Tradingview Lightweight Charts Lightweight Charts

Lightweight Charts skill

Works the same whether the project is a downstream npm consumer app or an upstream lightweight-charts source checkout. Detect which one you are in (below) and resolve every API name from whatever typings are locally available.

Source lookup order

Do not assume you are inside the upstream source repository.

1. In a consumer app, inspect the installed package first:
   • node_modules/lightweight-charts/package.json for the actual version.
   • node_modules/lightweight-charts/dist/typings.d.ts for the API surface.
2. In the upstream repo, inspect dist/typings.d.ts first, then src/ if generated output is unavailable.
3. Use official docs/examples as supporting evidence, but local typings win when they disagree.

Verify before answering (copy-paste):

node -p "require('./node_modules/lightweight-charts/package.json').version"
rg -n "createSeriesMarkers|addSeries|IPrimitivePaneRenderer|createImageWatermark|createOptionsChart" node_modules/lightweight-charts/dist/typings.d.ts
# upstream checkout instead of a consumer app:
rg -n "createSeriesMarkers|IPrimitivePaneRenderer|createImageWatermark|createOptionsChart" dist/typings.d.ts src

If the relevant file is unavailable, say what could not be verified. Do not invent option names, methods, exports, or wrapper behavior.

If the user's code imports from lightweight_charts import ... (underscore), uses pandas/asyncio, or talks about a Python window wrapper, they are probably using the third-party Python wrapper rather than the TypeScript package. Say that clearly before giving core lightweight-charts v5 JavaScript advice.

Mental model

Eight layers, in dependency order. Most bugs come from confusing one for another:

1. Chart — createChart(container, options) returns an IChartApi. One chart per container element.
2. Series — added via chart.addSeries(SeriesType, options, paneIndex?). Each call returns an ISeriesApi. v5 requires importing the series type explicitly.
3. Scales — chart.timeScale() (ITimeScaleApi), chart.priceScale(id) (IPriceScaleApi). Govern range, ticks, autoscale, visibility.
4. Data model — arrays of { time, … } points. Time = UTCTimestamp | BusinessDay | string. UTCTimestamp is seconds, not milliseconds. Times must be unique and ascending per series.
5. Interaction — chart.subscribeCrosshairMove, chart.subscribeClick, timeScale.subscribeVisibleLogicalRangeChange, etc.
6. Layout — chart.panes(), chart.addPane(). Adding a series with an out-of-range paneIndex auto-creates the pane.
7. Extension — pane primitives, series primitives, custom series, watermarks, custom renderers (canvas via fancy-canvas).
8. Wrappers — React/Vue/Web Components/iOS/Android. Lifecycle is the wrapper's responsibility; the core library is framework-agnostic.

v5 essentials

The conventions that catch people copying older snippets:

import { createChart, CandlestickSeries, LineSeries } from 'lightweight-charts';

const chart = createChart(container, { /* ChartOptions */ });
const candles = chart.addSeries(CandlestickSeries, { /* options */ });
candles.setData([
    { time: 1700000000, open: 1, high: 2, low: 0.5, close: 1.5 },
]);

Key v5 changes versus older snippets:

• chart.addSeries(SeriesType, options, paneIndex?) replaces addLineSeries / addCandlestickSeries / …
• createSeriesMarkers(series, markers) replaces series.setMarkers(...). The returned primitive owns .setMarkers(...) / .markers().
• createTextWatermark(pane, options) and createImageWatermark(pane, imageUrl, options) replace the watermark chart option.
• Series types are tree-shaken — importing them is required for ESM.

If a snippet calls addLineSeries or series.setMarkers, it's v4 or older. Confirm the user's version before forwarding it.

When the task targets v5, show the v5 API directly — do not hedge by mixing in v4 syntax.

Triage (problem → API → avoid)

v4 → v5 replacement matrix

Stale copy-pasted snippets are the most common failure. Map them directly:

Routing map (problem → where to look)

Use local typings first. The repo paths below exist only in the upstream lightweight-charts checkout; in a consumer app, translate them to the installed package typings or official docs.

If these repo paths do not exist in the user's project, do not ask them to create or clone the source tree just to answer a normal usage question. Use node_modules/lightweight-charts/dist/typings.d.ts and concise v5 examples.

