mplfinance vs mplchart¶
mplfinance is the standard matplotlib finance extension, with a lineage going back to matplotlib itself: it descends from the matplotlib.finance module that once shipped inside the library, was spun off as mpl_finance when deprecated out of core, and was then rewritten around the modern mpf.plot API and relaunched as mplfinance in 2019 — maintained under the matplotlib GitHub organization ever since. It is mature, feature-rich, widely deployed, and the natural default for financial plotting with matplotlib.
mplchart is an experimental library with a different design center: declarative chart composition with built-in indicators, on pandas or polars data. Both render classic financial charts with matplotlib, so they embed anywhere matplotlib does — scripts, notebooks, GUI apps, saved images.
This page is a factual comparison, with the same chart built in each. mplchart's interface is still evolving; mplfinance is the stable, established choice.
import mplfinance as mpf
from mplchart.chart import Chart
from mplchart.primitives import Candlesticks, HeikinAshi, Renko, Volume
from mplchart.indicators import SMA
from mplchart.samples import sample_prices
prices = sample_prices() # lower-case columns, full history
data = prices.tail(250).rename(columns=str.capitalize) # mplfinance wants Capitalized columns
prices.tail(3)
| open | high | low | close | volume | |
|---|---|---|---|---|---|
| date | |||||
| 2025-11-13 | 274.109985 | 276.700012 | 272.089996 | 272.950012 | 49602800 |
| 2025-11-14 | 271.049988 | 275.959991 | 269.600006 | 272.410004 | 47399300 |
| 2025-11-17 | 268.815002 | 270.489990 | 265.730011 | 267.459991 | 42862157 |
The same chart in both¶
Candlesticks, volume, and two moving averages — first mplfinance, then mplchart.
mpf.plot(data, type="candle", mav=(50, 200), volume=True, figsize=(10, 6))
Chart(prices, title="mplchart", max_bars=250).plot(
Candlesticks(), Volume(), SMA(50), SMA(200),
).show()
Same ingredients, two API styles: mplfinance is option-driven — one mpf.plot(...) call configured through keyword arguments (type=, mav=, volume=). mplchart is composition-driven — a chart is a list of primitives and indicators, extended by adding items or panes.
One difference is visible above: on the mplfinance chart the 200-bar average only covers the last 50 bars. That is structural, not a tuning issue — see The data is the view below.
Data conventions¶
| mplfinance | mplchart | |
|---|---|---|
| Backend | pandas | pandas or polars |
| Columns | Open, High, Low, Close, Volume (capitalized) |
open, high, low, close, volume (lower case) |
| Dates | DatetimeIndex |
datetime index (pandas) or date/datetime column |
| Mismatched names | rename yourself | Chart(..., normalize=True) |
The polars support is a real differentiator if your pipeline is polars-based: mplchart consumes polars frames natively (with polars expressions instead of indicators — see the backends reference page), no conversion to pandas needed.
The data is the view¶
In mplfinance, the data you pass is the view: mpf.plot shows exactly the frame it receives, and any computation happens on that window. Moving averages inherit this — mav=(50, 200) computes over the plotted slice, so on a 250-bar window the 200-bar average only exists for the last 50 bars, as seen in the first chart. The warm-up truncation is structural: there is no data outside the view to compute from. And beyond mav=/ema=, everything enters as pre-calculated data through mpf.make_addplot(...) — the user computes the indicator, slices it to the plotted window, and keeps it aligned by hand.
mplchart separates the data from the view: Chart(prices, max_bars=250) keeps the full history, indicators compute on the fly over that whole range, and only then are the results sliced to the visible window. A 200-bar average is fully formed at the left edge of a 250-bar chart, and resizing the window never changes indicator values.
Non-trading gaps¶
Both libraries can draw price history without weekend and holiday gaps, via different mechanisms. mplfinance controls it with a flag — show_nontrading=False is the default, and passing True restores calendar spacing. mplchart eliminates gaps by construction: bars are positioned by row number and the date axis maps tick labels back to dates, so gapless is the default; Chart(..., raw_dates=True) opts into raw calendar spacing instead.
Chart types¶
Both libraries now cover the box-based chart types. mplfinance renders them through type=: candle, ohlc, line, hollow_and_filled, renko, pnf. mplchart composes primitives: Candlesticks (including hollow modes), OHLC, line/area/bar plots, HeikinAshi, Renko, and PointFigure. Heikin-Ashi is mplchart-only; mplfinance draws hollow candles but not averaged ones.
The deeper difference is the design. mplfinance transforms inside the plot call — mav= is specially wired to compute over renko bricks, but that is where indicator support on box charts ends. mplchart's Renko and PointFigure are transform primitives: they rebase the chart onto bricks or columns, and everything after them — any indicator, volume, extra panes — composes in the transformed space with no special wiring. The window (max_bars) counts bricks, and an SMA(20) is an average of the last 20 bricks, time-free by construction.
mpf.plot(data, type="renko", renko_params=dict(brick_size=5), figsize=(10, 6))
Chart(prices, title="mplchart renko", max_bars=120).plot(
Renko(brick_size=5), SMA(20), Volume(),
).show()
Indicators¶
mplfinance ships moving averages (mav=, ema=) built in; anything else you compute yourself and attach with mpf.make_addplot(...), which gives full control at the cost of manual wiring (computed and window-aligned by hand — see The data is the view). mplchart ships a library of indicators (RSI, MACD, BBANDS, ATR-family, and more) that slot directly into the chart with automatic panes, plus polars expression equivalents, condition primitives (Stripes, Markers), and support for custom callables. If your charts are indicator-heavy, this is where the declarative style pays off.
Styling¶
Both ship a style system with named presets. mplfinance has a few dozen (yahoo, charles, binance, blueskies, ...) plus make_mpf_style for custom ones. mplchart resolves a single style= argument through several layers: shipped styles by name (nightclouds, chartist, modern, ...), any matplotlib stylesheet name (ggplot, dark_background, ...), provider-prefixed names — style="mpf:yahoo" renders mplfinance's own styles directly, style="mt:economist" a morethemes theme — or an ad-hoc spec combining rc params with finance-specific settings (candle.up.color, wicks.color, ...). mplchart styles are total: ambient rcParams never leak into a chart, so the same code renders identically everywhere.
Chart(prices, title="nightclouds", max_bars=250, style="nightclouds").plot(
Candlesticks(), Volume(), SMA(50), SMA(200),
).show()
