ClearBreaks
Sign inWatchlist

How ClearBreaks works

One job

ClearBreaks looks for classical chart patterns on the 4H, 1D and 1W charts of about 1,250 crypto assets (every coin with a liquid USDT or USD spot market on the exchanges below), drawn the way the classical charting literature draws them (Edwards & Magee, Schabacker, and practitioners such as Peter Brandt and Aksel Kibar): horizontal boundaries first, several tests of each boundary, weeks to months in the making, and a decisive close beyond the boundary: above it for a breakout, below it for a breakdown. Each pattern is shown only on the timeframe where it formed, drawn with its own boundary lines and nothing else.

Patterns

38 patterns from the classical literature (Edwards & Magee, Bulkowski's Encyclopedia of Chart Patterns, Brandt, Kibar). Each has its own page in the pattern library with the exact rules:

Sloping lines are held to the same standard as horizontal ones: each line runs through swing highs or lows with several distinct touches within about 0.4 ATR, no swing pokes through it, and no candle closes outside it before the break. When two patterns fit the same candle in the same direction, only one is listed, in this order: rectangles and right-angled triangles, head and shoulders, triple and double tops and bottoms, cups, flags and pennants, symmetrical triangles, wedges, diamonds, rounding patterns, broadening formations, channels, V-turns, measured moves.

Rules

Breakout and breakdown

We use two words for what price does at a pattern's boundary. Charts carry only the pattern's lines (boundaries, necklines, rims, poles and legs); the words appear on cards and lists.

Many breaks fail: price breaks out or breaks down, then closes back inside the pattern soon after (a false break). We remove such a pattern from the list when a candle closes back inside its boundary within 6 candles on 4H, 5 on 1D or 3 on 1W; a wick back to the line is a normal retest and does not count.

Failed breaks are common. In Thomas Bulkowski's US-stock data, 16% to 51% of these patterns (depending on pattern and direction) "bust", meaning price goes on to close beyond the opposite side of the pattern, and downward breaks bust more often than upward ones. Kibar records about 40% of his breakouts as failures. Classical traders cut the trade when a break fails. Some give the pattern one more chance, as Peter Brandt does; others trade the failure in the other direction. Sources and figures: Bulkowski on busted patterns, Kibar's breakout types.

The lists

Breakouts (bullish breakouts and bearish breakdowns) are patterns that completed recently: within about 5 days on 4H, 20 days on 1D and 8 weeks on 1W. A break that closes back inside its boundary soon after is removed. The watchlist shows mature patterns (length and tests met) that have not broken yet, where the latest close is within 3% or 1 ATR (at most 6%) of the level, sorted by distance. Each card says whether the pattern is heading for a breakout (near the upper boundary) or a breakdown (near the lower one). When a watchlist pattern breaks, it moves to Breakouts & Breakdowns. If price closes through the other side, or the pattern stops qualifying, it drops off.

Honest limits

Classical breakouts and breakdowns are rare by design, so the lists are often short. Rules like these also fire on random price series of the same volatility at a similar rate, so a pattern on its own is no evidence of what comes next; that is why many classical chartists manage the trade, not the prediction. The original five patterns were visually graded (about 61% clearly textbook); the patterns added later use strict rules but have not been graded the same way, and sloping-line patterns leave more room for judgment than horizontal ones.

Data and timing

Candles come from public market data of OKX, Kraken, Coinbase Exchange, Gate, KuCoin, MEXC, HTX and BingX, one exchange pair per coin and timeframe. 4H and 1D use the first exchange in that order that lists the coin with at least $10,000 of daily volume at a price matching CoinGecko's (within 8%); coins listed only on MEXC need $50,000. 1W uses the exchange with the longest weekly history. Stablecoins, wrapped and staked tokens and tokenized funds are left out. HTX and BingX daily and weekly candles open at 16:00 UTC (their exchange day).

A timeframe is scanned only once the coin has enough history for the trend filter: 200 candles on 4H and 1D, 104 weeks (two years) on 1W. Younger coins show the chart without patterns on that timeframe.

Each series is re-scanned shortly after each candle closes (usually within minutes; after the daily close at 00:00 UTC, when every coin's 4H and 1D candles close together, within about 15 minutes); only closed candles are analysed. The forming candle on a coin chart is display-only.