The Efficiency Landscape: Plotting Offensive vs. Defensive Rating

A standings table tells you who is winning. It doesn't tell you how. Two 50-win teams can be built completely differently — one bludgeoning opponents on offense, the other suffocating them on defense. The cleanest way to see that at a glance is to plot every team's offensive rating against its defensive rating on a single chart and split it into four quadrants. Build exactly that from the bundled NBA team ratings, and one picture separates the contenders from the rebuilders.

This pairs naturally with the net-rating dashboard table: that tutorial ranks teams; this one shows the whole landscape. The data is the bundled nba_ratings.csv (per-100-possession ratings, Basketball-Reference, retrieved June 2026), so it runs offline.

1. Load the ratings and find the league average

   Offensive rating (ORtg) is points scored per 100 possessions; defensive rating (DRtg) is points allowed per 100; net rating (NRtg) is the difference. Because both ends are measured on the same per-100 scale, league-average offense and league-average defense are the same number — the crosshairs of our chart.

   import pandas as pd
   
   df = pd.read_csv("nba_ratings.csv")   # Team, W, L, W/L%, ORtg, DRtg, NRtg
   lg_off = df["ORtg"].mean()
   lg_def = df["DRtg"].mean()
   print(df.sort_values("NRtg", ascending=False)[["Team", "ORtg", "DRtg", "NRtg"]].head().to_string())

   League average and the best net ratings

   League-average rating: 116.2 offense / 116.2 defense
   
   Best net rating:
                        Team    ORtg    DRtg   NRtg
   0          Boston Celtics  124.23  112.51  11.71
   1   Oklahoma City Thunder  120.43  113.12   7.31
   2  Minnesota Timberwolves  116.63  109.98   6.65
   3          Denver Nuggets  119.49  114.05   5.44
   4         New York Knicks  119.11  114.22   4.89

   The league averages out to about 116 points per 100 possessions on each end. At the top, the Celtics pair an elite offense (124.2) with a strong defense (112.5) for a net rating near +12 — the signature of a true contender. Minnesota gets there the opposite way, anchored by the league's stingiest defense (about 110).

2. A short label for each point

   Thirty full team names would turn the chart into spaghetti. We derive a three-letter tag from each nickname — the last word of the name, trimmed and upper-cased.

   df["Tag"] = df["Team"].str.split().str[-1].str[:3].str.upper()
   # "Boston Celtics" -> "CEL", "Denver Nuggets" -> "NUG"

   Chaining string methods with the .str accessor is the pandas way to transform a whole column at once, with no loop.

3. Draw the quadrant scatter

   Now the payoff. We scatter ORtg on the x-axis and DRtg on the y-axis, color each point by net rating, and draw dashed lines at the league averages. The one trick that makes it readable: invert the y-axis, because a lower defensive rating is better, and we want good defenses at the top.

   import matplotlib.pyplot as plt
   
   fig, ax = plt.subplots(figsize=(8, 7))
   sc = ax.scatter(df["ORtg"], df["DRtg"], c=df["NRtg"], cmap="RdYlGn",
                   s=70, edgecolor="#20242B", linewidth=0.5)
   ax.axvline(lg_off, color="#6C7079", ls="--")
   ax.axhline(lg_def, color="#6C7079", ls="--")
   for _, r in df.iterrows():
       ax.annotate(r["Tag"], (r["ORtg"], r["DRtg"]),
                   textcoords="offset points", xytext=(5, 2), fontsize=7)
   
   ax.invert_yaxis()   # better defense (lower DRtg) now sits at the top
   fig.colorbar(sc, ax=ax).set_label("net rating")
   fig.savefig("efficiency_quadrant.png", dpi=144, bbox_inches="tight")

   

   Read it by quadrant. Top-right is the promised land: above-average offense and defense (the green points). Bottom-left is a rebuild — below average at both ends. The other two corners are the lopsided teams: great offense with a leaky defense (bottom-right), or a defense-first grinder that can't score (top-left). One chart, and every team's identity is obvious.

Troubleshooting