Transparency
How scoring works
Every call on Stoa is graded by the market, not by us. The score is math on locked entry prices, resolved outcomes, and benchmark alpha. No manual edits. No reputation votes.
The loop
- An analyst publishes a call with ticker, direction, optional target, and horizon.
- Stoa locks the entry price from the live feed and captures SPY as the benchmark.
- When the horizon ends, the grading job fetches resolved prices and assigns an outcome.
- The Track Score (0-100) updates on the analyst profile.
Outcomes
HitMoved with the call and reached target (or +5% if no target).
NearMoved with the call but fell short of target.
PartialRoughly flat within the neutral band.
MissMoved against the stated direction.
OpenStill within the published horizon.
Track Score (0-100)
- Win rate: Wilson lower bound on time-weighted outcomes (recent calls count more).
- Profit factor: decay-weighted average win divided by average loss.
- Alpha: excess return vs SPY over the same window (needs at least five benchmarked calls).
- Consistency: penalties for miss streaks and outcome drawdowns.
- Sample confidence: a logarithmic ramp discounts small sample sizes, so a six-call streak and a sixty-call streak are never shown with the same confidence. Under 10 resolved calls, the score is marked provisional wherever it is displayed.
Sample confidence bands
0-9 callsProvisional. Score shown, but marked as early sample.
10-29 callsFull confidence label unlocks. Edge still forming.
30-74 callsMeaningful sample. Win rate and profit factor stabilize.
75+ callsLarge verified sample. Score moves slowly with each new call.
What we deliberately ignore
Firm prestige, follower count, media appearances, and self-reported win rates. Only verified calls that went through the lock-and-grade pipeline count.