Foot-guns (assertions, not categories)

Time

• UTCTimestamp is in seconds. Use Math.floor(Date.now() / 1000), never Date.now().
• There is no built-in timeScale.timezone option. For display-only timezone conversion, use timeScale.tickMarkFormatter for axis labels and localization.timeFormatter (or your own tooltip formatter) for crosshair/hover labels. Use Intl.DateTimeFormat(..., { timeZone }), Luxon, or date-fns-tz for DST-safe IANA timezone formatting.
• Pre-shifting timestamps changes chart semantics. It can make ticks line up in a chosen timezone, but it also changes where bars sit on the UTC time scale. Prefer formatter-based display unless the user intentionally wants shifted data.
• Series data must be strictly ascending by time, with unique values. Duplicates replace silently; out-of-order points are dropped or throw.
• One time format per series. BusinessDay ('YYYY-MM-DD' string or { year, month, day }) or UTCTimestamp, but do not mix within a single series.
• Whitespace points carry time only (no value / open / …). Use them to create gaps or align overlays; do not set values to null.
• setData replaces the dataset and can reset the visible range. update(point) appends if time > last and replaces if time === last. update(point, true) can update an existing historical point, but it is slower and does not insert missing old points. To prepend older history, build a new array and call setData again — there is no prepend method; save and restore the visible logical range to avoid the chart snapping.
• timeScale.setVisibleLogicalRange is logical (bar indices), not time-based. Use setVisibleRange for time bounds, or fitContent() to reset.

Scales and panes

• Two visible price scales requires both assigning series with priceScaleId: 'right' / 'left' and making the left scale visible. Use chart.applyOptions({ leftPriceScale: { visible: true }, rightPriceScale: { visible: true } }) or chart.priceScale('left').applyOptions({ visible: true }).
• Overlay series (priceScaleId: '' or any custom id) get an autoscaled, hidden scale by default. Make it visible explicitly if you want axis labels.
• autoscaleInfoProvider runs on every layout pass. Keep it cheap; do not allocate per call.
• addSeries(_, _, paneIndex) creates the pane if paneIndex is one past the current count. Pane order is creation order; sizes are controlled via chart.panes()[i].setHeight(px).
• Keep pane references if panes can move. const mainPane = chart.panes()[0] remains the same IPaneApi; mainPane.paneIndex() gives its current index after swapPanes or moves.
• Sync separate charts with public time-scale events. Use subscribeVisibleLogicalRangeChange and setVisibleLogicalRange; avoid _private__chartWidget, paneWidgets, or DOM offset hacks.
• Price-scale width is readable, not directly settable. Use series.priceScale().width() or chart.priceScale(id).width() to measure when aligning adjacent containers.

Markers and interaction

• v5 markers are a primitive. const m = createSeriesMarkers(series, [...]); m.setMarkers([...]). series.setMarkers does not exist on ISeriesApi in v5.
• Marker time must match an existing data point's time for that series. Misaligned markers are dropped silently.
• subscribeCrosshairMove fires with param.time === undefined outside the data range. Always null-check before reading.
• param.seriesData.get(series) returns undefined between bars or before the first bar.
• Click/crosshair coordinates need API conversion. Use series.coordinateToPrice(y) / series.priceToCoordinate(price) and chart.timeScale().coordinateToLogical(x) / logicalToCoordinate(logical) / timeToCoordinate(time); do not infer price/time from canvas DOM geometry.

Plugins and custom rendering

• Start plugins from the scaffold/examples. Prefer npm create lwc-plugin@latest and plugin-examples/ before writing a primitive from scratch.
• Interactive drawing tools need two parts: a primitive/custom series for rendering, and chart interaction handlers (subscribeClick, subscribeCrosshairMove, drag state) that convert coordinates with public APIs.
• CanvasRenderingTarget2D is imported from fancy-canvas, not from lightweight-charts.
• Use target.useBitmapCoordinateSpace(scope => …) for pixel-aligned strokes; use target.useMediaCoordinateSpace(scope => …) for CSS-pixel logic. Mixing them causes blur or sub-pixel jitter on HiDPI displays.
• IPanePrimitive views render across the whole pane; ISeriesPrimitive views are clipped to the series. Pick the right base.
• The renderer interface is IPrimitivePaneRenderer, not IPanePrimitivePaneRenderer. The view is IPanePrimitivePaneView and its renderer() returns IPrimitivePaneRenderer | null — the names are deliberately asymmetric. There is no IPanePrimitivePaneRenderer export; grep dist/typings.d.ts if unsure.
• ICustomSeriesPaneView is heavier than a primitive — only use it when you need a true series API (data, scales, autoscale). For decorations, use a primitive.

