We often tweak our formulas as we go along in an effort to become more efficient and accurate. Here’s a log of it:
- MAJOR – I’m totally revamping the formula used to calculate the value of a transferable currency. Up until this month, the value was comprised using two inputs. 90% of it came from the average non-zero value of their transfer partners. The remaining 10% came from the average of their transfer partners’ best redemptions. This method has several flaws:
- The 90/10 split was arbitrary
- A transferable point program that added a low-value transfer partner would then see a drop in value. This is not realistic. So long as high-value partners are not deleted, the addition of low-value ones shouldn’t have much, or any, effect on value.
- Using non-zero data as an input allows for carriers with terrible availability, but several high-value data points, to dominate.
- Going forward, the sole input when determining a transferable point’s ACRV will be it’s transfer partner’s ACRV. However, instead of just averaging, I’ll be summing them using exponentially decreasing weights. The result being, that if a high-value partner is added, the transferable point program will gain value. Conversely, if a low-value one is added, it should have very little, if any effect on the program’s value. More details/examples can be found here.
- Another tweak to the hypothetical maximum flight time. It will now be decreased using a logarithmic curve from 3x the rounded-up nonstop flight time at 0 hrs to 1.65x the nonstop flight time at 20 hrs. Results are still rounded up to the nearest hour.
- 09/28/19 – Adjustment to last months change; Hypothetical routes are now limited to 2x the nonstop time @475 kts, rounded up to the nearest hour.
- 09/28/19 – Cities with a distance >= 5,000 are permitted two stops
- 08/22/19 – Hypothetical routes are now limited to 2x the nonstop time @475 kts, rounded down to the nearest hour.
- 06/10/19 – Began universally presenting valuations in cents instead of fractional dollars. For example a currency calculated to be worth $0.023 will be presented as 2.3¢. No effect on values, just easier to read.
- SEMI-MAJOR: All hotel’s are now reduced in value by 10% instead of 25.
- 08/24/18 – SEMI-MAJOR: Will now compare award stays with cash+point stays at hotels, and use the greater of the two when averaging.
- 07/23/18 – Began hard-coding stop and hour limits on all city pairs. For example, our Atlanta -> Hong-Kong route is now limited to two stops and 32 hours of travel time. Previously, we would simply limit flights to subjective “reasonable” ones.
- 03/06/18 – Began resizing the distribution charts(bell curves), so the y-axis scale is 120% of the curves peak(frequency), rounded up to the nearest 10. This has no effect on valuation, just makes viewing a bit easier
- 02/20/18 – NEW FEATURE! We’ll begin including a non-zero average. We’ve been struggling with the presentation of very low ACRV’s when the carrier has some good redemption options. So we thought we’d include an average that doesn’t take them into account.
- 02/20/18 – SEMI-MAJOR – Will no longer average in a zero for a lack of international first class award availability across all routes. However, if a carrier does offer it, we’ll include it. In other words, we’ll give them credit if they do but won’t hold it against them if they don’t. Reason being similar to 02.18.18’s major update.
- 02/18/18 Replacing ATH->YYZ hypothetical redemptions with LON->PVG. Availability for that route was poor among all carriers.
- 02/18/18 – MAJOR: Removing first-class hypothetical redemptions from all non-North America-Asia routes. Reason being is that three-cabin layouts on non-ultra long haul routes are becoming rarer by the day. Moving to a two-class(economy/business) hypothetical redemption chart will allow us to provide more realistic valuations.
- 02/17/18 – MAJOR: Hypothetical dates are now standardized as the following: next Friday, Friday after next and 4 Fridays from the date of calculation. Previously, I would simply look at hypothetical awards 2, 7 and 21 days in the future. This way I’m always looking at the same day of the week, regardless of when the calc is being performed.
- 02/15/18 Started excluding Icelandair from business class cash price searches. Reason being is that their product is more akin to a domestic one than a true international business class
- 02/14/18 Began removing zeros from the best/worst redemption chart(but not the valuation).
- 02/13/18 Began including ULCC’s(ultra-low cost carriers) in economy cash price searches
- 02/13/18 Began entering a zero for negative redemption values. In other words, when the taxes/fees/carrier surcharges exceed the lowest cash price.
- Removed the IAD-JNB city pair – There was very little award availability across all carriers for this city pair, making calculations cumbersome.
- MAJOR: Began entering a zero instead of a null value for any case in which there is no award availability. Previously, the average would ignore any cell without a value. The rational was this: if there is no award availability, the currency is effectively worthless for that hypothetical redemption. By ignoring this, we’re effectively giving the currency in question a “pass” instead of reducing the average, as we believe should be the case.