Financial Charts

Financial Charts

Declarative TradingView-style charts. You never write chart code: you fill a JSON spec (what panes/series to show) and a JSON data file (the series values), inside a fixed HTML template that renders them with TradingView's Lightweight Charts v5 (CDN). The platform serves the files and displays the chart in the Workbench.

Workflow

1. Get the data. Pull OHLCV via the stock-analysis data sub-skill (or whatever source the task provides). Compute indicator series yourself in Python (recipes below).
2. Write the data file to /code/outputs/<slug>.json (create the directory if missing). Format below.
3. Copy the template from this skill directory to /code/outputs/<slug>.html and edit ONLY the <script id="chart-spec" type="application/json"> block: title, subtitle, dataUrl (must be ./<slug>.json), panes, notes, asOf. Do not touch anything else in the file.

mkdir -p /code/outputs
cp ~/.skills/financial-charts/template.html /code/outputs/aapl-backtest.html
# then edit the chart-spec block in it, and write aapl-backtest.json

4. Announce it in your final response exactly as the workbench contract requires:

<rebyte-output>
aapl-backtest.html
</rebyte-output>

Only list the .html (the JSON is fetched by it, same directory).

Spec reference

{
  "title": "AAPL — Momentum Strategy Backtest",
  "subtitle": "Daily bars, SMA(20) overlay, trade markers; volume and RSI(14) panes.",
  "dataUrl": "./aapl-backtest.json",
  "panes": [
    { "stretch": 3, "series": [
        { "type": "candlestick", "data": "ohlc", "title": "AAPL" },
        { "type": "line", "data": "sma20", "title": "SMA 20" },
        { "type": "markers", "data": "trades", "attachTo": "ohlc" } ] },
    { "stretch": 1, "series": [ { "type": "histogram", "data": "volume", "title": "Volume" } ] },
    { "stretch": 1, "series": [ { "type": "line", "data": "rsi", "title": "RSI 14", "color": "#7a5d94" } ] }
  ],
  "notes": ["Assumption lines shown under the chart."],
  "asOf": "2026-07-03"
}

• panes[0] is the main chart; stretch sets relative pane heights.
• series.type: candlestick | line | area | histogram | markers.
• series.data: key into the data JSON.
• markers need attachTo: the data key of a series in the SAME pane.
• color optional — the template has a coherent default palette; only set it when you need a specific meaning (e.g. benchmark grey).
• Typical layouts: price+SMA+markers / volume / RSI; strategy-vs-benchmark as two line series in one pane; equity curve as area with a drawdown histogram pane.

Data file format

One JSON object; each key is a series array sorted ascending by time, no duplicate times. time is "YYYY-MM-DD" for daily bars or a UNIX timestamp in seconds for intraday.

{
  "ohlc":   [ { "time": "2025-07-02", "open": 180.1, "high": 183.2, "low": 179.5, "close": 182.4 } ],
  "sma20":  [ { "time": "2025-07-30", "value": 184.2 } ],
  "volume": [ { "time": "2025-07-02", "value": 52000000, "color": "#cdd9e3" } ],
  "rsi":    [ { "time": "2025-07-22", "value": 61.3 } ],
  "trades": [ { "time": "2025-08-26", "position": "belowBar", "shape": "arrowUp",
                "color": "#1f7a5c", "text": "Buy @ 203.4" } ]
}

• Volume convention: #cdd9e3 on up days, #e3cfcd on down days.
• Markers: position belowBar/aboveBar, shape arrowUp/arrowDown/ circle/square; green #1f7a5c buys, red #b4443c sells.
• Indicator series start later than price (warm-up window) — that is fine, just omit the missing leading rows.

Indicator recipes (pandas)

import pandas as pd
df = pd.DataFrame(bars)  # columns: time, open, high, low, close, volume

df["sma20"] = df["close"].rolling(20).mean()
df["ema12"] = df["close"].ewm(span=12, adjust=False).mean()

# RSI(14), Wilder's smoothing
delta = df["close"].diff()
gain = delta.clip(lower=0).ewm(alpha=1/14, adjust=False).mean()
loss = (-delta.clip(upper=0)).ewm(alpha=1/14, adjust=False).mean()
df["rsi14"] = 100 - 100 / (1 + gain / loss)

# MACD(12,26,9): plot macd+signal as two lines, hist as histogram pane
ema26 = df["close"].ewm(span=26, adjust=False).mean()
df["macd"] = df["ema12"] - ema26
df["signal"] = df["macd"].ewm(span=9, adjust=False).mean()
df["hist"] = df["macd"] - df["signal"]

# Bollinger(20, 2): three line series in the main pane
mid = df["close"].rolling(20).mean(); sd = df["close"].rolling(20).std()
df["bb_up"], df["bb_lo"] = mid + 2 * sd, mid - 2 * sd

def series(df, col):  # -> [{time, value}] dropping NaN warm-up
    s = df[["time", col]].dropna()
    return [{"time": t, "value": round(v, 4)} for t, v in s.values]

Updating a chart

Re-running the analysis = rewrite the .json (and re-announce the same filename). The HTML template stays untouched; a fresh open shows the new data. Never inline data into the HTML.