Best Plays
What it is
Best Plays is the filtered top of the SharpSharks funnel. Out of thousands of edges the v4 engine grades each refresh, only a small fraction survive the Best Plays filters: minimum EV, minimum sharp consensus, tier requirement, no obvious traps. If you only check one page per day, this is it.
When to use it
- You want a curated shortlist, not a wall of every market on every book.
- You want picks that already passed multiple sanity checks (model agreement, sharp book presence, calibration).
- You are short on time and want the highest-confidence edges of the day.
How to read it
| Field | Meaning |
|---|---|
| Tier | A (highest confidence), B (solid), C (speculative). Drives sizing. |
| EV% | Expected value as a percentage of stake. EV% = (fair_prob × payout) - 1. |
| Sport / Market | What you are betting on (NFL spread, NBA total, MLB ML, prop, etc.). |
| Selection | The exact side / line / price. |
| Book | Where to place the bet. Soft books are where the largest edges usually live. |
| Sharp Consensus | The Pinnacle / Circa / LowVig blended fair price. Your anchor. |
| Time to Game | Markets tighten near tip. Earlier ≈ bigger edges, but more line movement risk. |
Worked example
Example
NBA: Celtics -4.5 -108 (FanDuel)
Sharp consensus (v4 blended): Celtics -5.0 -110. Fair prob ~54.2%. Implied at -108: 51.9%. EV ~+4.4%. Tier B (model agreement, multiple sharp books anchor).
Selection: Celtics -4.5 Book: FanDuel @ -108 Sharp anchor: Celtics -5.0 @ -110 (Pinnacle 45%, LowVig 30%, Circa 25%) Fair prob: 54.2% Implied: 51.9% EV: +4.4% Tier: B Size: 1u
You are getting a half-point and 2 cents better than the sharp anchor. The model agrees. Place it.
Common mistakes
- Sizing every Best Play the same. A-tier and C-tier are not the same bet. Use the tier system.
- Chasing the highest EV%. A 12% EV on a Polymarket-only longshot beats nothing. Read the books column.
- Ignoring the time-to-game. An edge with 8 hours to first pitch will move. An edge 20 minutes pre-game probably won't.
- Forgetting to log the bet. If you do not log it in Tracker, you cannot measure CLV. CLV is the only honest scoreboard.
- Stacking correlated picks. Two Lakers props on the same night are not two independent edges.