When the compliance engine vetoes a signal, that signal does not disappear. It goes into a tracking queue. From that moment forward, the system records what the price does — as if the trade had been placed. At the point where the trade would have closed (either the hypothetical target hit, the hypothetical stop triggered, or N days passed), the system logs the outcome: a win, a loss, or an expired signal with no clear resolution. This is counterfactual tracking — a complete record of what the AI wanted to do that it could not, and what would have happened if it had.
This data serves two purposes simultaneously. For you as a user, it provides transparency into the AI's analytical view even on signals that never traded — you can see the opportunities that were structurally blocked, and you can evaluate whether the compliance engine's caution was warranted. For the AI itself, the counterfactual outcomes feed directly into the learning system — the components that adjust the system's future behavior. If a pattern of profitable signals is being systematically blocked by a constraint (like PDT limits), the system detects this and can recalibrate its approach.
What counterfactual tracking is
Every vetoed signal becomes a counterfactual experiment. The system records the exact state at the moment of veto: the entry price that had been calculated, the stop-loss and target levels, the signal direction, and the veto reason. It then monitors the ticker's price for the duration the trade would have been active — typically until either the hypothetical stop or target is breached, or until a maximum hold time (based on the strategy type) expires.
This is distinct from simply watching a stock after a veto. The system applies the actual entry/stop/target levels to determine the hypothetical outcome. A signal vetoed at 10:15 AM with an entry of $42.50, stop at $41.00, and target at $44.75 — if the ticker reaches $44.75 before $41.00, the counterfactual records a win. If $41.00 is hit first, it records a loss. The tracking is mechanical and consistent, applying the same exit logic the execution engine would have used.
How to find vetoed signals
On the Markets page, open the filter panel and select Status: Vetoed. This narrows the feed to all signals that the compliance engine rejected in the current session (or the lookback period set in the date filter). Each vetoed row shows the veto reason in place of the Status pill. Expanding a vetoed row shows the same Levels panel, Score Breakdown, and trade explanation as an executed signal — plus the compliance panel showing which specific check failed and why.
For older vetoed signals where the system has had time to track the hypothetical outcome, the expanded row shows an Outcome column in the Levels panel. A green entry there means the signal would have been profitable; red means it would have hit the stop. Gray means it expired without resolution (neither stop nor target reached within the hold window). This gives you hard data on whether the veto was beneficial or costly in hindsight.
Interpreting the hypothetical outcome
This section covers key concepts related to Interpreting the hypothetical outcome. The system is designed to be fully transparent — every decision has a documented reason and every metric has a precise definition.
As you use Apex1819 over time, you will develop an intuition for how the AI behaves in different market conditions. The Learn hub is designed to give you the conceptual vocabulary to understand what the AI is doing and why.
How the AI learns from vetoed signals
The learning system uses counterfactual outcomes as negative/positive training signals for the analytical weights. A vetoed signal that would have been profitable is logged as a missed opportunity — evidence that the compliance engine's threshold might have been too restrictive for that regime and signal type. A vetoed signal that would have hit its stop is logged as a correct veto — evidence that the threshold was right.
Over time — typically measured in weeks, not days — the accumulation of these signals causes gradual shifts in the analytical weights. If the system is consistently missing profitable setups in a specific regime because the quality threshold is too high, the threshold may be adjusted down slightly. If vetoed signals in another regime are consistently unprofitable, the threshold in that regime stays high or increases. This is the self-improving loop at work: every veto generates information, and that information shapes the AI's future behavior.
When the veto was right vs. wrong
This section covers key concepts related to When the veto was right vs. wrong. The system is designed to be fully transparent — every decision has a documented reason and every metric has a precise definition.
As you use Apex1819 over time, you will develop an intuition for how the AI behaves in different market conditions. The Learn hub is designed to give you the conceptual vocabulary to understand what the AI is doing and why.