Every verdict starts at zero weight — in the shadows, watched but unheard. Its conviction rises only as its out-of-sample, overlap-adjusted accuracy justifies it, and it is capped at three. A confident-but-wrong call quietly silences itself.
The moment a verdict is issued it’s written to an immutable ledger, with its horizon and expected move fixed. Nothing is quietly revised after the fact.
When the horizon elapses, the call is graded on what actually happened — declustered so overlapping windows can’t inflate a lucky streak into false skill.
Only a genuine, repeated, out-of-sample edge lifts a verdict toward the cap of three — the chairman’s vote, given exactly the authority it has paid for.
Tulion measures every name against its sector, so a verdict isn’t washed out by a broad rally or panic — it isolates the part of the move a company earned on its own. The same call is issued across a week, a month, and one, two, and three quarters, because conviction that holds at one horizon rarely holds at all of them.
ADANIPOWER · conviction by horizon — strongest at one quarter, unproven at three.
Query a name, a sector book, or the ranked universe. Every response carries the verdict, its earned weight, the attribution, and the track record behind it — so your systems can trust it exactly as much as it has proven.
Request early access# best bets in Energy, this quarter { "sector": "Energy", "horizon": "1Q", "book": [{ "symbol": "ADANIPOWER", "bias": "LONG", "rel_move_pct": 4.1, "conviction": 2.2, // of 3.0, earned "hit_rate_90d": 0.581, "tradeable": true }] }