NewLaunch offer: 50% off for life with code SHARP50. Claim it

Research

Show the work. All of it.

Every call HalfSharp makes comes from a method that was tested before it was sold. Here's how we test, what passed, what failed, and how the live record is going, in plain English.

How we test an edge

  1. 1

    Write the test down first

    Before we look at any result, we commit the question, the data, the seasons and the bar a method has to clear. That stops us from moving the goalposts once we see the numbers.

  2. 2

    Build on old seasons, test on unseen ones

    Each model is fit on one stretch of seasons and graded once on seasons it never saw, usually both an older stretch and the most recent one. If it only works where it was built, it doesn't work.

  3. 3

    Correct for trying lots of ideas

    Test 40 ideas and one will look great by luck. We adjust for how many we tried, and treat a lone standout as noise until it proves otherwise.

  4. 4

    Check it against the market

    A real edge should beat the closing line, not just win games. We check that in testing and track it live on every call.

  5. 5

    Track it live before calling it proven

    Passing a backtest earns a method a public, timestamped live record. New ideas start as research plays, graded in public but never sold as bets.

Articles

What we found.

The live, always-current record is on the Results page. For the method in brief, see How it works.

Backtest figures are hypothetical, not results of actual bets, and past results don't predict future results. HalfSharp is an analytics and information service for adults 21+. Gambling problem? Call or text 1-800-GAMBLER.