Wrappers (React/Vue/etc.)

• Create the chart once in useEffect / onMounted and destroy it in cleanup with chart.remove(). Recreating on every render duplicates DOM and leaks listeners.
• Resize can be automatic in v5 with autoSize: true when ResizeObserver is available. If you need manual control, subscribe a ResizeObserver to the container and call chart.resize(width, height) (or applyOptions({ width, height })).
• Don't drive setData from props on every render. Use series.update(...) for incremental changes; only call setData when the dataset truly replaces.
• Next.js/SSR must be client-only. Put chart code in a 'use client' component, create it in useEffect, and import that component with next/dynamic(..., { ssr: false }) from server-rendered pages when needed.
• Plain HTML is not npm resolution. The standalone .js build exposes window.LightweightCharts; ESM in the browser must import an actual .mjs URL or use an import map. import { createChart } from 'lightweight-charts' only works when a bundler/runtime resolves the package name.

Canonical recipes

Realtime updates

import { createChart, CandlestickSeries } from 'lightweight-charts';

const chart = createChart(container, {});
const series = chart.addSeries(CandlestickSeries, {});
series.setData(initialBars); // full history once

ws.onmessage = ev => {
    const bar = JSON.parse(ev.data); // { time, open, high, low, close }
    series.update(bar); // appends if time > last; replaces if time === last
};

update() only handles the latest bar. To insert in the middle or prepend, rebuild the array and call setData.

Two price scales (left + right)

const right = chart.addSeries(LineSeries, { priceScaleId: 'right' });
const left  = chart.addSeries(LineSeries, { priceScaleId: 'left' });
chart.applyOptions({
    leftPriceScale: { visible: true },
    rightPriceScale: { visible: true },
});

Volume in its own pane

const price  = chart.addSeries(CandlestickSeries, {}, 0);
const volume = chart.addSeries(
    HistogramSeries,
    { priceFormat: { type: 'volume' } },
    1,
);
chart.panes()[1].setHeight(120);

Per-bar / per-point colors

Color individual bars by putting the color on the data point, not via a separate API:

// Candlesticks: per-point color / borderColor / wickColor override series defaults
candles.setData([
    { time: 1700000000, open: 1, high: 2, low: 0.5, close: 1.5,
      color: '#26a69a', borderColor: '#26a69a', wickColor: '#26a69a' },
    { time: 1700003600, open: 1.5, high: 1.8, low: 1.1, close: 1.2,
      color: '#ef5350', borderColor: '#ef5350', wickColor: '#ef5350' },
]);

// Histogram (e.g. up/down volume): per-point color
volume.setData([
    { time: 1700000000, value: 100, color: 'rgba(38,166,154,0.5)' },
    { time: 1700003600, value: 80, color: 'rgba(239,83,80,0.5)' },
]);

Line/Area points also take a per-point color. An omitted point color falls back to the series option. Do not create extra series or use markers to recolor bars.

Series markers

import { createSeriesMarkers } from 'lightweight-charts';

const markers = createSeriesMarkers(candles, [
    { time: 1700000000, position: 'aboveBar', color: '#e53935', shape: 'arrowDown', text: 'Sell' },
]);

markers.setMarkers([
    { time: 1700003600, position: 'belowBar', color: '#43a047', shape: 'arrowUp', text: 'Buy' },
]);

color is required on every marker (position, shape, time too); text, size, id, price are optional. Do not put marker objects into setData() points. Do not call series.setMarkers(...) in v5.

Timezone display

const zone = 'America/New_York';
const formatTime = (timestamp: number) =>
    new Intl.DateTimeFormat('en-US', {
        timeZone: zone,
        hour: '2-digit',
        minute: '2-digit',
        second: '2-digit',
        hour12: false,
    }).format(new Date(timestamp * 1000));

chart.applyOptions({
    timeScale: {
        timeVisible: true,
        secondsVisible: true,
        tickMarkFormatter: (time) =>
            typeof time === 'number' ? formatTime(time) : String(time),
    },
    localization: {
        timeFormatter: (time) =>
            typeof time === 'number' ? formatTime(time) : String(time),
    },
});

Do not invent timeScale: { timezone: 'America/New_York' }. If the user needs exchange-session boundaries to generate in a local timezone, explain the trade-off of shifting source timestamps separately from display formatting.

