Blended rate sanity check (March 2026)

[lineoffligbot]

Compared v2 blended rates (EUR base, March 20 2026) against Wise live mid-market rates (March 22 2026). Two-day gap accounts for most of the delta.

Pair	Frankfurter	Wise	Delta
EUR/USD	1.1519	1.1570	-0.4%
EUR/GBP	0.8632	0.8675	-0.5%
EUR/JPY	183.20	184.24	-0.6%
EUR/TRY	50.96	51.25	-0.6%
EUR/CHF	0.9098	0.9119	-0.2%
EUR/CAD	1.5794	1.5879	-0.5%
EUR/AUD	1.6315	1.6476	-1.0%
EUR/CNY	7.9176	7.9875	-0.9%
EUR/KRW	1722.52	1740.00	-1.0%
EUR/BRL	6.0401	6.1500	-1.8%
EUR/INR	107.30	108.49	-1.1%
EUR/MXN	20.58	20.73	-0.7%
EUR/PLN	4.2749	4.2764	0.0%
EUR/SEK	10.77	10.81	-0.4%
EUR/DKK	7.4636	7.4719	-0.1%
EUR/HUF	392.26	393.15	-0.2%
EUR/CZK	24.47	24.51	-0.2%
EUR/RON	5.0912	5.0944	-0.1%
EUR/ZAR	19.51	19.69	-0.9%
EUR/SGD	1.4749	1.4837	-0.6%
EUR/THB	37.71	38.01	-0.8%
EUR/NOK	11.02	11.07	-0.5%
EUR/KWD	0.3530	0.3547	-0.5%
EUR/KZT	553.99	556.81	-0.5%
EUR/GEL	3.1254	3.1470	-0.7%
EUR/BYN	3.4673	3.4886	-0.6%
EUR/PKR	321.36	323.17	-0.6%
EUR/UAH	50.57	50.33	+0.5%
EUR/AZN	1.9521	1.9657	-0.7%
EUR/RUB	96.50	96.18	+0.3%
EUR/MDL	20.04	20.20	-0.8%
EUR/MNT	4074.39	4122.39	-1.2%
EUR/ETB	179.47	181.10	-0.9%
EUR/TJS	11.00	11.04	-0.4%
EUR/KGS	100.45	101.15	-0.7%
EUR/EGP	59.97	60.44	-0.8%
EUR/TMT	3.9983	4.0495	-1.3%
EUR/RSD	117.20	117.47	-0.2%
EUR/KHR	4620.18	4640.15	-0.4%
EUR/TND	3.3834	3.3767	+0.2%
EUR/NPR	170.06	173.49	-2.0%
EUR/LBP	102857.84	103667.00	-0.8%
EUR/VND	29827.20	30428.90	-2.0%
EUR/UZS	14009.77	14074.40	-0.5%
EUR/DZD	151.69	152.87	-0.8%
EUR/MMK	2400.68	2429.70	-1.2%
EUR/CUP	27.42	27.77	-1.3%
EUR/ISK	143.62	143.80	-0.1%

Outliers

Pair	Frankfurter	Wise	Delta	Notes
EUR/BWP	15.34	16.42	-6.6%	BOB official rate diverges from market
EUR/NGN	1546.64	1600.19	-3.3%	Official vs parallel market
EUR/OMR	0.4598	0.4454	+3.2%	Pegged currency spread
EUR/IRR	1,560,012	48,594	~30x	Official rate (CBA/CBR/NBRB) vs parallel market

Most rates are within 1% of Wise mid-market. The consistent slight negative bias reflects the two-day lag. Outliers are explained by official vs market rate divergence in countries with capital controls.

Worth repeating periodically, especially when adding new data sources.

[lineoffligbot]

Sanity check on newly added currencies (March 24 2026). Only 4 of the new currencies have data in Frankfurter so far (BCEAO: XOF, BOI: JOD/EGP/LBP). NBP.B, Banguat, and BCCR backfills are still running.

Pair	Frankfurter	Wise	Delta
EUR/XOF	655.96	655.94	+0.0%
EUR/JOD	0.82167	0.82007	+0.2%
EUR/EGP	60.774	60.817	-0.1%
EUR/LBP	112393	103578	+8.5%

XOF, JOD, and EGP look solid. LBP is the outlier at +8.5%, which tracks with Lebanon's fragmented FX market. BOI publishes the official rate while Wise likely reflects market/parallel rates. Will re-run once NBP.B, Banguat, and BCCR data lands.

[lineoffligbot]

Consensus outlier analysis (April 2026)

Ran the consensus filter across all history (15,369 dates). The filter flagged 11,025 outlier rates. Summary of findings:

By provider

Provider	Outliers	%	Notes
NBU (Ukraine)	5,629	51%	Cross-conversion through UAH, amplified during crisis periods
BOJA (Jamaica)	1,508	14%	Counter/retail rates cross-converted through JMD
RB (Sweden)	1,200	11%	Mostly historical (2005-2013), clean since
Others	2,688	24%	Spread across 25 providers

Root causes

1. Cross-conversion noise (majority): Providers like NBU, BOJA, RB store rates against their local currency. Triangulation to EUR compounds retail spreads. This is expected and not a bug.
2. Source data errors (small subset): NBG RSD had intermittent quantity field inconsistency in their API (2016-2021). Confirmed by querying the source API directly.
3. Volatile/controlled currencies: IRR, BYN, etc. have legitimately different rates depending on which market the provider tracks.

Threshold validation

Compared the current 5% MIN_DEVIATION threshold against a hypothetical 7% threshold using 12 months of date-matched Wise mid-market rates as ground truth:

Quote	n	5% wins	7% wins	Avg error @ 5%	Avg error @ 7%
CHF	365	52	16	0.14%	0.16%
AUD	365	40	29	0.20%	0.21%
JPY	365	37	21	0.18%	0.19%
HKD	365	44	20	0.20%	0.22%
Total	3,650	174	89	0.154%	0.159%

5% threshold produces rates closer to Wise mid-market 2:1 over the relaxed threshold.

Decisions

• No provider exclusions. Let consensus filter data-driven rather than hardcoding opinions about provider quality. If a provider improves, it automatically passes.
• No threshold change. The 5% MIN_DEVIATION is validated against Wise mid-market.
• No code changes needed. Source data bugs (NBG RSD) are correctly caught by the filter.