airline-miles-value-ranker

Use when deciding how to spend frequent-flyer miles or credit-card points — comparing redemption options, judging a "deal," choosing between pay-with-miles, booking awards directly, or transferring points to partners — computes cents-per-point for each option with fare-comparison math, checks against 12 known partner sweet spots, prices the cash-fare floor, and issues BURN / HOARD / NEVER verdicts so a 300k-point balance never gets flushed at 0.4¢.

Points and miles die twice: they expire, and they die quietly when you spend 50,000 of them on a $200 flight — a 0.4¢-per-point redemption that a transfer partner would have turned into $1,000+ of value. The average leisure redeemer doesn't compute cents per point (CPP) at all; the airline apps make sure the worst option is the easiest button to press ("Pay with miles — 25,000 points!" on a $180 ticket).

This skill turns every redemption decision into arithmetic:

text

CPP = (cash fare − taxes/fees paid on award) / points spent × 100

It ranks your actual options, compares them to anchor values (~1.0–1.5¢ for major currencies), flags the sweet spots worth chasing, and says BURN / HOARD / NEVER plainly.

When to Use

• "Should I use 60,000 miles for this flight or just pay $580?"
• Comparing an airline's own "book with miles" price vs a transfer partner award vs paying cash and saving the miles
• Deciding whether to move credit-card points to a partner now or wait for a transfer bonus
• Planning a big redemption (business class, family of four) and wanting the ordering: which airline, which route, which cabin
• Points expiring — "dump them on something" mode: best floor values to burn into rather than $0 after devaluation
• Post-devaluation triage: an award chart just got worse — what's the new best use of the balance?

Don't use for: hotel points (different economics — nightly value vs fare comparison; the CPP math transfers but the anchors don't), or cash-back cards (no redemption complexity — just take 2%).

The Value Model

Step 1 — Establish the cash fare honestly

The comparison fare is what you would actually pay for that exact itinerary in that exact cabin:

• Same flights, same cabin, same date — from a search, not memory
• If you'd never pay $6,000 for business class, you may still use it as the ceiling, but ALSO price premium-economy on the same route: the honest "value over what I'd buy" number
• Include the taxes/fees the award requires you to pay in cash: award value = (cash fare) − (award's cash copay)

Step 2 — Compute CPP for every option on the table

For each candidate: (fare − copay) / points × 100. The script (scripts/miles_ranker.py evaluate) does this and ranks them.

Step 3 — Judge against anchors

Currency	Weak	Fair	Good	Excellent
Amex MR / Chase UR (transferable)	<1.0¢	1.0–1.25¢	1.25–1.8¢	>1.8¢
Airline miles (UA/DL/AA class)	<0.8¢	0.8–1.2¢	1.2–1.6¢	>1.6¢
Hotel points (context only)	<0.5¢	0.5–0.8¢	0.8–1.2¢	>1.2¢

A transferable point spent at 1.0¢ via portal credit is the floor — spending transferrable points BELOW the portal rate is a pure loss with extra steps.

Step 4 — Verdicts

• BURN — CPP at or above "Good" for that currency, or points expiring, or a known devaluation announced: book it
• HOARD — below Fair but balance is young, no expiry pressure, and a plausible big redemption within 12 months
• NEVER — any airline-program redemption below ~0.7¢, nearly all "pay with points at checkout" options, merchandise/gift cards (0.5–0.8¢ fixed), or paying with miles for taxes only

Known Sweet Spots (verify current charts — they shift)

The script ships a sweet-spot table; treat it as a starting map:

1. Short-haul partner awards — 5,000–7,500 miles for hops that cost $150+ cash (1.5–2.5¢ CPP routinely)
2. Business class to Europe on partners — 55k–70k one-way vs $3,000+ fares
3. ANA / Lufthansa first via partners — the classic 2–4¢ ceilings (fuel surcharges can gut it — compute, don't assume)
4. Off-peak economy (AA-style charts) — dated charts keep off-peak pricing long after dynamic cash pricing moves up
5. Transfer bonuses 20–30% — effectively multiply CPP of any partner plan; rarely worth chasing for sub-1¢ plans though

The Expiry and Devaluation Clock

• Points that expire (many airline programs: 18–24 mo activity) are worth ~15% less book value — plan a burn before the wall
• A announced chart devaluation (happens ~yearly somewhere) converts HOARD to BURN for the affected program
• Transferable bank points don't expire — they are the correct parking lot for indecision

Pitfalls

1. Valuing business class at retail. $8,000 first-class fare ≠ $8,000 of value if you'd never pay it. Use "value over the cabin you'd actually buy" as a second line — the script supports both via --ceiling-fare vs --realistic-fare.
2. Forgetting the copay. "Free" award with $560 in surcharges (LONDON) can be worse than a cheap cash ticket.
3. Dynamic award pricing bait. Programs now show "miles from 12,500" — your date may price at 95,000. Always evaluate the actual priced option.
4. Transfer before confirming award space. Points move one way. Confirm the award seat exists, THEN transfer.
5. Sunk-mile fallacy. "I have the miles so the flight is free" — no: miles spent below 0.7¢ are cash set on fire. Compare to the floor redemption every time.
6. Mixing currencies in your head. 100k Amex ≠ 100k Delta ≠ 100k Air France. Rank per-currency, then choose.

Verification Checklist

• Cash fare pulled from a live search (not remembered)
• Award copay/taxes included in the math
• CPP computed per option, ranked
• Compared to the currency's anchor table
• Expiry / devaluation pressure factored into BURN vs HOARD
• Transfer options checked for bonuses before moving points

Quick Recipe

text

python3 scripts/miles_ranker.py evaluate \
  --label "SFO→NRT business" --points 75000 --copay 42 \
  --realistic-fare 2600 --ceiling-fare 6800
python3 scripts/miles_ranker.py evaluate \
  --label "Same route economy" --points 40000 --copay 42 \
  --realistic-fare 850 --ceiling-fare 850
python3 scripts/miles_ranker.py compare

The compare view prints the ranked table with verdicts, so the decision is a one-glance read.