Research Plot Formatting

• serif typography
• light grids
• muted spines
• consistent line widths
• restrained color palette

Standard Moves

Typography and Axis Style

• Prefer serif fonts such as STIX Two Text, Times New Roman, or DejaVu Serif.
• Remove top and right spines.
• Use light horizontal grid lines only.
• Keep labels explicit and short.
• Format shares as percentages when the x-axis is ownership or exposure.

Titles

• Use one short main title.
• Remove explanatory subtitle text unless it adds real information.
• If multiple panels are present, panel titles should be short and consistent.
• Add a), b), c) markers only when they help cross-reference the text.

Legends

• Prefer one shared legend for multi-panel figures.
• Move the legend outside the data region when it competes with the curves.
• Bottom-of-figure legends usually work better than repeated panel legends.

Panel Layout

• Keep comparable panels the same size.
• Use a single row when width is available and panels are simple.
• Use a 2 + 1 triangular layout when the figure looks too wide or generic, but keep the third panel the same size as the others.
• Avoid stretching one panel across a full row unless it represents a different object.

Data Layer Additions

Use only if empirically justified:

• confidence bands
• shaded support range
• vertical lines for knots or turning points
• labeled optimum markers

Avoid adding explanatory text like “lower is better” inside panels unless the user explicitly wants teaching-style figures.

Notes by Figure Type

Fitted-Line Plots

• Emphasize the curve shape.
• Mark turning points only when they are inside support.
• Shade support lightly, not heavily.

Spline Multi-Panel Plots

• Show 95% confidence bands if available.
• Mark interior knots with thin dashed vertical lines.
• If optima are important, label them directly on each panel.
• Keep panel titles and legend from competing with the data.

Score or “Good Range” Plots

• Highlight the top range with a light band.
• Mark the optimum with a vertical line and point annotation.
• Use a compact note box for the best range when helpful.

Editing Rules

• Edit the source plotting script, not the exported SVG directly.
• Reuse any existing style helpers if they exist.
• Prefer incremental layout edits over rewriting the whole script.
• If the user asks for repeated refinement, change only the requested elements.

Verification

After rerunning:

• confirm the output path was written
• check that titles, legend, and panel sizes match the request
• mention any estimation warnings separately from plotting success

Typical Summary Back to User

Keep the close-out short:

• what visual changes were made
• which file was regenerated
• whether underlying estimation warnings still remain