Data Playground

Our method, tested in public.

We validate trading signals and forecasts for a living. Here the same method runs on everyday questions, day and night: forecasts locked before the outcome, data saved before it changes, and every step committed where anyone can check it.

  • Locked before the outcome
  • Rules published first
  • Scored against data we don’t control
Live

What the lab did today

Read from public commits and GitHub runs every 15 minutes. No result appears here before its test is finished and checked.

Reading the pulse…

01

Three steps, every experiment

The same rules we use when a client asks us to judge a signal. Nothing here is tuned after the fact.

01 · LOCK

Before the outcome

Every forecast is saved, hashed and chained to the one before it, before the outcome exists. Nobody can change it afterwards, including us.

02 · MEASURE

Against independent data

Scored against measurements we don’t control, with rules published before the first data point and not changed while the test runs.

03 · COMPARE

With a lazy guess

A forecast only counts if it beats a simple baseline, such as “tomorrow will be like today”. Most people never check this.

02

What the lab is testing

Six investigations. Each one publishes its method and code, works only from public sources, and leaves a trace on GitHub at every step.

Experiment 01 · since 28 Sep 2026

Which weather forecast is right most often?

Allforecast downloads locked

DMI, MET Norway, OpenWeatherMap and the pilots’ own airport forecast, for five Danish cities, locked four times a day before the weather happens. Pilots don’t check a weather app: is the forecaster on duty better than the apps?

Day – · chain checked daily · first scoreboard around 28 October

Live status, stations and rules →
Pre-registered · since 2 Oct 2026New

Do offshore wind farms take the rain from the coast?

Locked before the rain data

A 2026 study projects 10–15 % less coastal rain after a huge build-out. We test today’s farms against 35 years of Danish and German rain gauges, every rule locked before the first rain value is read. On the map: all 2,848 turbines and every gauge in the test.

Weather model (ERA5), then the power check, then the test · no rain values read

Method, progress and the time machine →
Experiment 03 · since 30 Sep

Energy data, locked as published

Locked as published

Energinet’s CO₂ forecast every hour and the official price list every day, saved before they are overwritten. The picture: the electricity tax per kWh, 88 øre in 2019 and 0.8 øre in 2026.

Live status →
Experiment 02 · finished

Did the experts get it right?

60 forecasts scored

Ten years of GDP and inflation forecasts from the Government, the Nationalbank and the Economic Councils, against what happened and against a lazy guess.

See the scoreboard →
Open data · published 1 Oct

Where Denmark’s wastewater goes

36 of 4,195 overflows measured

Treatment plants, sewer overflows and rainwater outlets in every municipality, from the national data. Only 36 of 4,195 overflows report volumes from measured flow and concentrations.

Pick a municipality →
Pre-registered · finished

Farming’s share of the nitrogen

4–137 %

The usual figure is about 70 %. With open measurements and the method locked before the first value, more farmland clearly means more nitrogen, but the share itself can’t be pinned down.

See why →
04 · DATA ACCESS

Want to explore the data yourself?

The scoreboard will be public. Access to the full data is by application: browse every forecast against the measurement, hour by hour, and check any download against the hash chain. It is for students, journalists, researchers and anyone curious.

Applications open together with our client portal. Email us with “Playground” in the subject line and we will tell you when. We only use your address for that.

Email validation@kvantix.tech

Independent validation of signals and forecasts. An independent company registered in Denmark.

Kvantix · CVR 46296036 · Denmark · Servers in the EU© 2026 Kvantix