Kevin Brennan

Writes for Lotto Edge AI

About

My coworker Jarrod ran our office Powerball pool and picked numbers off the break room birthday board every week. I'm a data analyst. Four months in, I built a spreadsheet to test whether birthday-based picks were worse than random. They weren't meaningfully worse. They weren't better either. The whole thing is pretty much random, which is the point. But now I had a Powerball frequency tracker and I kept running it.

That's been going for about two years. The spreadsheet has a frequency tab, a picks log, a draw-history sheet covering Powerball and Mega Millions back to when I started, and a hit-rate column I update Wednesday and Saturday nights after the draws close. My wife has named it the Spreadsheet, capital S, with a tone she insists is affectionate. The whiteboard on the back wall of the home office has running tally columns by game. Updating them with a dry-erase marker late at night is a specific kind of quiet: overhead light off, just the monitors and the desk lamp.

About a year into the tracking project, I started noticing ads for AI lottery tools: platforms claiming to run pattern detection on historical draw data and surface better number selections. Bought three. Each one got at least 30 draw cycles before I wrote anything: pick suggestions logged before each draw, actual results in the next column, hit rate calculated with no early stops when results looked promising. One tracked about four percent above random selection over eight weeks. One came in roughly two percent below. The third was statistically indistinguishable from a random number generator with better graphic design.

Data analysis is the day job and the methodology here. Powerball and Mega Millions use gravity-mixing machines built by Smartplay International, with ball sets and machines selected at random before each draw; results are independently audited. The expected value on a lottery ticket is negative regardless of how the numbers were chosen. What the tracking produces is an honest picture of whether a given platform's picks outperform chance over a real sample. Sometimes they don't by a notable margin. Sometimes the margin is small enough that the data is flat. Both outcomes go in the review without adjustment.

Content here is for entertainment and informational purposes only. Financial decisions around lottery play should stay within your actual means. If gambling is causing harm to you or your family, the National Council on Problem Gambling helpline is 1-800-522-4700 (US, free, confidential, available any time).

Articles by Kevin Brennan

Disclosure

Some outbound links here are affiliate links, meaning I get paid a small commission when you buy through them, at no extra cost to you. I don't collect or store any data about which links you click. For more on how the editorial side handles affiliate relationships, see my Editorial Policy.