catch% and peak% — the two numbers that matter

Every signal BullAlert publishes is described by two numbers, scoped to a single (ticker, day, session) triple:

catch% — the price change at the moment we caught the ticker, measured against the correct session reference. Frozen at catch. Once written, it never updates.
peak% — the highest price the ticker reached in that session after we caught it. A running max, refreshed every five minutes, frozen at session close (with bar-resolution truth from the daily reconciliation pass).

The three sessions — pre-market (4:00–9:30 AM ET), regular hours (9:30 AM–4:00 PM ET), and after-hours (4:00–8:00 PM ET) — are independent. A pre-market signal that closes its session is not carried forward into regular hours. Each session opens with its own catch and peak.

Why two numbers, not one? Because catch% tells you how early the scanner caught the move, and peak% tells you how far it went. The gap between them — the catch-to-peak delta — is the only honest measure of edge. We sort every leaderboard by this delta, and you can verify it yourself on the public alert history.

Three sessions, three threshold sets

Pre-market, regular hours, and after-hours have different liquidity profiles, different average volumes, and different microstructure. A signal that's noteworthy at 4:30 AM is unremarkable at 10:30 AM, and vice versa. Rather than apply one threshold across the day, the scanner applies three session-aware sets — the concept matters more than the exact numbers, which we keep internal.

At a session boundary, every live row resets. Live candidates.catch_session switches to the new session; the historical per-session row gets frozen into candidates_daily_snapshot a few minutes before the boundary. There is no cross-session inheritance, no “today's catch” that spans pre-market into regular hours.

Sixteen internal scoring gates

Before a candidate becomes a signal, it passes through sixteen gates. Each gate measures one thing. Below is what each group checks — not the thresholds, formulas, or weights. The recipe stays opaque on purpose; the categories are public on purpose.

Liquidity & structure

Tradable session — the ticker is currently in an active trading session
Volume floor — enough turnover that prices reflect real demand, not a single retail order
Float guardrails — share structure is within a band where momentum can sustain without instant exhaustion
Price band — the universe is constrained to small-caps where catalysts move price meaningfully

Momentum & flow

Relative volume regime — today is structurally heavier than the recent baseline at this time of day
Momentum acceleration — recent move is expanding rather than fading
Session-aware change — the move is measured against the correct prior-session reference
Trend confirmation — multi-bar agreement, not a one-print spike

Quality & catalyst

Sentiment density — a meaningful number of independent social sources have surfaced the ticker recently
Catalyst presence — a fresh news/filing/event tied to the ticker, with sane recency
Scoring threshold — internal momentum score crosses a session-aware bar
Tier eligibility — the catch quality earns S, A, B, C, or D classification

Late-catch & fader rejection

Late-catch guard — the move has not already exhausted (no chasing local tops)
Fader rejection — patterns that historically reverse quickly are filtered out
Drawdown brake — early-cycle weakness disqualifies before it becomes a loss
Per-session cap — a finite number of catches per session prevents over-firing

Pattern grading (S / A / B / C / D)

Once a candidate clears the gates, the resulting signal is graded on a five-level scale based on how many confirming layers stack up. Grades are outcome-described, not recipe-described:

Highest conviction. Multiple catalyst signals, clean structure, strong session context.

High conviction. Catalyst plus structure, with one fewer confirming layer than S.

Medium conviction. Either catalyst or structure dominates, not both.

Low conviction. The signal crossed gates but the structure is thin.

Marginal. Surfaced for completeness; treat as watchlist context, not a setup.

What we don't do

We don't show charts, raw price feeds, or order-book depth.
We don't execute trades. We are not a broker.
We don't recommend buys, sells, exits, or position sizes.
We don't claim “win rates.” We don't trade — you do.
We don't use insider data, off-exchange flow, or anything that isn't public.
We don't predict gains. We measure session-truth catch% and peak%, after the fact.

What you see

The user-facing surface is intentionally narrow. For every signal:

Ticker and session (pre-market, regular, or after-hours).
catch% and peak% against the session reference.
Number of gates passed (out of sixteen) and tier grade.
Pattern labels if any cleanly applied (e.g., gap-and-go, VWAP reclaim).
Catalyst tag when a fresh news/filing event was tied to the catch.

That's it. We don't show our internal scores, weights, thresholds, or vendor stack.

Verify it yourself

Every signal we've ever published is on the public alert history. catch% and peak% are byte-identical across the in-app feed, email digests, social posts, and the landing page — the same (ticker, day, session) row drives all of them. If you find a mismatch, email adrian@bullalert.ai and we'll fix it.

Not financial advice

BullAlert is an educational data tool. We are not a registered investment advisor, not a broker, and not a financial planner. Nothing on this page or in our signals is investment advice. Always do your own due diligence. Past performance does not guarantee future results.