Methodology

How we value points and miles

An Award Rate valuation is the median cents-per-point across a fixed basket of 49 route-and-cabin cells and 44 hotel properties, each award priced against the cash cost of the same seat or room on the same date, with taxes and carrier surcharges subtracted from the cash side. It is a benchmark for choosing between points and cash — not a price you can sell at.

Updated Card terms modeled, not yet reconciled against issuer disclosures — what that means

What an Award Rate valuation is — and what it is not

A valuation answers one question: when should I pay with points instead of money? If a redemption returns more cents per point than our published figure for that currency, spending points is the better of the two options in front of you. If it returns less, pay cash and keep the points for something better.

That is the whole of it. A valuation is a decision threshold, and it is deliberately a boring one. It is not a market price, it is not an appraisal of your balance, and it is not a promise that any particular seat exists.

The formula, in full

One observation is one award redemption priced against one cash alternative. Its value in cents per point is:

cash comparable − cash paid at award bookingpoints required× 100

The numerator is what you avoided spending. The denominator is what it cost you. Everything contentious in points valuation is a disagreement about how to fill in those two terms — which is why the rest of this page is about exactly that.

The published figure for a currency is the median of every observation for that currency in the quarter, taken across the whole basket rather than per route. We report the median of observations, not the median of route medians, because the second of those silently upweights thin markets where award space barely exists.

The sampling frame: which routes and properties we price

The basket is fixed. It was set in the Q1 2025 model and is reviewed once a year, in December, for route relevance only. Freezing it matters more than choosing it well: a valuation that moves because we changed which routes we price is not a measurement, it is an edit.

Air: 49 route-and-cabin cells

The air basket, unchanged since the Q1 2025 modelEach cell is priced on four departure dates per month — 21, 45, 90 and 180 days out — plus one fixed peak date per quarter (the Wednesday before US Thanksgiving, the Sunday after Christmas, the first Friday of July, and Golden Week in Japan). Round trips are priced as two one-ways wherever the program permits one-way awards.
BandRoutes pricedCabinsCells
US domestic short-haul (under a thousand miles)LGA–ORD · DAL–HOU · LAX–SFO · ATL–MCOEconomy4
US transcontinentalJFK–LAX · EWR–SFO · BOS–SEAEconomy, domestic first / transcon business6
US–HawaiiLAX–HNL · SEA–OGGEconomy, first4
US–Caribbean and MexicoJFK–SJU · LAX–SJD · MIA–PUJEconomy3
US–EuropeJFK–LHR · ORD–FRA · BOS–CDG · IAD–ZRHEconomy, business8
US–AsiaSFO–NRT · LAX–ICN · SEA–HKG · JFK–SINEconomy, business8
US–Middle East and IndiaJFK–DOH · EWR–DELEconomy, business4
US–South AmericaMIA–GRU · IAH–SCLEconomy, business4
US–OceaniaLAX–SYD · SFO–AKLEconomy, business4
Intra-Europe and intra-AsiaLHR–FCO · CDG–ATH · NRT–BKK · SIN–DPSEconomy4
Total49

Hotels: 44 properties across three segments

The hotel basket, unchanged since the Q1 2025 modelEach property is priced for a one-night stay on four check-in dates per month — 14, 30, 60 and 120 days out — plus one compression date per quarter drawn from a fixed list of citywide events in each market. Five-night stays are priced separately so that fifth-night-free and fourth-night-free benefits are measured where they actually apply rather than assumed across the board.
SegmentRepresentative brandsMarketsProperties
Select-service and airportHyatt Place · Courtyard · Hampton · Holiday Inn ExpressChicago, Denver, Dallas, Orlando, Manchester (UK)14
Upper-upscale urbanHyatt Regency · Marriott · Hilton · Kimpton · Crowne PlazaNew York, London, Paris, Tokyo, Chicago, Los Angeles18
Luxury and resortPark Hyatt · Ritz-Carlton · Waldorf Astoria · InterContinental · ConradMaui, Cancún, Dubai, Bangkok, Paris, Tokyo12
Total44

The basket is deliberately weighted toward trips people actually take. There is more domestic economy in it than a points enthusiast would choose, and less first class than a highlight reel would contain, because the median is meant to describe an ordinary redemption. If you only ever fly transatlantic business class, your realized value will run above our published figures — and the per-currency notes on the valuations page tell you which currencies reward that pattern most.

How the cash comparable is chosen

The cash comparable is the fare or rate you would have bought instead, captured in the same session as the award price so the two are contemporaneous. Five rules govern the choice, and they are the rules that do most of the work in this model.

