Plotly error bands: shaded confidence interval (tested)

Plotly has no error-band option: add the upper bound as an invisible line, then the lower bound with fill="tonexty" (it fills to the trace right before it), then the mean line. Or draw one closed polygon with fill="toself".

1. Upper line, lower line with fill="tonexty", then the mean

b1.py:

import numpy as np, pandas as pd, plotly.graph_objects as go
x = np.arange(5)
mean = np.array([2.0, 3.0, 2.5, 4.0, 3.5])
sd = np.array([0.4, 0.5, 0.3, 0.6, 0.5])

fig = go.Figure([
    go.Scatter(x=x, y=mean + sd, mode="lines", line=dict(width=0), showlegend=False),
    go.Scatter(x=x, y=mean - sd, mode="lines", line=dict(width=0), fill="tonexty",
               fillcolor="rgba(31,119,180,0.2)", name="±1 sd"),
    go.Scatter(x=x, y=mean, mode="lines", line=dict(color="rgb(31,119,180)"), name="mean"),
])
for t in fig.data:
    print(t.name, "| fill:", t.fill, "| y:", np.round(t.y, 2).tolist())
fig.write_html("band.html")

# Wrong order: fill="tonexty" on the FIRST trace fills down to y=0 instead
bad = go.Figure([go.Scatter(x=x, y=mean + sd, fill="tonexty"),
                 go.Scatter(x=x, y=mean - sd)])
print("bad first trace fill:", bad.data[0].fill, "(no trace before it)")

# Single trace alternative: closed polygon with fill="toself"
poly = go.Scatter(x=np.concatenate([x, x[::-1]]),
                  y=np.concatenate([mean + sd, (mean - sd)[::-1]]),
                  fill="toself", line=dict(width=0), name="band")
print("toself x:", poly.x.tolist())
print("toself y:", np.round(poly.y, 2).tolist())

Output (plotly 7.1.0, numpy 2.5.4, Python 3.13.5):

None | fill: None | y: [2.4, 3.5, 2.8, 4.6, 4.0]
±1 sd | fill: tonexty | y: [1.6, 2.5, 2.2, 3.4, 3.0]
mean | fill: None | y: [2.0, 3.0, 2.5, 4.0, 3.5]
bad first trace fill: tonexty (no trace before it)
toself x: [0, 1, 2, 3, 4, 4, 3, 2, 1, 0]
toself y: [2.4, 3.5, 2.8, 4.6, 4.0, 3.0, 3.4, 2.2, 2.5, 1.6]

• Plotly has no error_band argument. Draw the upper bound as an invisible line (line=dict(width=0), showlegend=False), then the lower bound with fill="tonexty": it shades between itself and the trace right before it. Add the mean line last so it sits on top of the band.
• Order matters. Plotly's own docstring for fill: "if there is no trace before it, they behave like 'tozerox' and 'tozeroy'", so tonexty on the first trace shades all the way down to y=0 instead of to the other bound.
• Use a semi-transparent fillcolor (rgba(..., 0.2)) in the same hue as the mean line.
• Single-trace alternative: fill="toself" on a closed polygon, x forward then reversed, upper bound then the lower bound reversed (printed above). This is handy with many groups because each band is one trace.

2. Which spread to shade

• ±1 standard deviation shows the spread of the data; standard error (sd / sqrt(n)) or a 95% interval (about ±1.96 × se) shows the uncertainty of the mean. Say which one in the legend name.
• For error bars on individual points instead of a band, see Plotly error bars: error_y. seaborn draws a band automatically in lineplot: seaborn errorbar sd vs ci.

Written with AI assistance. Measured on 2026-10-23 with plotly 7.1.0, numpy 2.5.4 and Python 3.13.5 on Linux. Scripts ran verbatim and the output was pasted unedited. PlainTables is our extension; Plotly is free.