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

Research

Month one: the live record.

Our first month of live, timestamped calls, Oct 16 to Nov 13, 2026. What went right, what didn't, and why a month is far too short to judge either way.

Precision Sports Mathematics · Nov 14, 2026

The short version

  • Halftime BETs: 16–10, +21.09 units. College football 11–4, NFL 5–6.
  • November was rough: 5–6, −6.74 units, after a strong October.
  • Everything else is near even: basketball leans 14–14, the deep dive 11–9, opening-line calls 0–1.
  • 26 graded bets is not a verdict. Judge us over hundreds of plays, and on the closing line first.

This is a snapshot from Nov 14, 2026. The live, always-current record is the Results page, and it's the source of truth.

How to read these numbers

  • Units at size. Halftime BETs are 2 to 5 units by how big their edge is (5 units before Nov 4), the deep dive's best bet 3 units, everything else 1 unit. A 5-unit win counts five times a 1-unit win.
  • Halftime calls are graded at −115, the extra juice books charge at the half (break-even 53.5%). Pregame calls at the price posted when they were made.
  • The halftime record starts with three "Replay" calls from Oct 15–17 (2–1, +3.7 units): what the updated college football model would have called, replayed with its settings frozen. Everything from Oct 19 on was called live.
  • Each product has its own record, so one can't prop up another.

The numbers

Halftime BETs (the main product)16–10 · +21.09u

About +17% return on units risked. College football 11–4, NFL 5–6. November: 5–6, −6.74u.

Basketball halftime leans14–14 · −1.83u

1 unit each. Labeled LEAN, never BET, because they never beat break-even in testing. This is about what that predicts.

Daily deep dive11–9 · +1.27u

New and unproven, on its own record. November: 6–7, −6.73u.

Opening-line calls0–1 · −1u

College basketball just started. These calls wait until both teams have played four games.

The board checked 796 halftimes in the month and made 35 BET calls. Saying no about 95% of the time is the method, not a slow month.

The honest read

October went well and November gave a chunk of it back. Neither tells you much. The college football model's own testing had 4 of 19 seasons go 37–48%, and a stretch of 11 bets can swing either way on luck alone. A record this short can't prove an edge, and it can't disprove one.

What it does show: every call was timestamped before it was graded, the losses are on the board next to the wins, and the products we label unproven (leans, the deep dive) are behaving like unproven products, near even.

What we're watching

  • The closing line. If our number keeps beating the market's final number, the edge is real long before wins and losses settle. Our research plays have beaten the close on 41 of 75 so far, matched it on 14 and trailed on 20. Only 2 recommended plays have a closing line yet, too few to read.
  • Hundreds of plays, not dozens. At 1–3 college football halftime calls a week, a fair verdict on the main model takes at least a full season.
  • Opening-line basketball, which should add about 15–25 calls a week once both seasons are going, and will be graded on closing-line value from day one.
  • The research ledger: 47–49 so far, −2.28 units. These ideas are graded in public but never recommended until they pass a test on seasons they weren't built on.

Related research

See every live call on the Results page, or how the whole system works on How it works.

Past results don't predict future results, and no call is guaranteed. Figures are a snapshot of the live record on Nov 14, 2026; the Results page is current. HalfSharp is an analytics and information service for adults 21+; we don't accept wagers. Gambling problem? Call 1-800-GAMBLER.