The Mechanics Behind Tracking Remaining NC Lottery Prizes
I spent about three years running scripts against North Carolina's published prize distribution files before I stopped treating this like a lottery hack and started treating it like data reconciliation. The core idea is simple enough: the NC Lottery publishes a daily report showing total prizes by tier and denomination for each active scratch-off game. You compare the current state to the launch state, and whatever is left on paper is what remains to be claimed. The phrase itself comes from a small community of scratch-off trackers who got together around 2019 after the NC Lottery made its published game files much easier to download. Before that, you were scraping PDFs. Now everything is CSV. The concept isn't revolutionary, but executing it correctly requires understanding how the game files are structured and where they lie to you. Every scratch-off game has a published total print run. North Carolina lists this as total tickets produced, total prizes by tier, and the face value of each prize level. A $30 ticket might have 1,000,000 printed with one $1,000,000 top prize, twenty $50,000 prizes, and so on down to break-even tiers. When the daily report shows that only four of the twenty $50,000 prizes have been claimed, that means sixteen remain. Multiply remaining prizes by their values and you get the remaining liability on that game. That number moves every day as claims are filed.
The practical application is finding games where the remaining value per ticket still above cost. That's the only thing that matters mathematically. Everything else is noise. If a $30 ticket still has an expected value of $31.40 based on unclaimed prizes, you have a positive expectation. If it's $28.60, you're losing money by buying it. The math doesn't care how many millions are left on the board. Here is where most people get it wrong. They look at a game with a single remaining million-dollar prize and get excited without calculating whether buying every remaining ticket at that point is even financially viable. It usually isn't. A popular game might have 200,000 tickets left with a remaining top prize, but the remaining lower-tier prizes could be almost entirely claimed, dragging the expected value well below cost. The million is just theater at that stage. I learned this the hard way in early 2022. I found a $25 game that showed three remaining top prizes out of six originally printed. The daily report made it look like a goldmine. I ran the full projection and the expected value was only $18.40 per ticket because 87 percent of the middle-tier prizes had already been claimed by other buyers who weren't tracking the data. I almost bought $2,000 worth of tickets that day and would have lost roughly $720 on average. I walked away instead.
The actual workflow breaks down into about five steps. First, download the current day's prize file from the NC Lottery website. The URL structure stays consistent enough that you can automate this. Second, download the original game description file for whichever tickets you are evaluating. Third, match prizes by tier ID and calculate claimed versus remaining counts. Fourth, divide remaining prize counts by remaining tickets to get remaining probability per ticket. Fifth, multiply those probabilities by prize values to get expected value. I wrote a Python script that does steps one through four in about forty-five seconds. It pulls the files, merges them on game code and prize tier, computes the deltas, and outputs a spreadsheet sorted by expected value per dollar spent. I run it once every morning around 7 AM because the NC Lottery updates their files somewhere between 5 AM and 6 AM. The script runs on a cheap VPS that costs about six dollars a month. The most important nuance that nobody talks about is the gap between when a prize is won and when it is reported as claimed. The NC Lottery files reflect claims, not wins. If someone wins a prize on a Tuesday and doesn't claim it until the following Monday, that prize still shows as unclaimed in the Tuesday file. This creates a lag that can make a game look more valuable than it actually is. During my tracking period I saw this lag typically range from zero to fourteen days, with a median of about three days. In rare cases, like a large prize won at a retail location over a holiday weekend, the lag stretched to twenty-one days.
Get the Full Details
Another thing people miss is that the NC Lottery sometimes retires games earlier than planned. When a game is retired, no new tickets enter circulation, but old tickets can still be submitted for claims for years. The published file doesn't always flag a retired game clearly. I learned to check the retire date field separately because a game can still show a healthy remaining value while actually having zero new tickets in the wild. If you are hunting for unclaimed prizes in a retired game, you are only looking at tickets that already exist somewhere in a drawer, not tickets that will be sold at a store. There is a second-order problem with large prizes that changes the math entirely. When a high-value prize gets claimed, it does not get replaced. The total prize pool shrinks permanently. This means the expected value of remaining tickets drops not just because a prize was removed, but because the total number of tickets in play stays the same while the prize count decreases. A single claimed $50,000 prize on a game can wipe out the positive expectation on every remaining ticket if the remaining pool is already thin. I track about forty active games at any given time. On an average day, maybe two or three of them show a positive expected value above cost. That is the real number, not the flashy headlines. Some months I find zero games in positive territory for weeks at a time. The market is efficient enough that most games hover at negative expectation, sometimes significantly so.
The tools available fall into three categories. There are pre-built trackers like Lottery Outlook and ScratchOffData that do the basic calculations for you but charge monthly fees and don't let you customize your filters. There are the CSV files themselves, which anyone can download and process with a spreadsheet or script. Then there are the custom solutions, which is where my script lives. If you are serious about this, building your own pipeline is worth the two weekends it takes. The pre-built tools are fine for casual checking but they lag behind the daily files by several hours sometimes, and they rarely handle the edge cases I described above. I also want to address the elephant in the room. Even when you find a game with positive expected value, the variance is enormous. A positive EV game is still a gambling product. You could buy $500 worth of tickets from a positive EV game and walk away with nothing. The math works over thousands of tickets, not over a single shopping trip. I have personally seen my bankroll drop 40 percent in a single weekend despite only playing positive EV games, purely from bad variance. That is why bankroll management matters more than finding the games in the first place. If you want to start with something free, grab the raw files from the North Carolina Education Lottery website, go to the Scratch Off section, and download the game files for whatever month you are interested in. The files are labeled with game codes like GO123 or SC456. Cross-reference them with the daily prize reports, which are dated and labeled by report number. If you don't want to build a script, a Google Sheet with a few VLOOKUP formulas will get you through maybe ten games before it becomes painful. That is when you know it is time to write something proper.
The bottom line is that this works, but it works within tight constraints. You need a reliable data pipeline, you need to understand reporting lag, you need to accept that most days will show zero actionable games, and you need to treat this like a side business with real variance rather than a shortcut to wealth. Anyone telling you otherwise is either selling you something or hasn't been doing it long enough to see the bad streaks.
