Table: repricing of the input instruments, as-of 2026-09-18 (implied minus market, bp).

| Instrument | Market (%) | A implied (%) | A error (bp) | B implied (%) | B error (bp) |
|---|---:|---:|---:|---:|---:|
| O/N | 3.8500 | 3.8500 | -9.1e-12 | 3.8500 | -9.1e-12 |
| T/N | 3.8500 | 3.8500 | 5.2e-10 | 3.8500 | -2.8e-10 |
| 1M | 3.8974 | 3.8974 | 1.4e-11 | 3.8974 | -1.3e-11 |
| 2M | 3.9683 | 3.9683 | -4.2e-11 | 3.9683 | -3.3e-12 |
| 3M | 4.0319 | 4.0319 | -4.6e-11 | 4.0319 | -2.5e-12 |
| 4M | 4.1031 | 4.1031 | -1.5e-10 | 4.1031 | -2.6e-12 |
| 6M | 4.2057 | 4.2057 | -3.9e-09 | 4.2057 | -2.1e-13 |
| 1Y | 4.4987 | 4.4987 | -9.5e-12 | 4.4987 | 1.4e-12 |
| 18M | 4.5868 | 4.5868 | -6.2e-13 | 4.5868 | -2.1e-13 |
| 2Y | 4.6358 | 4.6358 | 9.2e-10 | 4.6358 | 9.7e-13 |
| 3Y | 4.6342 | 4.6342 | 6.9e-14 | 4.6342 | 4.2e-13 |
| 4Y | 4.6107 | 4.6107 | 4.4e-12 | 4.6107 | 6.9e-14 |
| 5Y | 4.5968 | 4.5968 | 4.9e-13 | 4.5968 | 2.8e-13 |
| 6Y | 4.5917 | 4.5917 | -2.8e-13 | 4.5917 | 4.9e-13 |
| 7Y | 4.5923 | 4.5923 | -2.1e-13 | 4.5923 | -6.9e-14 |
| 8Y | 4.5964 | 4.5964 | -5.6e-13 | 4.5964 | 3.5e-13 |
| 9Y | 4.6107 | 4.6107 | -5.4e-12 | 4.6107 | 2.8e-13 |
| 10Y | 4.6208 | 4.6208 | -1.5e-12 | 4.6208 | 4.2e-13 |
| 12Y | 4.6582 | 4.6582 | -3.5e-11 | 4.6582 | 3.5e-13 |
| 15Y | 4.7085 | 4.7085 | -1.4e-13 | 4.7085 | 2.1e-13 |
| 20Y | 4.7477 | 4.7477 | 0.0e+00 | 4.7477 | 1.4e-13 |
| 30Y | 4.6637 | 4.6637 | 3.5e-11 | 4.6637 | 2.1e-13 |

Source: out/repricing.csv, bootstrap.py.

Table: curve nodes and discount factors.

| Instrument | Quote (%) | Node date | DF (A) | DF (B) | Zero A (%) | Zero B (%) |
|---|---:|---|---:|---:|---:|---:|
| O/N | 3.8500 | 2026-09-21 | 0.99967927 | 0.99967927 | 3.9028 | 3.9028 |
| T/N | 3.8500 | 2026-09-22 | 0.99957237 | 0.99957237 | 3.9030 | 3.9030 |
| 1M | 3.8974 | 2026-10-26 | 0.99590577 | 0.99589852 | 3.9407 | 3.9477 |
| 2M | 3.9683 | 2026-11-25 | 0.99256500 | 0.99256231 | 4.0058 | 4.0072 |
| 3M | 4.0319 | 2026-12-24 | 0.98926020 | 0.98925711 | 4.0631 | 4.0643 |
| 4M | 4.1031 | 2027-01-26 | 0.98539645 | 0.98539331 | 4.1305 | 4.1314 |
| 6M | 4.2057 | 2027-03-24 | 0.97863739 | 0.97863384 | 4.2149 | 4.2156 |
| 1Y | 4.4987 | 2027-09-24 | 0.95572273 | 0.95572071 | 4.4555 | 4.4557 |
| 18M | 4.5868 | 2028-03-24 | 0.93322386 | 0.93322614 | 4.5615 | 4.5614 |
| 2Y | 4.6358 | 2028-09-26 | 0.91120400 | 0.91120732 | 4.5928 | 4.5926 |
| 3Y | 4.6342 | 2029-09-26 | 0.87035500 | 0.87035722 | 4.5907 | 4.5907 |
| 4Y | 4.6107 | 2030-09-25 | 0.83221605 | 0.83221601 | 4.5666 | 4.5666 |
| 5Y | 4.5968 | 2031-09-24 | 0.79573496 | 0.79573440 | 4.5523 | 4.5523 |
| 6Y | 4.5917 | 2032-09-24 | 0.76045965 | 0.76045897 | 4.5473 | 4.5473 |
| 7Y | 4.5923 | 2033-09-26 | 0.72641480 | 0.72641346 | 4.5484 | 4.5484 |
| 8Y | 4.5964 | 2034-09-26 | 0.69383432 | 0.69382998 | 4.5534 | 4.5535 |
| 9Y | 4.6107 | 2035-09-26 | 0.66194101 | 0.66194129 | 4.5703 | 4.5703 |
| 10Y | 4.6208 | 2036-09-24 | 0.63168916 | 0.63168734 | 4.5823 | 4.5823 |
| 12Y | 4.6582 | 2038-09-24 | 0.57321602 | 0.57320626 | 4.6279 | 4.6281 |
| 15Y | 4.7085 | 2041-09-25 | 0.49406408 | 0.49405758 | 4.6912 | 4.6913 |
| 20Y | 4.7477 | 2046-09-26 | 0.38682467 | 0.38688308 | 4.7405 | 4.7397 |
| 30Y | 4.6637 | 2056-09-26 | 0.25196788 | 0.25260335 | 4.5881 | 4.5798 |

