A standardized stake size, typically 1% of bankroll, used to communicate bet size without revealing dollar amounts.
The unit convention solves a coordination problem. A picks service publishes a wager to a subscriber base whose bankrolls range from a few hundred dollars to six figures. If the service publishes the stake as $100, the small-bankroll subscriber is over-risked and the large-bankroll subscriber is under-risked. If it publishes "1 unit," each subscriber sizes the bet to their own bankroll.
The standard convention is that one unit equals 1% of bankroll. A subscriber with a $1,000 bankroll bets $10; a subscriber with $50,000 bets $500. Unit sizing also enables clean comparisons across services and across time — a 5-unit week is unambiguously a strong week, regardless of dollar amounts.
Units come in fractional sizes for variance management. A high-confidence pick might be staked at 1.5 or 2 units; a lower-confidence pick at 0.5 units. Reputable services tie unit size explicitly to a confidence tier or implied edge band, not to gut feel.
The honest practice with units is to size based on edge, not on past performance. A service that retroactively grades cold streaks as half-unit losses and hot streaks as full-unit wins is cooking the books. The Signal Labs diamond tier system pre-commits to unit size at the moment of publication — 3 diamonds, 4 diamonds, or 5 diamonds — and the size is locked before the result is known.
Our diamond tiers map directly to unit sizes — 3 diamonds = 1u, 4 diamonds = 1.5u, 5 diamonds = 2u — and the mapping is fixed before the pick is graded. Track-record units are the truth; W/L percentage is cosmetic.