Media
The research behind the dashboards in two watchable and readable formats: movies, daily 4K time-lapses of twenty-one years of the Polish sovereign curve, and shorts, two-page summaries of the underlying papers with one headline figure and one key table each.
Movies
jump to shorts ↓Daily 4K time-lapses of the Polish sovereign zero-coupon yield curve, ACM/BRW term premia, fit diagnostics and the NSS/SPF forecast family, plus a lexical-evolution time-lapse of NBP MPC minute language. Built from the same data behind the dashboards. Watch on YouTube for full quality.
Lexical evolution of NBP MPC minutes — 19-year time-lapse
2007-04 → 2026-04 · 212 meeting frames at 2 fps · 1m46s · 1080p
What you see on each frame. Top section: TF-IDF-weighted top terms aggregated over the trailing six MPC meetings — roughly six months of distinctive Polish central-bank vocabulary. Bottom strip: per-meeting net tone, (hawkish − dovish) / (hawkish + dovish), with the 3-meeting moving average as the heavy line and a navy playhead line that sweeps left-to-right marking the current frame's meeting. Background bands derive mechanically from the 6-month change in the NBP reference rate (green accommodation, salmon tightening, grey neutral). The interactive version of the same panels lives on the MPC tab.
Episodes to watch for:
- 2008-09 — GFC vocabulary forms
- 2012-15 — deep-cuts era
- 2020-03 — COVID vocabulary spike
- 2022-02 — Ukraine war language
- 2022-09 — inflation hike-cycle peak
- 2024-26 — post-hike pause
Polish Sovereign Yield Curve & Term Premia — 21-year time-lapse
2005-01-07 → 2026-04-21 · 5,290 daily frames at 30 fps · 4K
What you see on each frame. Top-left: LW-NSS-fitted zero-coupon curve with the underlying coupon-bond panel as bubbles sized by outstanding amount and coloured by curve segment. Top-right: Adrian-Crump-Moench and Bauer-Rudebusch-Wu term premia at 1y/2y/5y/10y horizons. Bottom-left: daily fit diagnostics — MAE in bp, number of bonds in the daily fit, NSS β and τ parameters. Bottom-right: forecast family — NSS 1-year forwards, ACM and BRW expected-rate paths, and the NBP Survey of Professional Forecasters implied path.
Episodes to watch for:
- 2008-09 — GFC
- 2013-05 — taper tantrum
- 2015-16 — ECB QE drift
- 2020-03 — COVID dislocation
- 2021-22 — inflation surge
- 2023-26 — normalisation
Polish Sovereign Curve, Term Premia & Liquidity — Last 5 Years
2021-04 → 2026-04 · ~1,300 daily frames at 30 fps · 4K
Same four-panel layout as the full 21-year version, zoomed onto the most recent five years. This is the most expensive window in twenty-one years of Polish term-structure data: the fastest steepening in the sample, a clean Greenwood-Vayanos-Vila supply-pressure episode at the long end, the survey-versus-model horizon split visible in real time, and a venue-cap-bound bid-ask spread that hides the actual liquidity stress in 2022. The 10-year term premium is the cleaner stress signal here.
YouTube ↗ Subscribe @yieldcartography ↗Both movies render the same daily slides served behind the curves and term-premia dashboards. New videos are uploaded as the dataset extends. Subscribe to the YouTube channel for notifications.
Shorts
back to movies ↑Two-page summaries of the working papers behind the yieldcartography dashboards. Each short carries a catchy title, ~200 words of setup, one headline figure, one key table, ~200 words of interpretation, and a 50-word "what this means for practitioners" line. Comments are open at the bottom of every short.
There is no simple recipe to beat the bond benchmark
Time TBSP duration on the term premium and a full-sample fit looks like a winning carry rule, information ratio +0.28. Estimate the same premium in real time and the identical rule flips to losing, the edge was look-ahead. The only honest survivor, a contrarian fade of the premium, fails a Hansen SPA correction (p = 0.47), loses two thirds of its return to costs, vanishes at monthly frequency, and reverses in the US and euro area.
Underlying analysis · Dec, M. (2026). Term-premium duration timing and the look-ahead trap · recursive ACM, security-level TBSP book, SPA multiple-testing, cross-market
Does the curve know the next MPC move?
An oracle from bias-corrected forwards and minutes tone. The ACM-BRW market path, the professional-forecaster survey, and the lexical tone of the MPC minutes combine into one next-move score. Inside the tightening band four months before the 2021 liftoff, deep in the loosening band before the September 2023 cut.
Underlying machinery · ACM-BRW term premia, MPC minutes lexicon, NBP Survey of Professional Forecasters implied paths · interactive Oracle tab
The Polish duration trade pays over cash, except in the 2022 storm
TBSP compounded at 4.14% per year since end-2006. Carry and roll-down contributed +413 bp per year on average; the macro-driven curve-move bloc averaged -17 bp per year and is statistically indistinguishable from zero. Condition out the four 2022 NBP-hike quarters and the TBSP-versus-NBP-deposit excess is +46.5 bp per quarter at HAC p = 0.003.
Underlying paper · Dec, M. (2026). Does the term premium pay for the duration? Evidence from Polish sovereigns. Research Square preprint v1, 1 June 2026 · doi:10.21203/rs.3.rs-9849665/v1
Why naive Nelson-Siegel underprices the Polish belly — and what to do about it
Equal-weight NSS leaves a 1.6 bp / day mean fit MAE on the table over twenty-one years of Polish sovereign data. A weight matrix derived from BondSpot turnover and outstanding amounts cuts that almost in half. Information-matrix derivation included.
Underlying paper · Parsimonious yield curve modeling in less-liquid markets (LW-NSS, FAME|GRAPE WP, 2021)
The pure expectations hypothesis dies in three different ways in PL, US and EA
Polish curves drift mildly anti-PEH. The US sits roughly consistent with PEH at long horizons. The euro-area AAA panel rejects strongly under asymptotic Newey-West but barely under the wild block bootstrap — the cleanest evidence for the Bauer-Hamilton (2018) concern in our sample.
Underlying paper · Closer to New York than to Frankfurt? (SSRN 6695444)
Are NBP forecasters informative? At 1y yes, at 3y and 5y no
Diebold-Mariano, Clark-West and forecast-encompassing tests on the NBP Survey of Professional Forecasters versus the ACM/BRW expected-rate path. The survey wins at 1y, the model wins at 3y and 5y, and encompassing rejects the survey at long horizons.
Underlying paper · Are Survey-Based Rate Expectations Informative? (SSRN 6644222)
Six microstructure measures, one composite, and the venue-cap puzzle
Bid-ask spread, zero-trading-day frequency, Amihud, Roll, Pastor-Stambaugh γ, Corwin-Schultz, and a composite z-index across the BondSpot panel from 2005 to today. Aggregate term premia absorb supply pressure that the venue-capped bid-ask spread cannot.
Underlying paper · Supply, habitat and the price of liquidity in less-liquid sovereign bond markets (Borsa Istanbul Review, 2026, open access)
Each short is one self-contained letter-format page. The underlying working papers are linked from the about page with full DOIs and abstracts.