Test it before you bet it.

Where we are right now

Crak'd is in beta. That is not a disclaimer we are hiding behind, it is a description of what you will run into, so here it is plainly.

• The engine is the finished part. The maths is tested against a reference implementation on every build, and it does not look at the future. What it cannot fix is the data underneath it: public box scores contain errors and gaps, and a correct calculation on a wrong row is still wrong. When we find one, we rebuild the season.
• The parser is the young part. It reads a lot of plain English and it will still fail on sentences that seem obvious to you. When that happens the tool says so rather than guessing, and there is a box to tell us what you meant. That is the single most useful thing you can do for it.
• Leagues come and go. We switch a league off rather than leave it half working, so the list on the homepage is what genuinely runs today, not what we hope to have.
• No historical betting lines yet. Every bet is priced at one number you choose, not the price a book actually offered on the day. This is the biggest gap between Crak'd and a finished product, and we say so on every result.

Why we built it

Almost every betting tool sells confidence. Picks, locks, plays of the day, someone's record in a screenshot you can't verify. Very little of it tells you how an idea has actually performed over hundreds of games, and none of it lets you check your own.

We wanted the opposite: a place to take the hunch you already have and put a number on it before any money is involved. Most hunches don't survive that, which is exactly the point. Finding out an idea doesn't work is worth more than being told one does.

What we do

• Read plain English. Describe the bet the way you'd say it out loud. We show you how we read it before it runs, so you can correct us.
• Run it against real games. Real box scores, from public data sources we are allowed to use, priced at whatever odds you set. Every league names its source on its own page.
• Test sides and totals, not just props. Spreads, moneylines, game totals and team totals, graded against final scores at the number you choose. Without historical lines this answers "how often did it clear this number", not how it did against the line a book actually posted.
• Never peek at the answer. Conditions only ever use games that happened before the one being bet. A backtest that can see the future is worthless.
• Compare against the base rate. We show whether your conditions actually changed the odds, or whether the bet just hits that often for everyone.
• Say when the sample is too small. A 70% hit rate across nine games is noise. We label it rather than let it look like an edge.
• Tell you where the data is weak. Every league lists its source, how often it refreshes, and the eras where the sport itself changed enough to make old seasons misleading.

What we don't do

• We don't give picks. No plays of the day, no bet slips, no telling you what to back tonight.
• We don't predict. Everything here is a record of what already happened. The past is not a forecast, and a strategy that worked for five years can stop working tomorrow.
• We don't take bets or handle wagers. We are not a sportsbook and we are not affiliated with one.
• We don't sell your data or take money to feature a book.
• We aren't affiliated with any league or team. The NFL, NBA, MLB, WNBA and NCAA names and marks belong to their owners. We name them to say which games the data covers, nothing more.
• We don't do parlays yet, on purpose. Legs from the same game move together: a quarterback throwing for 300 makes his receiver going over far more likely, and the two are not independent bets. Multiplying the individual hit rates, which is how almost every parlay tool does it, quietly overstates the edge in exactly the spots people care about most. We would rather not ship it than ship a number that flatters you. When we do, it will account for that correlation.
• We don't pretend the house edge isn't real. At -110 you need to hit 52.4% just to break even, and at -130 it is 56.5%. Every result also tells you the price it would need to profit, so you can compare it against the number your book is actually showing. Test enough ideas and one will look good by luck alone. We would rather say that than let a green number imply otherwise.

Where the money goes

Subscriptions are the only thing that funds Crak'd. No sportsbook pays us, nothing here is featured because someone bought the slot, and there are no investors whose returns have to come out of your results. That leaves one way to stay in business: build a tool worth paying for. A subscription buys access to it, and nothing else. What subscriptions pay for:

• Better programming. A smarter parser that understands messier questions, faster results, and more ways to slice a situation.
• Licensed data. Some of the best sports data is not free, and some free data can't legally be used in a paid product. Paying for proper licences is how leagues get added honestly and how live in-season data reaches sports that only have history today.
• Expansion. More leagues, deeper history, and eventually real historical betting lines, so a result reflects the price offered on the day rather than one price applied to every game.

They also pay us for the work. That is the whole model, and we would rather write it down than let you assume a product like this funds itself.

Please gamble responsibly

This is a research tool, not betting advice, and nothing here is a promise of profit. Only bet what you can afford to lose. Crak'd does not take bets and is not a sportsbook. If gambling has stopped being fun, help is available: in the US, call or text 1-800-MY-RESET (1-800-697-3738), or visit ncpgambling.org.