$399 and $401 to Seoul: How Same-Destination Fare Divergence Splits the 17 October Price Window

Executive Summary: Two Dollars, Opposite Signals

The OSINTRAI price-window engine run of 09:06 Beijing time on 17 October 2026 evaluated 59 routes on a five-month fare window from October 2026 through February 2027 and returned 13 BUY signals, 18 AVOID signals and 28 neutral ratings. [source-1] Buried in that distribution is the cleanest possible demonstration of why absolute price is the wrong first question: New York to Seoul is rated BUY at $399, which is 0.444 of its own $898.60 mean, while Chengdu to Seoul is rated AVOID at $401, which is 1.566 of its own $256 mean. [source-1] Two fares separated by two dollars sit at opposite ends of the ratio spectrum, because fare divergence is measured against each route's own average, not against the market. [source-1] The same-destination spread repeats across the schedule: Taipei pairs Los Angeles at 0.500 against Shenzhen at 1.410, Kansai pairs Bangkok at 0.572 against Singapore at 1.351, Hong Kong pairs Chengdu at 0.573 against Shenzhen at 1.684, and Tokyo pairs London at 0.845 against Singapore at 1.472. [source-1] Five destinations, ten origin-side verdicts, one lesson: on 17 October the ratio, not the fare, is the signal, and fare divergence between two origins serving one destination is the sharpest expression of it. [source-1]

1. The Engine and Its Vocabulary

The engine is a numpy 2.4.6 build with a fixed five-month window from 2026-10-01 through 2027-02-01, and its headline output per route is a ratio: the current fare divided by the route's own mean across the window. [source-1] The ratio is a dimensionless measurement key, osintrai_price_window in the house vocabulary, and it is explicitly not a search-popularity metric, not a demand index and not comparable to any external heat score. [source-1] A ratio below roughly 0.75 with a negative trend slope and tolerable variance earns BUY; a ratio above 1.0 with a positive slope earns AVOID; the middle band is WATCH or WAIT and is the honest majority at 28 of 59 routes. [source-1] The month-on-month change is reported alongside the ratio, and the two can disagree: LAX-TPE is 0.500 of its mean yet moved minus 69.2 percent month on month, meaning the fare is far below its window average and still falling. [source-1] [source-2] Reading the ratio without the slope, or the slope without the ratio, inverts decisions in exactly the pairings this report names. [source-1]

Sources: [source-1]

2. The Five Divergences, Route by Route

The strongest form of fare divergence is two origins pricing the same destination at opposite ends of their own windows on the same morning. [source-1] Seoul is the extreme case: JFK-ICN at $399 against an $898.60 mean, ratio 0.444, month on month minus 60.8 percent, trend slope minus 0.1069, coefficient of variation 0.297; against CTU-ICN at $401 against a $256.00 mean, ratio 1.566, month on month plus 67.1 percent, slope plus 0.1971. [source-1] Taipei repeats the pattern with Los Angeles at $442, ratio 0.500, against Shenzhen at $330, ratio 1.410, where the Shenzhen fare is the cheaper ticket in absolute dollars and still the AVOID. [source-1] Kansai pairs Bangkok at $200, ratio 0.572, against Singapore at $540, ratio 1.351; Hong Kong pairs Chengdu at $121, ratio 0.573, against Shenzhen at $636, ratio 1.684, the widest divergence in the run; and Tokyo pairs London at $647, ratio 0.845, against Singapore at $549, ratio 1.472, where again the nominally cheaper Singapore fare is the AVOID. [source-1] In three of the five pairs, the cheaper absolute ticket carries the worse signal, which is the entire argument for ratio-first reading. [source-1]

Sources: [source-1]

3. Fare Divergence Is Not a Discount Claim

A ratio of 0.444 invites the phrase "56 percent off", and that phrase is where most readers go wrong. [source-1] [source-3] Fare divergence lives in the ratio, not in the discount headline, and the denominator is the route's own mean over a five-month window whose reliability varies with its variance. [source-1] JFK-ICN carries a coefficient of variation of 0.297, a tolerable spread, but LAX-TPE at 0.500 ratio carries a coefficient of variation of 0.467, meaning its own observations span widely enough to make the mean itself soft. [source-1] SIN-TPE, the third-cheapest ratio at 0.532, carries the widest variance in the BUY set at 0.497, so its apparent discount is the least anchored of the five and its fare divergence claim the weakest. [source-1] The disciplined read order is variance first, trend slope second, then the weather and schedule context: a cheap ratio on a soft mean is an observation, not an opportunity, while a cheap ratio on a tight mean with a negative slope is the only combination that justifies acting today. [source-1] [source-3] BKK-KIX is the cleanest of the five on that discipline, with ratio 0.572 and the tightest coefficient of variation in the BUY set at 0.219. [source-1]