  1. Same product, not the cheapest product. An award that earns a full main-cabin seat is compared with a main-cabin fare, never with the basic-economy fare beneath it. An award night in a standard king is compared with the cash rate for a standard king, not with the chain’s cheapest room type on the night.
  2. Same operating carrier and flight numbers. A partner award on ANA metal is priced against ANA’s own cash fare for that flight, not against the cheapest itinerary between the same two cities.
  3. Cheapest publicly bookable fare, all-in. Taxes and mandatory fees are included on the cash side. Optional ancillaries — seat selection, bags, Wi-Fi — are excluded from both sides, because a valuation should not depend on how much luggage you carry.
  4. Winsorised at the 95th percentile. If the contemporaneous cash fare sits above the 95th percentile of that route-and-cabin’s trailing twelve-month distribution, it is replaced by the 95th-percentile value. This is the single most consequential rule in the model and the main reason our figures are lower than most published elsewhere.
  5. Refundability is matched, not ignored. Awards in most programs cancel back to points for free. Where they do, the cash comparable is the cheapest fare with equivalent change terms if one exists within a fifth of the non-refundable price; otherwise the non-refundable fare is used and the difference is not credited to the award. We would rather understate the currency than launder a flexibility premium into it.

Why we publish the median rather than the best redemption

Award value distributions are severely right-skewed. A handful of redemptions in every program return several times the typical figure, and a mean would let those few observations set the number for everyone. The median is the honest summary of a skewed distribution, and it has a property a mean does not: it is the value you have a fifty-fifty chance of beating.

The cost of that choice is that our numbers look pessimistic beside sites that publish something closer to a best case. They are meant to. If your redemptions consistently beat our figure, you are redeeming better than the median — which is the point of learning any of this.

We also publish the interquartile spread internally and use it to decide what to publish at all. A currency whose middle fifty per cent of observations spans more than a full cent is flagged as unstable and its per-currency note says so, because a single median hides a great deal when the distribution is that wide.

How taxes, fees and carrier surcharges are handled

Every award has a cash component. Government taxes, security fees, airport charges, and — the expensive one — carrier-imposed surcharges. All of it is subtracted from the cash comparable before the division, which means a program that levies a large surcharge on a transatlantic business award scores lower than one that does not, on the same seat, on the same day.

This is not how most published valuations treat surcharges, and it is a large part of why our figures differ. A mileage price quoted without its cash companion describes half a transaction. On the routes where surcharges bite hardest, the difference between the two treatments is worth several tenths of a cent per mile — enough to reorder the table.

Hotel awards get the same treatment in reverse: where a program waives resort fees on award nights and a competitor does not, the waived fee is part of what the points bought and is credited to the currency. That single rule accounts for a meaningful share of the gap between the strongest and weakest hotel currencies in our model.

How fixed-value redemption floors enter a bank currency’s number

A transferable bank currency is two things at once: an option on every partner it reaches, and a fixed-value redemption you can always fall back on. Most people use the fallback more often than the option. A model that prices only the transfers would describe an enthusiast, not a cardholder.

So for each bank currency we sample both, and combine them in the proportion that real redemptions occur. That proportion comes from two sources: the redemption mix issuers disclose in their own reporting, and our annual reader survey — 3,412 responses in January 2026, weighted to the balance distribution the same survey reports. Neither source is perfect. Both are named here so you can discount the number accordingly.

This is why Chase Ultimate Rewards is valued at 1.9¢ rather than at the value of its best partner: the fixed-value floor is a large share of actual redemptions, and it pulls the median down. It is also why the floor is worth understanding as a benefit rather than a limitation — it is the reason a bank point can never be worth nothing.

What the model deliberately excludes

What a model leaves out determines its number as much as what it includes. Here is the complete exclusion list, with the reasoning for each.

