Three layers, and only one of them is a prediction

The DAYTRADE tab on the terminal is built from three separate layers. Each one is measured on its own, and the numbers below come straight from the same engine that serves the panel — recomputed from stored market data every time new bars land, never hand-edited.

Layer 1 — direction. At every market close, three frozen pattern rules run against the daily bar history:

thrust2_up — two consecutive days each up 1% or more
big_up — a single day up 2% or more
run3_up — a three-day run totalling +2.5% or more

If any rule fires, the next session is a LONG signal day. If none fires — and that is most days — the system's answer is no trade. Across the full history (888 sessions of SPY daily bars), the rules fire on 52 unique days, about 6% of all sessions. Those 52 signal days closed higher the next day 78.8% of the time, averaging +0.31% per trade (+17.4% compounded) — against a baseline of 56.7% and +0.08% for an average day. A day that fires two rules at once counts once; cumulative returns are compounded, not summed.

Layer 2 — timing. Minute-bar replay of trend days shows a repeating structure: the low of the day prints early — between 09:30 and 11:15 ET — and the high prints late, after 15:00. The playbook that follows from it: on a signal day, buy weakness in the first ninety minutes, exit in the final hour. The panel shows each recent session's actual optimal window so you can judge the pattern yourself.

Layer 3 — the option itself. The first two layers are measured. This one is a translation, and it deserves honesty: converting an underlying move into an option's profit depends on strike, expiry, implied volatility, and gamma. The useful yardstick is the implied daily move — at recent implied vols (~13–14%), SPY's options price in roughly a ±0.8–0.9% day. A signal day that beats that (2026-08-04 ran +1.79%, about twice the implied move) pays near-the-money calls roughly +50% to several hundred percent on premium depending on strike and entry. A signal day that stalls inside the implied move loses most or all of the premium to time decay and spread.

What a loss looks like — read this part twice

The honest projection for a defined-risk call bought on a signal day (the hit rates here are the two-thrust cohort's next-day base rate; the panel's scoreboard labels each basis — the conditional close-to-close statistic and the executable next-open-to-close expression — separately):

Win case (historically ~2 of 3 signal days): the day trends, the option returns +50% to +300% on premium for near-the-money short-dated strikes.
Loss case (~1 of 3 signal days): the pattern misses, and the option loses 40–100% of the premium paid. Options can and regularly do go to zero.
Expected value lives across MANY signal days, not any single one. A 67% hit rate means one in three of these trades loses. Position sizing — risking only what a full loss cannot hurt — is the entire game.

Two more things the system will tell you that most services won't:

"Do nothing" is the most common output. In the three weeks before 2026-08-04 the rules fired zero times — fifteen straight no-trade days. That was correct: the tape was chop, and short-dated options bleed to costs in chop. Not trading those days is the edge.
It doesn't call every move. The −1.4% breakdown on 2026-07-29 came with no signal — the ruleset is long-only today, so it kept you flat (no call losses) but did not capture puts. Short-side rules exist in the research layer and will only be promoted if their measured base rates earn it.

Where to see it live

The DAYTRADE tab on the terminal carries the direction engine, the replay panel (the scoreboard above plus each recent session's day return, optimal window, and which rule fired the night before), gamma regime, and unusual-flow reads. The same replay data serves on the Options tab for every visitor, with deeper history for members.

Everything here is derived from delayed and end-of-day market data and is published for information only — it is not investment advice, and it is the same impersonal output for every reader. Past pattern frequencies do not guarantee future results. Options involve substantial risk of loss.