Line plots with matplotlib

A figure, an Axes, a line

A chart is the quickest way to see that a price series is clean, trending or broken. In matplotlib a Figure is the whole canvas and an Axes is one chart on it, with its lines, labels and legend. The object-oriented way is fig, ax = plt.subplots() followed by calls on ax: ax.plot(x, y) draws a line, ax.set_title, ax.set_xlabel and ax.set_ylabel name the chart. Drawing on an explicit Axes, and returning it, makes a plotting function easy to reuse and to test.

import matplotlib
matplotlib.use("Agg")            # draw without a window (no screen needed)
import matplotlib.pyplot as plt
import pandas as pd

prices = pd.Series([100.0, 102, 101, 105, 107, 106, 110], index=pd.bdate_range("2026-03-02", periods=7))

fig, ax = plt.subplots()
ax.plot(prices.index, prices.values, label="AAPL")
ax.set_title("Closing price")
ax.set_xlabel("Date")
ax.set_ylabel("Price (USD)")

print(ax.get_title(), "|", ax.get_xlabel(), "|", ax.get_ylabel())   # Closing price | Date | Price (USD)
print(len(ax.lines), ax.lines[0].get_ydata().tolist())               # 1 [100.0, 102.0, 101.0, 105.0, 107.0, 106.0, 110.0]

Labels and the legend

With several lines on one chart, each needs a name. Pass label="..." to ax.plot and call ax.legend() once at the end: the legend lists the labels in the order the lines were drawn. A chart without labels, a title and units is a picture, not a result. Lines also have style: linestyle="--" makes a dashed line, handy for an average that should not hide the prices, and color, linewidth and marker refine it further.

ma = prices.rolling(3).mean()
ax.plot(ma.index, ma.values, label="MA3", linestyle="--")
ax.legend()

print([line.get_label() for line in ax.lines])                  # ['AAPL', 'MA3']
print([text.get_text() for text in ax.get_legend().get_texts()])   # ['AAPL', 'MA3']
print(ax.lines[1].get_linestyle(), ax.lines[1].get_ydata().tolist()[:3])   # -- [nan, nan, 101.0]   NaN points are simply not drawn

Reference lines, and charts you cannot see

A horizontal line marks a level such as a strike, a stop or the average: ax.axhline(105, linestyle="--") draws it across the whole chart (axvline does the same vertically). Matplotlib can also draw with no window at all: with the Agg backend, fig.savefig("chart.png") writes the chart to a file, which is how reports and servers produce charts. Because everything you draw stays on the Axes object, a program (or a test) can read it back: lines with ax.lines, their data with get_ydata(), labels with get_label(), the title with get_title().

ax.axhline(105, linestyle=":", label="level")

level_line = ax.lines[-1]
print(list(level_line.get_ydata()), level_line.get_linestyle(), level_line.get_label())   # [105, 105] : level

fig.savefig("prices.png")   # writes the chart to a file; no window, no plt.show() needed
plt.close(fig)              # release the figure when you are done with it

def plot_prices(prices):
    fig, ax = plt.subplots()
    ax.plot(prices.index, prices.values)
    ax.set_title("Closing price")
    return ax                                   # the caller can add lines, save the chart or inspect it

ax = plot_prices(prices)
print(len(ax.lines), ax.get_title())            # 1 Closing price
plt.close("all")

• fig, ax = plt.subplots() then ax.plot(x, y); name the chart with set_title, set_xlabel and set_ylabel.
• Give each line a label= and call ax.legend(); linestyle="--" draws a dashed line; ax.axhline(y) a horizontal one.
• Draw on an Axes and return it: ax.lines, get_ydata() and get_label() read the chart back; use savefig and plt.close instead of show in programs.

Next up: candlestick charts, drawing open, high, low and close for each day.

Lab · what's graded

• Draw the price line with plot_prices25 pts
• Compare two series on one chart with plot_comparison25 pts
• Add a dashed moving average with plot_with_average25 pts
• Mark a price level with plot_with_level25 pts