How PUTKI HQ measures itself.
A publication that publishes numbers should publish the numbers about itself first. Below: three editorial principles, three live datasets refreshed on every read, and four long-form methodology pieces. Nothing here is marketing copy — every block links to a primary artefact.
Three principles
Source before opinion
Every number on this site is bound to a named public source or an internal document with a date and version that can be traced. No anonymous "insider tips".
Measurement before marketing
Mittari, The Read and Voita are presented as numbers and quartiles, not exclamations. When a back-test is not ready yet, we say so — we do not fabricate numbers.
Separated commercial layer
Affiliate links are clearly marked in every context. Sponsors do not influence editorial choices — tested twice during 2025.
Three live datasets
Fetched fresh on every page load. No fabricated rows. When the underlying ledger is empty, the page says so — never invents numbers to fill the table.
Diagnostics dataset
Anonymised quartiles + N for every completed diagnostic, across the three games. Refreshed on every read.
Voita raffles ledger
Every concluded Voita raffle with winner stamp + timestamps. Append-only.
Mittari accuracy back-test
Rolling 90-day per-class hit rate. Today shows "back-test in progress" — flips to live numbers automatically once a class crosses N ≥ 30.
Four methodology pieces
How The Read scores you
The 3-axis multi-question instrument, the Cronbach-α calibration cycle, what the result card claims and does not claim.
The 28 named sources
6 categories, published weights, the contested-signal handling rule, and how source-list changes are governed.
Voita raffle FAQ
10 questions on eligibility, draw mechanics, taxation and the publication’s position under Arpajaislaki §27.
Dioni Bouropoulos Q&A
8 on-the-record questions — why launch ahead of 2027, how PUTKI HQ makes money, what would shut it down.
Where the numbers fall short today
- Mittari back-test: currently in scaffold mode. Signals are being snapshotted daily; an editor grades them weekly. The public table flips to live numbers once any class crosses N ≥ 30 inside the rolling 90-day window.
- Diagnostics sample: currently small (~12 400 runs aggregated by Feb 2026). The quartile bars on the dataset page are real but the tails are sparse. Re-calibration runs quarterly with a published Cronbach-α threshold.
- Voita pseudonymity: winners pick how they appear in the ledger. The timestamp is the public integrity proof. Anonymous winners do not show a name — they show "Winner drawn" + the timestamp.
Five sharpest picks, every morning.
Join 525+ Finnish bettors · 6/7 hit last week · 18+