Sources: [source-1]

4. Why the Same Destination Prices Both Ways

Fare divergence between two origins flying to the same destination is normal market structure, not an anomaly, and October 2026 sharpens it for identifiable reasons. [source-4] [source-5] The origins sit in different market basins with different competition density, different seasonality and different currency exposure, so their window means diverge before the current fares do, and the divergence compounds into the spreads this report measures. [source-4] The month-on-month fields show the AVOID sides inflating on their own baselines: SIN-TYO moved plus 155.3 percent month on month, SIN-ICN plus 104.6 percent, LHR-KIX plus 92.0 percent, while the BUY sides deflate, and the combination produces spreads of 0.6 to 1.1 ratio points between two airports serving one destination. [source-1] For the traveller, the actionable consequence is to price the origin as a variable: a positioning flight or a surface leg to the BUY-side origin can be cheaper than the AVOID-side premium, and in the Hong Kong case the 1.684-versus-0.573 gap dwarfs any realistic positioning cost. [source-1] [source-5] The engine does not model positioning costs, so that comparison is the traveller's arithmetic, not the engine's output. [source-1]

5. What To Do With The 13 BUY And 18 AVOID

The portfolio read is deliberately unheroic: 13 BUY and 18 AVOID out of 59 means 22 percent of the schedule is actionable on the buy side and 31 percent on the avoid side, with a 47.5 percent honest middle that is neither. [source-1] A consumer who quotes only the BUY list has presented a distorted market, and so has one who quotes only the AVOID list, because the middle band is information too. [source-3] Within the BUY set, the five cheapest ratios are JFK-ICN at 0.444, LAX-TPE at 0.500, SIN-TPE at 0.532, BKK-KIX at 0.572 and CTU-HKG at 0.573, and every one carries a negative trend slope, which is the confirmation condition. [source-1] The AVOID set's worst three ratios are SZX-HKG at 1.684, CTU-ICN at 1.566 and SIN-TYO at 1.472, and all three pair an inflating month-on-month with a positive slope, which is the confirmation condition in the other direction. [source-1] Treat AVOID as a wait instruction rather than a prohibition, and treat the fare as a perishable: the ratio valid at 09:06 Beijing time is not a promise about tomorrow. [source-1] [source-6] A car rental or a city activities pass booked against a BUY-side fare is where the saving compounds, and a travel insurance policy is what carries the schedule risk the fare does not.

6. The Portability Of The Method

Nothing in this report is specific to October 2026 or to these five destinations. [source-3] The method travels: take the route's own window mean, divide the current fare by it, check the variance of the denominator, check the slope, and only then compare across routes. [source-1] The failure modes are equally portable: comparing absolute fares across different-mean routes, quoting a ratio as a discount percentage without the variance caveat, and treating a single morning's snapshot as a trend. [source-1] [source-3] On 17 October the snapshot yields a 4.5-times spread between the best and worst BUY-side conditions and a 3.5-times spread between the cheapest and dearest AVOID-side ratios, and those spreads, not the fare table, are the finding. [source-1] Any reader who remembers one sentence should keep the Seoul pair: $399 and $401, two dollars apart, at opposite ends of the spectrum, because the ratio carried the signal and the price did not. [source-1]

Sources: [source-1] [source-3]

Conclusion And Outlook

The 17 October window closes with a clean division of labour between the two numbers every fare carries. [source-1] The absolute price says what the ticket costs today; the ratio says where today sits inside the route's own five-month distribution, and fare divergence between two origins flying one destination is what happens when those distributions have nothing to do with each other. [source-1] The outlook follows the slopes: the BUY set slopes negative across the board, with JFK-ICN at minus 0.1069 and LAX-TPE at minus 0.2148, so the cheap ratios have room to persist or improve, while the AVOID set's worst slopes, SZX-HKG at plus 0.6131 and PVG-CDG at plus 0.3908, point the other way. [source-1] For the traveller, the standing instructions are short: subscribe to fare alerts on the BUY-side pairings, register the dates with more than one search engine so the snapshot discipline survives a single engine's caching, and follow the ratio board rather than the fare board when the two disagree. [source-1] [source-6] [source-7] The next engine run, not this article, is the authority on whether any of these verdicts still stand tomorrow morning. [source-1] [source-27]