Awards we could not actually book
Every observation is carried to the program’s own payment screen before it counts. Phantom space — an award that displays and then fails at ticketing — is the single most common way a published valuation gets inflated, because the seats that never ticket are disproportionately the cheap ones.
Cabins with negligible award availability
A cabin enters the basket only if the programs we track released bookable saver space on at least a fifth of sampled dates. Lufthansa first class fails that test and is excluded, which is why our Star Alliance numbers look lower than valuations built on it.
Elite-only and status-dependent value
Suite upgrade awards, Globalist breakfast, waived close-in fees and elite award discounts are worth real money and are not in the model. A valuation should describe the currency, not the member.
Transfer bonuses
A transfer bonus changes how many points you receive, not what one is worth after it lands. Folding bonuses into a valuation double-counts them the moment you multiply the valuation by a bonused balance.
Speculative and manufactured itineraries
No positioning flights, no married-segment tricks, no fuel-dumping, no error fares, no fifth-freedom routings priced against a cash fare a normal traveler would never buy.
Cash fares above the 95th percentile of their own route history
Comparables are winsorised at the 95th percentile of that route-and-cabin’s trailing twelve-month fare distribution. Pricing an award against a last-seat walk-up fare is how hotel currencies get valued at two cents and airline miles at eight.
Basic economy and non-comparable fare products
An award that earns a full main-cabin seat is compared with a main-cabin fare, not with the stripped basic-economy fare underneath it. Comparing unlike products inflates every domestic number in the model.
Cash-out floors that do not exist
Only currencies with a published fixed-value redemption carry a floor in the model. Airline miles have none, which is exactly why they are riskier to hold than the arithmetic alone suggests.

How often the model is revised, and what triggers an off-cycle change

The model is republished quarterly, on the first business day of the quarter. The current model is Q3 2026, published . The previous quarter’s figure is published alongside the current one for every currency, so a devaluation appears as a visible change rather than as a quietly edited number.

What triggers a revision outside the quarterly cycleBetween quarters we correct facts, not estimates. A ratio or a partner list is fixed the week it changes; a valuation waits for enough new observations to move a median.
TriggerWhat we do
A program publishes, withdraws or rewrites an award chartImmediate re-sample of that currency, published within ten business days with a changelog entry.
A transfer ratio changes, or a partner is added or droppedThe partner tables are corrected the same week. The valuation is revised only if the change moves the trailing median.
The trailing 90-day median moves more than a tenth of a cent from the published figureFlagged for the next quarterly publication. We do not chase noise between quarters.
A reader or another publisher demonstrates an errorCorrected and logged under our corrections policy, with the previous figure left visible rather than overwritten.

Every revision is logged publicly in the valuation changelog, with the direction of the change and the reason for it. The log is not a marketing artefact — it is the only way to check whether the claims on this page are true over time, which is why it includes the quarters where we were wrong.

Where this model can be wrong

Four known weaknesses, stated plainly, because a methodology page that only lists strengths is marketing.

  1. The basket is American. It is built around US originations and it will understate programs whose value lies in intra-Asia or intra-Europe short-haul. A reader based in Singapore should treat our airline numbers as directional rather than local.
  2. Availability is measured, not modeled. We record whether space was bookable on the dates we sampled. We do not attempt to model how hard it was to find, and “findable with two hours of searching” and “findable instantly” score identically. That flatters the programs with the worst search tools.
  3. Hotel award nights are priced one property at a time. Footprint is not in the arithmetic. A currency with a fixed chart and few hotels can score above a currency you can actually spend where you travel, and the per-currency notes have to carry that caveat rather than the number.
  4. The floor mix is estimated. The share of bank-currency redemptions that go through a fixed-value portal is the least certain input in the model. If it is wrong, every bank currency is wrong in the same direction — and by more than any other single assumption here.

Conflicts of interest, and who pays for this research

Award Rate is funded by affiliate commission from credit card issuers. When a reader is approved for a card through a link on this site, we are usually paid. That arrangement is disclosed on every page that displays an offer, and it creates an obvious incentive to publish valuations that make card bonuses look larger than they are.

Four controls exist against that incentive, and they are checkable rather than promised.

  • The valuations are computed by the data team, who have no access to revenue reporting by card or by issuer. The people who know which cards pay best do not touch the model.
  • No issuer, airline or hotel group is given sight of a figure before publication, and none has ever been offered the opportunity to comment on one.
  • Bonus values across this site are derived from these numbers rather than typed in, so a cut propagates automatically to every card page, including the most commercially valuable ones. This quarter 4 currencies were cut, and the largest of those cuts reduced the headline value of the card bonuses that earn it immediately and without discussion.
  • Valuations are never used as a ranking input in a way that advertising could reach. Where a commercial relationship exists with a card whose currency we downgraded, the downgrade stands and the ranking moves.

If you believe a number here has been shaded commercially, the changelog and the sample sizes are the evidence to argue with, and we would rather have the argument in public. Our editorial policy and advertiser disclosure set out the rest.

Reproducing, citing and challenging these numbers

The valuation table is published as a citable dataset with a license, a canonical URL and a quarter stamp, and the how to cite this page block on the valuations page gives the exact form. Quote the quarter with the figure; a valuation without its quarter is a rumour.

