026-09-10) . A trader who waits for three-of-four agreement is waiting for something that happens most days. That is a weaker filter than it feels like. The bottom row of the table is the more interesting one. MACD and the 50/200 moving average pair disagreed more often than they agreed. One reads a few weeks, the other several months. That disagreement is information: two genuinely different windows are describing the market differently. The 89.8% overlap at the top is not, since it would appear in almost any sample.

Horizon does more work here than the choice of tool.

Does the Timeframe You Pick Change the Answer?

Price above its 20-day average and price above its 200-day average agreed on only 63.3% of days. The same asset, the same tool and the same instant produced opposite readings on more than a third of the sample, purely from the lookback chosen.

The same tool, three lookbacks

Running one indicator at three settings isolates the effect of the window. RSI at 7, 14 and 21 periods is stable: the 14 and 21 settings agreed on 93.3% of days and the 7 and 21 settings on 82.0% across 23,442 asset-days. Moving averages are not. Price above SMA20 and price above SMA50 agreed on 79.8% of days and SMA50 and SMA200 on 72.4%. SMA20 and SMA200 managed just 63.3%, with all three aligned on 57.8% (Binance Spot API, 2026-09-10) . Short oscillators converge because they are all reading recent change. Trend tools diverge because they are answering about genuinely different spans of time.

Choosing a horizon first

The practical order is therefore backwards from how most people learn it. Decide the horizon first, then pick the tools that measure it. Most people pick familiar tools and discover afterwards which horizon those tools describe. A swing trader and a long-term holder looking at one chart are not disagreeing about the asset. They are reading different windows and getting the correct answer for each. Stating the horizon out loud before adding an overlay removes most of the apparent contradiction between technical readings.

Patterns sit outside this arithmetic entirely.

What Do Chart Patterns Add That Indicators Do Not?

Chart patterns describe the shape of price movement across a stretch of bars, which no single-value indicator captures at any setting. That is a real addition to the toolkit. It also makes patterns far harder to test than a threshold rule, and the two facts turn out to be connected.

Patterns as structure, not signal

A head and shoulders, a triangle or a double bottom is a claim about how a sequence of highs and lows relate. Nothing in an RSI reading encodes that relationship, so patterns genuinely carry information the oscillator panel discards. CoinPaprika's Predicting Bullish or Bearish Price Movements With Classic Chart Patterns holds the full catalogue, and What Are Japanese Candlestick Patterns? covers the single-bar and two-bar formations. This article's contribution is only to place them. Patterns are structural descriptions rather than computed values, and they sit in a different part of the toolkit than anything in the table above.

Why patterns resist testing

A pattern is recognised rather than calculated. Two analysts can look at the same chart and disagree about whether one is present. That subjectivity is exactly what makes pattern claims difficult to evaluate honestly. An indicator crossover has an unambiguous timestamp; a triangle has a drawing. Nothing here says patterns do not work. The evidence for them is simply harder to assemble than the evidence for a threshold rule. A reader who wants the measured version of that question will find it in the companion article on which indicators actually survive testing ↗.

Every one of these tools has conditions it needs.

Where Does Technical Analysis Stop Working?

Technical analysis needs a liquid, continuously traded market whose price series reflects real transactions at prices someone could have hit. Crypto supplies that for a small number of assets and supplies something considerably worse for most of the rest, which is the single largest practical limit on the method.

The conditions TA needs

The method assumes the printed price is a price someone could have traded at. In a deep book that holds. In a thin one it does not. A handful of trades can set a close that no meaningful size could have transacted at. Two companion articles cover how to measure that depth directly before trusting it — one on crypto liquidity ↗, one on reading the order book ↗. Reported volume compounds the problem. Wash trading inflates the one column that is supposed to confirm participation. A chart drawn from a venue with fabricated volume is a chart of fabricated events, and no indicator setting repairs that.

Where the chart goes blind

Even on a healthy market, a price series cannot see a scheduled token unlock, a protocol exploit, an exchange failure or a regulatory decision. Those arrive as gaps rather than as trends, and no lookback anticipates them. The chart is also silent on why a move happened. That matters, because the same 10% rally means different things after an exploit and after a listing. On-chain data answers a different slice of that question; companion articles cover the on-chain metrics themselves ↗ and the analytics platforms that publish them ↗. Recognising when the chart has gone blind is a technical skill in its own right, and it is mostly a matter of knowing which questions the price series was never able to answer.

That leaves the question of how the lenses fit together.

How Does Technical Analysis Fit With Fundamental and On-Chain Analysis?

Three lenses answer three different questions, and the common mistake is asking one of them a question that belongs to another. Sequenced properly they compose into a single view. Stacked carelessly they double count, producing a confidence that the underlying evidence does not support.

Three lenses, three questions

Fundamental analysis asks whether the asset is worth owning. What it does, who uses it, what the supply schedule looks like. On-chain analysis asks what holders are actually doing: flows to exchanges, holding periods, realised profit. Technical analysis asks where price is now relative to its own recent history, and nothing more. Fundamental Analysis vs. Technical Analysis: Which One Is Better? works through the first pair in detail. The useful framing is that the lenses have different subjects rather than competing answers about one subject.

Combining without double counting

The overlap risk is the same one measured in section eight, one level up. Three momentum indicators agreeing is one reading; a fundamental thesis and a technical entry are two genuinely separate inputs. Sequencing helps: settle whether to own the asset first, then use the chart for timing and sizing, not the reverse. Where an on-chain metric and a technical stance conflict, note that they measure different things over different windows. Declaring a winner is the wrong move. A view that requires all three lenses to align will rarely be actionable, for the same arithmetic reason that three-of-four indicator confluence turns out to be common rather than rare.

The routine that follows from all this is short.

How Should a Beginner Build a Technical Analysis Routine?

Four steps cover the working method: state the horizon, read the bare chart, add one tool per question, then check whether the venue's data deserves the analysis at all. Most of what gets added beyond those four steps turns out to be redundant with something already on the screen.

A four-step reading routine

Name the horizon before opening the chart, because that decision determines every setting afterwards. Read price and volume with no overlays, marking the levels where price has repeatedly turned. Then add at most one tool per family. A second tool from the same family mostly repeats the first. Finish by asking whether the venue's depth and volume support the reading at all. That is a data-quality question rather than a charting one. The routine is deliberately smaller than a default charting layout, and the measurements above are the reason.

Data current as of September 2026.

What to skip at the start

Skip multi-indicator dashboards, which mostly measure the close repeatedly at slightly different smoothings and present the repetition as agreement. Skip pattern hunting before support and resistance make sense, since patterns are built from the same highs and lows. Skip the question of which indicator is best until the horizon is settled. The answer changes entirely with it. And skip treating any of this as evidence that the tools produce an edge. That is a separate, measurable question, and a companion article tests it directly ↗ rather than assuming an answer.

FAQ

References / Sources