VALTAVAHTI ← Back to front page

About & Methodology

An independent political transparency platform — open data, transparent sources.

Valtavahti (valtavahti.fi) · Updated: 20 July 2026 · The service itself is in Finnish; this page is an English summary for international readers.

1. What is Valtavahti?

Valtavahti ("Power Watch") is an independent, free transparency service that tracks the Parliament of Finland (Eduskunta) and its 200 members. It presents the public record of democratic decision-making — votes, attendance, legislative initiatives, speeches, lobbying meetings, declared interests and campaign finance — in an approachable interface: politicians have a promise-tracking score, player-style cards, comparison tools and leaderboards.

The service requires no registration or login, uses no cookies, and collects no personal data about its visitors (see the privacy statement, in Finnish).

2. Data sources

All politician data comes from public, official sources:

Most datasets are refreshed automatically every night; registers based on periodic reporting (such as lobbying disclosures) follow their own reporting cycles. Each view in the service shows its source and latest update date in a footer row. The full machine-generated source list is published at valtavahti.fi/lahdeluettelo.html and the complete calculation methodology (weights, formulas, known limitations) at valtavahti.fi/metodologia.html — both in Finnish.

3. Key metrics and how they are computed

Activity score (Aktiivisuus)

A points total out of 230, built from five fixed-scale components: plenary voting participation (max 60), legislative initiatives (max 50), plenary speeches (max 50), media mentions (max 40) and absences (max 30, fewer absences = more points). The Speaker of Parliament is structurally excluded from voting and absence scoring, and this is labelled in the interface.

Overall rating (OVR, politician cards)

The player-card style overall rating is a weighted composite: voting participation 55 %, lobbying exposure (inverse) 20 %, legislative initiatives 15 % and absences (inverse) 10 %. Stats use fixed scales — MPs are scored against the scale, not against each other.

Ethics risk estimate (Etiikkariski)

A 0–6 point heuristic combining three signals: register cross-check findings (max 3 pts — discrepancies between the declarations of interests and the Transparency Register or business register, the strongest signal because it rests on two official registers), unexplained campaign finance gaps (max 2 pts — a large difference between reported 2023 election funding and expenses), and a combined high-lobbying-plus-high-absence pattern (max 1 pt). Levels: LOW (≤2), ELEVATED (3–4), HIGH (5–6). The basis for each MP's score is always shown.

The ethics section additionally shows an informational media signal: the number of crisis-toned news items mentioning the MP in the last 30 days, detected with a weighted keyword vocabulary from the same public news feed as the media-mentions view. It is deliberately not counted into the score, because keyword detection cannot distinguish the subject of a scandal from politicians commenting on it; it exists to surface cases that official registers do not capture, with the underlying headlines always visible.

Promise index (Lupausindeksi)

Compares each MP's own answers to the 2023 voting-advice applications (published by Yle and Helsingin Sanomat) with how they actually voted in plenary sessions — in short, whether they keep their own promises. The comparison is made by a language model against an anchored rubric in which votes are weighted by significance, and the reasoning behind every score is shown. It is an interpretation, not a verdict. Party election programmes are not used for this index.

Implementation index (Toteutumisindeksi)

A separate, party-level index that answers two questions: how much of a party's 2023 election programme ended up in the government programme (measured by statistical text similarity), and whether there is evidence in parliamentary voting for those promises being implemented (assessed by a language model, promise by promise). It covers the governing parties. Text similarity does not prove causation, and the model's assessment is an interpretation whose basis is always shown.

Group discipline (Ryhmäkuri)

Measures how often each MP votes against their own parliamentary group's majority position, computed from the full plenary voting record.

Standing and form (Poliitikkopörssi)

A sports-style composite: an MP's standing derives from their promise index, activity, attendance, independence and experience, and their form from recent momentum. MPs are grouped into five types (star player, veteran, riser, core player, underperformer). It is a presentation device for real underlying metrics — nothing is traded and no money is involved.

Conflict-of-interest observations (Hillotolpat)

Automated cross-referencing of declared interests against the Transparency Register, campaign finance data and the business register. All findings are probabilistic name matches produced by software.

Important: automated observations are leads, not conclusions. They are never claims of wrongdoing. Every observation links to its underlying official records so readers can judge for themselves, and misidentifications are corrected on notice.

4. Corrections

If you appear in the service and find an error in your data, contact [email protected] — we correct errors to match the official source material. Readers can report technical bugs directly from the site.

5. Who runs this?

Valtavahti is built and maintained by Tuomas Kuusisto. It is independent: not affiliated with, funded by or endorsed by any political party, organisation or media outlet.

General contact: [email protected] · Privacy: [email protected]