Text watermark

import { createTextWatermark } from 'lightweight-charts';

createTextWatermark(chart.panes()[0], {
    horzAlign: 'center',
    vertAlign: 'center',
    lines: [{ text: 'DEMO', color: 'rgba(0, 0, 0, 0.25)', fontSize: 48 }],
});

The old watermark chart option is not the v5 API.

Whitespace gaps

line.setData([
    { time: '2024-01-01', value: 10 },
    { time: '2024-01-02' }, // whitespace point: time only
    { time: '2024-01-03', value: 12 },
]);

Do not use value: null, undefined, or NaN to create gaps.

Sync separate charts or panes

let syncing = false;

priceChart.timeScale().subscribeVisibleLogicalRangeChange((range) => {
    if (!range || syncing) return;
    syncing = true;
    indicatorChart.timeScale().setVisibleLogicalRange(range);
    syncing = false;
});

indicatorChart.timeScale().subscribeVisibleLogicalRangeChange((range) => {
    if (!range || syncing) return;
    syncing = true;
    priceChart.timeScale().setVisibleLogicalRange(range);
    syncing = false;
});

For v5 panes inside one chart, prefer chart.addSeries(SeriesType, options, paneIndex) and chart.panes()[i].setHeight(px). If the original pane can move, keep the IPaneApi reference and call pane.paneIndex() instead of assuming index 0 forever.

Coordinate conversion

chart.subscribeClick((param) => {
    if (!param.point || param.time === undefined) return;

    const price = series.coordinateToPrice(param.point.y);
    const logical = chart.timeScale().coordinateToLogical(param.point.x);
    if (price === null || logical === null) return;

    // Store { time: param.time, price } or convert logical back later:
    const x = chart.timeScale().logicalToCoordinate(logical);
    const y = series.priceToCoordinate(price);
});

Use the public conversion APIs for drawing tools and annotations. DOM canvas coordinates alone cannot tell you the chart price/time.

Next.js client-only component

'use client';

import { createChart, LineSeries } from 'lightweight-charts';
import type { ISeriesApi, LineData, Time } from 'lightweight-charts';
import { useEffect, useRef } from 'react';

export function LwcChart({ data }: { data: LineData<Time>[] }) {
    const ref = useRef<HTMLDivElement | null>(null);
    const seriesRef = useRef<ISeriesApi<'Line'> | null>(null);

    useEffect(() => {
        if (!ref.current) return;
        const chart = createChart(ref.current, { autoSize: true });
        const series = chart.addSeries(LineSeries);
        seriesRef.current = series;
        return () => {
            seriesRef.current = null;
            chart.remove();
        };
    }, []);

    useEffect(() => {
        seriesRef.current?.setData(data);
    }, [data]);

    return <div ref={ref} style={{ width: '100%', height: 320 }} />;
}

From a server-rendered Next.js page, import the component with:

import dynamic from 'next/dynamic';

const LwcChart = dynamic(() => import('./LwcChart').then((m) => m.LwcChart), {
    ssr: false,
});

Plain HTML loading

Use the standalone IIFE build when there is no bundler:

<div id="chart" style="width: 600px; height: 300px"></div>
<script src="https://unpkg.com/lightweight-charts/dist/lightweight-charts.standalone.production.js"></script>
<script>
  const chart = LightweightCharts.createChart(document.getElementById('chart'));
  const series = chart.addSeries(LightweightCharts.LineSeries);
</script>

Choose the loading shape by environment:

• Bundler (Vite/webpack/Next): bare import { createChart } from 'lightweight-charts'. The package exports resolve the right ESM/CJS build.
• Browser <script> (no bundler): the standalone IIFE build above exposes the LightweightCharts global; there is no bare-specifier resolution.
• Browser native ESM (<script type="module">): import from a concrete .mjs URL (e.g. the standalone .mjs) or add an import map; a bare 'lightweight-charts' specifier will not resolve.
• SSR frameworks: load client-only (see the Next.js recipe above) — never import at module top level in a server component.