Source: out/nodes.csv, bootstrap.py.

Table: comparison metrics.

| Metric | A: log-linear DF | B: monotone convex |
|---|---:|---:|
| Max abs repricing error (bp) | 3.9e-09 | 2.8e-10 |
| Largest day-to-day jump in overnight forward, as-of date to last node (bp) | 59.71 (2046-09-26) | 1.68 (2029-09-28) |
| Day-to-day forward jumps over 2 bp (count) | 18 | 0 |
| Total variation of the daily forward to the last node (bp) | 421.3 | 365.0 |
| Largest change of the forward over 5 business days, to the last node (bp) | 59.73 (2046-09-28) | 5.34 (2026-12-31) |
| Forward at the 3Y node and 3 business days later (%) | 4.4319 / 4.4319 | 4.4705 / 4.4317 |
| Average overnight forward (Act/360) over business days from the 20Y node to the day before the 30Y node (%) | 4.2249 | 4.2013 |
| Overnight forward (Act/360) at the 20Y node and at the 30Y node (%) | 4.2248 / 4.2248 | 4.6127 / 3.9957 |
| Min / max overnight forward to the last node (%) | 3.8500 / 4.8778 | 3.8500 / 4.8930 |
| Jump in overnight forward at the last node (bp) | 0.00 | 0.00 |
| DF at 4.5Y (2031-03-24) | 0.81397156 | 0.81397425 |
| PV of USD 100m at 4.5Y (USD) | 81,397,155.57 | 81,397,425.03 |
| Max abs DF difference A-B on the monthly grid, and its date | 2.275e-03 (2050-04-18) | (one value for both) |
| Same difference as PV of USD 100m paid on that date (USD) | 227,458.20 | (one value for both) |
| Zero rate on that date (%) | 4.6715 | 4.7006 |
| Same difference as zero rate on that date, B minus A (bp) | 2.9 | (one value for both) |
| 5Y +1 bp: max forward change inside [4Y, 6Y] (bp) | 5.5367 | 7.6650 |
| 5Y +1 bp: max forward change outside [4Y, 6Y] (bp) | 0.0305 | 4.1475 |
| 4.5Y cash flow: share of abs PV change outside the 4Y and 5Y inputs | 5.86% | 11.94% |
| Payment lag 0 vs 2: max abs DF change (monthly grid) | 1.33e-05 | 4.79e-06 |
| Quarterly vs annual payments: max abs DF change (monthly grid) | 6.62e-03 | 6.61e-03 |
| Fixed leg read as Act/365 Fixed instead of Act/360: max abs DF change (monthly grid) | 5.00e-03 | 5.00e-03 |
| Overnight anchor +1 bp: max abs DF change (monthly grid) | 1.11e-06 | 1.31e-06 |
| Extrapolated DF at 40Y | 0.16426043 | 0.16853911 |
| Extrapolated overnight forward at 35Y (%) | 4.2248 | 3.9957 |

Source: out/results.json, bootstrap.py.

Table: interpolation variants compared with method B.

| Variant | Bootstrap | Max abs DF difference vs B (monthly grid) | Max daily forward jump (bp) | Jump at last node (bp) |
|---|---|---:|---:|---:|
| B_without_positivity_collar | converged | 0.00e+00 | as B | as B |
| quantlib_convexmonotone_default_0.3_0.7 | converged | 8.01e-05 | 20.56 | 0.00 |
| quantlib_convexmonotone_paper_setting_0_1 | failed: 1st iteration: failed at 5th alive instrument, pillar December 24th, 2026, maturity December 22nd, 2026, reference date September 18th, 2026: root not bracketed: f[-1,1] -> [2.375741e-01,4.775343e-02] | n/a | n/a | n/a |

Hagan-West sectors used on the 22 intervals: i: 10, ii: 4, iii: 3, iv: 5. Positivity collar binding at 0 node(s).

Source: out/results.json, bootstrap.py, hagan_west.py.
