Most people build a Same-Game Parlay by picking 3 legs they individually LIKE, then hoping they're correlated. That's backwards, and it's why most SGPs lose even when the "reasoning" sounds smart.

Here's the fix — a 90-second checklist I run before every parlay:

1. Pick your anchor first.

Before touching individual legs, write one sentence describing how you think the game plays out (the "game script"). Example: "Home team leads big, forces garbage-time passing for the trailing team." That sentence is your anchor.

2. Cascade from the anchor — don't shop for unrelated legs.

Every leg in your parlay should be something that becomes MORE likely specifically because your anchor happened. If a leg would hit regardless of your game script, it's not adding correlation — it's just adding variance.

3. Watch for legs that compete for the same resource.

This is the #1 silent killer. Two running backs on the same team both going Over rushing yards sounds "correlated" (same offense) but they're actually competing for the same finite touches. That's negative correlation disguised as positive. Same logic applies to two WRs chasing the same targets, or two bench guys chasing the same garbage-time minutes.

4. Only bet it if the payout undervalues how tightly linked the legs are.

Sportsbook SGP pricing engines simplify correlation at scale. If your legs are all driven by the exact same mechanical outcome (one game script), the true joint probability is usually higher than what the odds imply. If they're only loosely related, the book's price is probably fine — pass.

That's it. Anchor → cascade → resource check → price check. Most "sharp" parlay building is really just this checklist done consistently instead of once in a while.

Happy to break down a specific parlay idea in the comments if you want a second set of eyes on the correlation logic.