We will supply the underlying observation set — route, date, cabin, program, points required, cash comparable, cash paid — to researchers, journalists and other publishers who ask, under the same attribution terms. Write to editors@awardrate.com with the quarter and the currencies you need. If you find an error in it, tell us and we will log the correction with your name on it.

Where a guide works a single redemption end to end, the record behind it is published in the same form: the query as entered, where it was entered, the prices returned, the derivation, and an explicit statement of which figures are independently confirmable today and which are point-in-time observations. The worked example on the beginners guide is served at /evidence/worked-example-2026-07-14.json. We do not publish screenshots of an airline's or a hotel's own booking interface: a dynamically priced fare on a date that has passed cannot be re-derived from an image, so a screenshot would imply a proof it cannot carry.

Frequently asked questions

  • Is a point valuation a price I can sell my points at?

    No. A valuation is a benchmark for deciding between two ways of paying for the same trip. It is not a market price, it is not an offer to buy, and most programs prohibit selling points outright. If you cash a balance out at the program’s own fixed rate you will almost always get less than our valuation, because our figure includes transfer redemptions that the cash-out option does not.
  • Why is your valuation lower than the one on other points sites?

    Two reasons, both structural. We publish the median rather than a headline redemption, so half of our sampled redemptions come in below our own number. And we winsorise cash comparables at the 95th percentile of each route’s own twelve-month fare history, which removes the walk-up fares that make a mile look like it is worth several cents. A valuation built on the best case is a description of the best case.
  • How large is the sample behind each number?

    Sample sizes are published beside every figure and range from 3,120 to 22,400 priced redemptions per currency for Q3 2026, across 184,620 observations in total. A currency with fewer than a thousand usable observations is not published at all; it is marked as derived and stated conservatively.
  • Do you include taxes and carrier surcharges?

    Yes, and they reduce the valuation rather than being footnoted away. The cash you pay at the time of an award booking is subtracted from the cash comparable before the division, so a program that adds several hundred dollars of carrier surcharge to a transatlantic business award scores lower than one that does not. That single choice is most of the gap between our numbers and valuations that quote the mileage price alone.
  • How often do the valuations change?

    Quarterly, published on the first business day of the quarter — the current model is Q3 2026, published August 6, 2026. Off-cycle revisions happen when a program publishes or withdraws a chart, or when a demonstrated error needs correcting. Both the current and the previous figure are shown on the valuations page so a devaluation is visible rather than quietly edited away.
  • Does advertising influence the valuations?

    No. The valuations are computed by our data team, who have no access to revenue reporting by card or issuer, and no issuer, airline or hotel group sees a figure before publication. Because welcome-bonus values across this site are derived from these numbers, a cut lowers the headline value of the cards that pay us most — and we publish it anyway. Our changelog exists so you can check that claim against our own record.

About the author

Editor, Award Rate

Nicholas Miles writes and maintains every page on Award Rate, and the site is built so that none of those pages has to be taken on faith: every card figure is rendered from a structured record that is checked against the issuer’s own pricing-and-terms disclosure, and the build fails if a number appears in prose instead. The points side of the site is lived-in rather than theoretical — Nicholas holds and tracks balances across ten loyalty programs — transferable bank currencies and the airline and hotel programs they feed — and recently took the family on a multi-stop award trip across Korea, Japan and Hawaii, flown mostly on points — the multi-seat, fixed-dates redemption problem most points advice quietly ignores. The conflicts that come with that are disclosed rather than hidden: the editor personally collects and redeems points in programs this site covers, the site is funded by affiliate commission disclosed on every offer page, and no issuer, airline or hotel group has any other relationship with it.

  • Personally earns, tracks and redeems points and miles across ten loyalty programs
  • Built the Award Rate card-data model and the verification gates every published figure must pass
  • Booked and flew a multi-stop family award trip across Korea, Japan and Hawaii, mostly on points

Disclosures and sourcing

How Award Rate makes money

Award Rate is an independent publisher and comparison service, not an investment or financial advisor. Some of the card offers on this site are from advertisers who compensate us when you are approved through our links. That compensation may affect where and in what order offers appear, but it never affects our ratings, our point valuations, or what we recommend. We do not review every card on the market, and card offers displayed here may not include all available offers.

Where the numbers come from

Our editorial team writes and scores every review before any commercial arrangement is considered. Issuers do not review, approve, or endorse our content.

Dollar values are derived from the published Award Rate valuation model rather than from issuer marketing, and the model is republished with its sample sizes each quarter.