Plugin starter

npm create lwc-plugin@latest

Use the generated project and plugin-examples/ as the base. Decide the extension type first:

• Primitive (IPanePrimitive / ISeriesPrimitive, attached via attachPrimitive) — for drawings, overlays, labels, bands, annotations, hit-testing. This is the default; reach for it unless you need real series behavior.
• Custom series (ICustomSeriesPaneView) — only when the extension needs its own data, autoscale, or price/time-scale integration. It is heavier; do not use it for decorations.

For draggable drawings, attach a primitive with series.attachPrimitive(...) or pane.attachPrimitive(...), then manage interaction with subscribeClick / subscribeCrosshairMove and coordinate conversion APIs. Import renderer-only CanvasRenderingTarget2D from fancy-canvas. The renderer interface is IPrimitivePaneRenderer (not IPanePrimitivePaneRenderer).

Non-financial data

import { createOptionsChart } from 'lightweight-charts';

const chart = createOptionsChart(container);

Decide first whether the horizontal axis is truly non-time:

• True custom horizontal scale (option chains, sensor sweeps by a non-time key): use createOptionsChart for a numeric x-axis, or createChartEx with a custom IHorzScaleBehavior (verify both in local typings).
• Just non-financial y units (temperature, load) on an ordinary time/category axis: keep the normal createChart series model and set priceFormat / localization.priceFormatter so labels match the domain.
• Categorical x that is really ordered: mapping categories to ascending BusinessDay strings is simpler than a custom scale — but say that the axis is then date-like, not truly categorical.

Custom price formatter

chart.applyOptions({
    localization: {
        priceFormatter: (p: number) => `$${p.toFixed(2)}`,
    },
});

For per-series formatting, set priceFormat: { type: 'custom', formatter, minMove } in the series options — the chart-level priceFormatter is the fallback.

Minimal pane primitive

import type {
    IPanePrimitive,
    IPanePrimitivePaneView,
    IPrimitivePaneRenderer,
    Time,
} from 'lightweight-charts';
import type { CanvasRenderingTarget2D } from 'fancy-canvas';

class Overlay implements IPanePrimitive<Time> {
    paneViews(): IPanePrimitivePaneView[] { return [new View()]; }
}
class View implements IPanePrimitivePaneView {
    renderer(): IPrimitivePaneRenderer { return new Renderer(); }
}
class Renderer implements IPrimitivePaneRenderer {
    draw(target: CanvasRenderingTarget2D): void {
        target.useMediaCoordinateSpace(({ context, mediaSize }) => {
            context.fillStyle = 'rgba(0,0,0,0.05)';
            context.fillRect(0, 0, mediaSize.width, mediaSize.height);
        });
    }
}

chart.panes()[0].attachPrimitive(new Overlay());

Working examples live under plugin-examples/src/plugins/ — copy structure from there before writing one from scratch.

Worked example

"My chart snaps to the right edge every time new data arrives."

1. Layers involved: data model (4) and time scale (3).
2. Suspect: the code calls series.setData(...) on every tick instead of series.update(...). setData resets the visible range.
3. Verify: ask whether the call path is setData or update. If setData, ask whether the visible logical range is captured before and restored after — the v4-era workaround.
4. Fix: switch to series.update(bar) for incremental ticks; reserve setData for full-history replacement; only use chart.timeScale().scrollToRealTime() if the user explicitly wants follow-the-latest behavior.
5. References: website/tutorials/demos/realtime-updates.js, website/docs/time-scale.md.

Code-generation rules

• Default to v5. Assume v5 APIs unless the user's installed package or prompt pins an older version. If they're on v4, explain the v5 migration point instead of mixing syntaxes.
• Use real option names. In consumer apps, grep node_modules/lightweight-charts/dist/typings.d.ts; in the upstream repo, grep dist/typings.d.ts or src/api/options/. Many similarly-named options exist across chart/series/scale — confirm which level owns the option.
• Minimal snippets. One feature per code block. Combining markers + watermark + custom pane primitive in one snippet hides which API does what.
• Match the user's framework. A React user wants the useEffect lifecycle; a vanilla user does not.
• Don't invent. If an API name does not appear in the installed typings or upstream source, it does not exist.

Answer contract

When answering a lightweight-charts question:

1. Name the relevant v5 API.
2. Show one minimal snippet, not a mega-demo.
3. Call out the main foot-gun for that task.
4. Say what local source was checked (version/typings), or state that it could not be verified.

Source: tradingview/lightweight-charts — distributed by TomeVault.