A verdict is in one of three states — setup, no-setup or suspended — and carries one of five signals: strong-buy, buy, no-trade, sell or strong-sell. Signals are derived deterministically from the engine’s own numbers; the language model never writes one. Every no-trade names its single deciding reason.
The three states
Suspended means the newest bar is too old to stand behind — the read is not wrong, it is simply no longer current. The product distinguishes a closed exchange from a delayed feed here, because "the market is shut" and "our data is late" are different facts and only one of them is about us.
No-setup means the read completed and refused. There are four deciding reasons: a rule pack's hard rule failed, the chart held no setup — no pattern qualified, or the only thing on it was an extended chase — reward-to-risk fell below 1, or the grade landed at or below the veto floor.
Setup means a plan stands.
The five signals
Signals are computed from the engine's numbers by a deterministic function. The narration layer never sees the signal and never writes one, and directive buy-or-sell language is banned from generated prose.
This separation is the point. A model that can write "strong buy" can write it for reasons that are not in the data, and no reader can tell the difference from the outside.
The strong tier
A signal is promoted to its strong form only when four conditions hold together: the grade is at least A−, calibrated confidence is at least 70, the multi-timeframe score is at least 0.5, and reward-to-risk is at least 2. Failing any one of them leaves an ordinary buy or sell, and the verdict names which condition was missed.
Tested setups: Beta and Proven
Since engine 1.12.0 a plan — entry, stop and targets — is given only for a setup that has beaten random entries in a registered test. The test compares how often the setup reached its first target before its stop with the same trade, same distances, taken at random moments in the same market and month. A read with no such setup is a chart read: it describes the trend, structure, location and stretch of the chart, and shows no plan, grade or confidence.
A tested setup carries a tier. Beta means it passed the test on past data and is being tested again on bars that did not exist when it was defined. Proven means it has passed that second test too. At launch every tier is Beta: the failed breakout on crypto and forex 1H and 4H, and the top-ranked reads in crypto, forex and stocks. The second tests read no earlier than December 2026; a setup that fails its test loses its plan.
Every tested setup ORIN issues is tracked on its report card, against thirty random moments per result. A report card shows a rate only once it holds 100 results over 30 market days. Gold has no tested setup yet, so gold charts get a chart read.
A tier is necessary for a plan, not sufficient: stale data, a pack hard rule, reward below the floor, a refused grade and the higher-timeframe veto still refuse a tiered read.
The higher-timeframe veto
When the charts above the graded one disagree strongly enough, the trade is refused outright rather than merely marked down.
The veto only fires when at least one higher leg was genuinely fetched, which today means crypto. For markets where higher legs are aggregated roll-ups rather than separate fetches, disagreement can discount the multi-timeframe score but can never veto — an approximation is not allowed to cast the deciding vote.
Stored versus live
The stored signal is the historical record of what was said at the time. Every surface re-derives the current one, which is why an aged read shows no-trade without its record being altered.
Frequently asked
- Why does a chart show no plan when it looks like a setup?
- Because ORIN only gives a plan where a setup has beaten random entries in a registered test. On the replayed record most untested setups played out no more often than chance, so a plan on them would be a guess in a confident format. The chart read still tells you what the chart shows.
- Is a no-trade verdict an error state?
- No — it is a complete answer. The read names its single deciding reason, draws no levels, and is recorded like any other. Passing on it is a decision the journal can later credit, so a refusal still earns its place in your record.
- Does the higher-timeframe veto apply to stock charts?
- No. The veto exists only off a genuinely fetched higher leg, which today means crypto. Equity and forex legs are aggregated roll-ups: they can discount the multi-timeframe score but never veto, because an approximation should not cast the deciding vote.
Updated Sep 1, 2026