ARM vs Fixed.

An ARM is rational for some buyers. Most buyers should not have one.

An adjustable-rate mortgage trades stability for a lower initial rate. The trade is rational when your hold period is shorter than the fixed period, when you have strong income trajectory, or when you intend to refinance. The trade is irrational when you're betting that rates will be lower later — that's a forecast, not a plan.

How an ARM is structured

• Initial fixed period. Rate is locked for 5, 7, or 10 years. The initial rate is typically 50–125 basis points below 30-year fixed.
• Reset. After the fixed period, rate adjusts annually based on a benchmark index (SOFR, CMT) plus a margin (~2.5–3.0%).
• Caps. Modern ARMs typically cap the first adjustment at 2 percentage points, annual subsequent adjustments at 1 point, and lifetime adjustments at 5–6 points. Read the loan documents — caps vary.

When ARM is rational

• Confirmed shorter hold. Job that's reliably going to relocate you in 5 years. Plan to upsize before year 7. Confirmed retirement timing. Use the fixed period that matches your hold length.
• Strong refinance backstop. You have meaningful equity already (large down payment) and credit/income that would qualify you for a competitive refinance if rates were higher at reset.
• Substantial rate spread. If the 7/1 ARM is 75 basis points below fixed, the math is meaningfully different than a 20-basis-point spread. Run the actual numbers.
• Income trajectory. Junior associate at a law firm, medical resident heading toward attending status — confirmed major income jumps before reset can absorb a worst-case reset.

When ARM is a trap

• You're using the lower payment to qualify for a more expensive home than you could afford fixed. The ARM let you stretch; the reset will not be kind.
• You don't have a confirmed exit before reset and your reasoning is "rates will probably be lower by then."
• Your reserves are thin and a worst-case reset would force a refinance regardless of rates at that point.
• You don't fully understand the rate caps in your loan and the math behind a worst-case reset.

Modeling the worst-case

The "worst-case rate after reset" input matters. Most buyers default to "the index can't possibly go that high" — and many were proven wrong in 2022–2024. A realistic worst-case is the lower of (a) loan's lifetime cap above starting rate, or (b) initial rate + 4 percentage points. For a 5.875% start with a 5-point lifetime cap, that's 10.875% as the absolute ceiling; 9.875% is a reasonable stress test.