← Back to Number Lab

Methodology

Exactly how Lotto Pick HQ turns official draw records into the charts and Smart Picks you see on the Number Lab — and where the limits of that analysis lie.

We believe statistics are only trustworthy when the method behind them is open. This page documents every step between the raw lottery results published through the public feed and the numbers rendered on our homepage. Nothing here is proprietary or hidden: if you had the same public dataset, you could reproduce every figure yourself.

1. Where the numbers come from

Powerball and Mega Millions results come from official public data distributed through the Socrata platform. A scheduled job fetches each game's JSON feed once per day after drawings are certified. For every historical drawing we store the draw date, five white balls, and special ball. Our Data Sources page explains the fields and update cadence.

2. How the data is cleaned and stored

Raw open-data rows are normalized before storage. We resolve columns by their field names rather than by position (so a change in column order never corrupts a record), validate that every white ball falls within the game's legal range and that the special ball is valid, and sort the five white balls in ascending order for consistency. Records are written to a MySQL table keyed on (game, draw_date), which means re-running the importer never creates duplicates — it simply updates any row that changed. Draws are always kept in chronological order, oldest first, because our "overdue" math depends on that ordering.

3. The core statistics

Every visualization on the Number Lab is derived from the stored draws using plain, well-understood arithmetic. Here is each one.

White-ball frequency

We count how many times each numbered ball (1 through 69 for Powerball, 1 through 70 for Mega Millions) has appeared across the entire history. That count is the bar height in the frequency chart and the shade in the heatmap. Because five balls are drawn per game, over a long history each number's expected count is roughly (5 / poolSize) × totalDraws. Real counts scatter around that expectation; the scatter is normal random variation, not a signal.

Overdue (gap since last seen)

For each ball we record the index of the most recent draw in which it appeared, then compute the gap as the number of draws that have happened since. A ball drawn in the latest game has a gap of zero; a ball absent for 40 games has a gap of 40. This is a factual description of the past. It is not a countdown — a large gap does not raise the ball's probability on the next draw.

Sum distribution

For every draw we add the five white balls together and place that total into a histogram bucket. The distribution is bell-shaped and centered near the midpoint of the possible range, because there are simply many more number combinations that add up to a middling total than to an extreme low or high one. This is a property of arithmetic, not a strategy.

Special-ball frequency

The same counting method applied to the smaller special-ball pool, drawn from its own machine independently of the white balls.

Top number pairs

For each draw we look at all ten unique two-ball combinations among its five white balls and tally how often each pair has co-occurred across history. The most frequent pairs rise to the top of the chart. With thousands of draws, some pairs inevitably appear more than others purely by chance.

Average sum over time

We group draws by calendar month, average each month's winning-line totals, and plot the monthly means as a trend line. The line hovers close to the long-run average and wanders only within the bounds of ordinary randomness — there is no upward or downward "momentum" to ride.

4. Smart Picks

The Smart Pick generator samples numbers according to the mode you choose (Hot, Cold/Overdue, Weighted, or Quick Pick). The precise algorithms are documented on How Smart Picks Work. To keep a given day's featured picks stable for every visitor, we seed the random generator with the latest draw date, so the numbers only change when a new drawing is added.

5. Sample size and confidence

People often ask which numbers are "statistically hot." The honest answer is that, in a fair lottery, no number is genuinely hotter than another. If you ran a formal test — for example a chi-square goodness-of-fit test against a uniform distribution — a legitimate lottery's frequency counts would almost always be consistent with pure randomness. The apparent leaders on any frequency chart are the expected peaks of random noise, and they reshuffle over time.

The single most important caveat: every statistic here describes what has already happened. None of it predicts, improves, or influences a future draw. Each drawing is an independent event in which every valid combination is equally likely.

6. Reproducibility and corrections

Because we key records on game and draw date and pull from an official source, our history should match the state's published results exactly. If you ever spot a discrepancy between our data and an official result, please tell us through the contact page so we can investigate and correct it.