How I Built a 1,472-Draw Lotto Archive

Kia ora! A lot of you ask where my numbers come from. Is it a magic box? A crystal ball? Nope - it's a database of 1,472 NZ Lotto draws, and I want to show you exactly how it's built, and where it falls short.

Where the data comes from

Every NZ Lotto draw is public record. I pull results from official sources - Lotto NZ's published results and the public results archives - covering every draw from late 2018 to today. That's two draws a week, Wednesday and Saturday, for over seven years.

Each record contains: - The 6 main balls (1-40) - The Powerball (1-14 since 15 September 2026; 1-10 before that) - The bonus ball and Strike numbers

How I keep it honest

The most important rule I follow: no future data leaks. When I backtest a draw from 2023, I only use the draws before it. If I peeked at the result and then "predicted" it, the whole thing would be a lie.

That's why you'll see honest numbers on my site - like the fact that our 30 lines hit a prize in about 77 of the 100 most recent draws (mostly small Division 5-7 wins), and plenty of weeks where nothing hit at all.

What the data can't tell you

Here's the part I want to be fully transparent about: the data can't predict the future. It describes the past - beautifully, in detail - but each new draw starts from zero. The hottest ball in history has the same chance as the coldest.

What the data can do is give you context: which numbers are popular, how streaks behave, what's statistically normal. That's valuable - as long as you know the difference between context and prophecy.

I update the database after every draw, so the numbers you see are always current. Want to explore them yourself? Dive into the full 1-40 frequency explorer ->

A deeper look at the pipeline

Let me walk through what actually happens from draw night to the numbers you see on this site, because transparency matters more than any fancy claim.

Step one: capture. After each Wednesday and Saturday draw, the official results are published. I pull the six main balls, the Powerball, the bonus ball and the Strike numbers from the public results, along with the draw number and date.

Step two: validate. Every record is checked against the official publication. If a ball is missing or a draw number looks wrong, the record is rejected rather than guessed at. A database full of confident errors is worse than an empty one.

Step three: store. Each clean draw joins the archive. The archive now runs from 2001 to today: 1,472 draws, more than 8,800 main balls recorded, each in its own slot.

Step four: analyse. The statistics you see - hot numbers, cold numbers, appearance percentages - are computed from this archive on demand. When you open the explorer, the page is reading the same database the backtests use.

What the data genuinely cannot do

This is the part I want every reader to engrave somewhere: historical frequency has no predictive power. The archive describes the past exactly and cannot say a single thing about the future. Every draw is independent - the machine resets to zero each Wednesday and Saturday.

I have made peace with this. In fact, it is the reason you can trust what I put together. A lottery predictor that promised certainty would be lying; something that shows you honest history is the only kind worth building.

A promise about the numbers

Every statistic on this site is reproducible. You could take the public draw archive, run the same counts, and get the same percentages. There is no hidden formula, no proprietary magic - just careful counting of public facts. That is the whole story, and it is a good one.

Why I document the limitations out loud

Talking about what the data cannot do is not modesty for its own sake - it is the foundation of trust. Anyone in this space has enormous temptation to overclaim: the audience wants magic, and magic sells. But every overclaim I have seen in this space collapses the moment someone checks it against a real draw archive.

That is why the limitations section is not buried at the bottom of this article - it is the second thing a reader meets. If you understand what the data cannot do, you can use what it can do without being fooled. You can enjoy the frequency charts, respect the honest track record, and walk away from any site that promises more.

The same discipline applies to the backtest archive: every draw, every line, every result - including the empty weeks. Reproducible, auditable, and honest. That is the whole product, and it is the only product worth having in a game where the odds never change.