Skip to main content

Floating-point comparisons

Floating-point values are not stored with perfect decimal precision. That means direct equality checks can fail even when two results look like they should match.

What is happening?​

print(0.1 + 0.2 == 0.3)
print(0.1 + 0.2)

Output:

False
0.30000000000000004

What you might expect: 0.1 + 0.2 is exactly 0.3.

What actually happens: The internal binary representation is only an approximation.

Why this matters​

This shows up in:

  • comparisons in tests
  • money-like calculations
  • thresholds and tolerances
  • scientific or numeric code

The problem is often not the arithmetic itself. The problem is assuming exact equality when the real concept is "close enough."

Prefer math.isclose()​

import math

print(math.isclose(0.1 + 0.2, 0.3))

Output:

True

math.isclose() lets you compare values using a tolerance, which better matches how floating-point numbers behave.

note

For exact decimal behavior, especially with money, decimal.Decimal may be a better fit than float.

Rules of thumb​

  • Do not assume float equality is exact.
  • Use math.isclose() when approximate comparison is the real goal.
  • Choose Decimal when exact decimal semantics matter.
  • Be especially careful with float equality checks in tests.