> ## Documentation Index
> Fetch the complete documentation index at: https://docs.tradionlabs.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Chart Analyzer
> Drop in a chart image and get a verdict with entry, stop, target, and the reasoning behind the call.
Chart Analyzer reads a picture of a chart and returns a one-word verdict, three price levels, and a written case for the call. Any image works — a TradingView screenshot, a capture from your broker, a chart posted in a Discord.
Included on **Starter** and every plan above it, with no monthly cap. See the [plan comparison](/reference/plan-comparison).
## Upload a chart
**Swing** (days to weeks), **Day Trade** (intraday), or **Scalp** (minutes). Levels scale to the mode: a scalp stop sits much closer to entry than a swing stop.
The **Ticker** field anchors the analysis to that symbol, overriding the label printed on the chart. Leave it blank and the model reads the symbol off the image. Fill it in when the symbol is small, abbreviated, or cropped out.
Click the panel to browse, drag a file onto it, or paste from the clipboard — on macOS, `⌘+Shift+4` then `⌘+V`. One image per analysis; candlestick, line, and Heikin Ashi all read fine. Images above 10MB are rejected.
Three stages run — scanning, patterns, trade plan. Most charts come back in under a minute.
## What makes a chart readable
The model is reading pixels, so anything you can't make out clearly, it can't either.
| Include | Why it matters |
| ----------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------- |
| The **ticker**, visible or typed in | Without it, nothing can be pulled for that symbol — no analyst data, no history |
| The **timeframe** label (1H, 4H, D) | There is no field for this. A wrong guess scales every level to the wrong horizon |
| **Enough price history** — 100 bars or so | Support and resistance are prices where buying or selling has repeatedly appeared. A chart zoomed into a dozen candles has none |
| The **full price axis**, uncropped | Cropped axes produce reports describing levels that aren't on your screen |
| Native resolution | Screenshots scaled down before upload lose the candle bodies |
Leave out the clutter. Overlaid indicators, drawing tools, and coloured zones cover the price action the model needs to read. **A clean candlestick chart with volume beats a decorated one.** For indicator values in the analysis, [Lens](/research/lens) computes them from live market data instead.
If the model returns an incomplete response you'll see a **Degraded Analysis** banner. The levels and confidence figures underneath are defaults, not a read of your chart. Recapture and run it again rather than acting on them.
## The verdict card
The card leads with **BUY**, **SELL**, **HOLD**, or **NO TRADE**, then three numbers: **Setup Quality** out of 10, **Risk/Reward**, and **Confidence** as a percentage. Under them sits a headline, a suggested action, the technical breakdown, and up to three key insights.
### In plain English: the card says HOLD
**HOLD means wait.** Not "hold your position", not "we're unsure" — there is no clear direction on this chart right now. The line under the verdict says the same: *Default: No action until conditions change*. Elsewhere in Tradion this verdict is written **WAIT**; the chart card prints HOLD.
**NO TRADE** is a different answer. HOLD is "not yet". NO TRADE is "this chart cannot support a decision at all".
## Trade levels and critical levels
Every verdict carries **entry** (where the setup makes sense), **stop** (where you exit for a loss), **target** (where the move should run out), and **risk/reward** — the gap to the target divided by the gap to the stop. A **Critical Levels** panel adds a **pivot point** (the previous session's high, low, and close averaged), **R1** and **S1** (prices above and below where selling and buying have tended to appear, each with a **break probability**: the estimated chance price pushes through), and **ATR** — the typical distance the symbol travels in one period.
[Reading a verdict](/concepts/reading-a-verdict) explains what to do with all of it, including why a stop tighter than the ATR gets taken out by noise.
When price hasn't reached the zone yet, the panel reads **Conditional Levels**, marked *not actionable — waiting for trigger*, with **Conditional Micro-Plans** beside it: what to do if price breaks up, or down.
## The confidence breakdown
Confidence is a weighted average of five factors, each scored 0–100 with a one-line reason: trend clarity (30%), pattern quality (25%), volume (20%), risk/reward (15%), market context (10%).
### In plain English
Two charts can both read 68% and mean nothing alike. One might be a textbook pattern nobody is trading; the other, five mediocre scores averaged. **Read the five factors, not the headline.** A wide spread between them means the evidence disagrees with itself.
## The rest of the report
The **Price Forecast** panel draws a cone — a fan of possible prices that widens the further out it reaches — ending in bull, base, and bear points, each with a percentage change, a probability estimate, and a reason you can expand.
For listed stocks, **Analyst Consensus** shows the Wall Street Buy / Hold / Sell split, a price target with the implied upside, and recent upgrades and downgrades by firm. A **consensus estimate** is the average of what professional analysts covering the stock publish — a crowd opinion, not a measurement. The panel hides itself when no rating data exists, normal for crypto, foreign exchange pairs, and small listings.
Below that: a market overview, a table of **key invalidation levels** (prices that would prove the idea wrong), three trade scenarios with triggers, a game plan covering entries and exits, and a market structure section ending in **price receipts** — the observations the verdict rests on.
## Analysis history
Every analysis with a detected ticker saves automatically, chart image included. History is the landing view — click any entry to reopen the full report; Delete removes it permanently.
Two things go wrong. A banner reading *"Analysis displayed but could not be saved to history"* means the result on screen is fine but nothing was stored. And entries created before full-report saving reopen with a notice rather than a report; those need re-running.
Chart Analyzer reads the image and nothing else. There is no follow-up question box, no live price check against your levels, and no strategy code to export. To ask questions about a chart in conversation, use [Lens](/research/lens), which keeps the analysis in context.
It produces analysis, not instructions. It does not know what else you hold or what news breaks tomorrow. The decision stays yours.
Entry, stop, target, break probability, and how confidence misleads.
Live charts inside Tradion, backed by real market data.