@@ -1402,6 +1402,7 @@ def fetch_sensor_data_cars(self):
 
                 completed = []
                 planned = []
+                started = []
 
                 if entity_id and "octopus_intelligent_slot_action_config" in self.args:
                     config_entry = self.get_arg("octopus_intelligent_slot_action_config", None, indirect=False)
@@ -1415,10 +1416,22 @@ def fetch_sensor_data_cars(self):
                     try:
                         completed = self.get_state_wrapper(entity_id=entity_id, attribute="completed_dispatches") or self.get_state_wrapper(entity_id=entity_id, attribute="completedDispatches")
                         planned = self.get_state_wrapper(entity_id=entity_id, attribute="planned_dispatches") or self.get_state_wrapper(entity_id=entity_id, attribute="plannedDispatches")
+                        # Not merged into octopus_slots or used for any rate/plan decision yet - read
+                        # only for the #4516 Stage 1 diagnostic timeline log below, to observe whether
+                        # this is a trustworthy earlier-than-completed confirmation signal before
+                        # building any gating logic on it (Stage 2, deferred).
+                        started = self.get_state_wrapper(entity_id=entity_id, attribute="started_dispatches") or self.get_state_wrapper(entity_id=entity_id, attribute="startedDispatches")
                     except (ValueError, TypeError):
                         self.log("Warn: Unable to get data from {} for car {} - octopus_intelligent_slot may not be set correctly in apps.yaml".format(entity_id, car_n))
                         self.record_status(message="Error: octopus_intelligent_slot not set correctly in apps.yaml for car {}".format(car_n), had_errors=True)
 
+                # #4516 Stage 1: diagnostic dispatch-timeline log. Purely observational - see
+                # build_dispatch_timeline()'s and dispatch_timeline_should_log()'s docstrings.
+                timeline = self.build_dispatch_timeline(car_n, completed, started, planned)
+                should_log, marker = self.dispatch_timeline_should_log(car_n, timeline)
+                if should_log:
+                    self.log("Octopus: Dispatch timeline car {} @ {} [-4h..+24h]: {}{}".format(car_n, self.time_abs_str(self.minutes_now), timeline, marker))
+
                 # Completed and planned slots - merge from all cars
                 if completed:
                     self.octopus_slots[car_n] += completed
 
@@ -2910,9 +2910,14 @@ def load_saving_slot(self, octopus_saving_slots, rate_dict, export=False, rate_r
                         if not export:
                             self.load_scaling_dynamic[minute] = self.load_scaling_saving
 
-    def decode_octopus_slot(self, car_n, slot, raw=False):
+    def decode_octopus_slot(self, car_n, slot, raw=False, boundaries_only=False):
         """
         Decode IOG slot
+
+        boundaries_only skips the kWh/source/location handling below - including the "remove
+        empty slots" check, which real HA-integration started_dispatches entries (start/end only,
+        no kWh, source or location) would otherwise trip, decoding them as empty (#4516 Stage 1's
+        build_dispatch_timeline() review). Only start/end are meaningful there.
         """
         if "start" in slot:
             start = datetime.strptime(slot["start"], TIME_FORMAT)
@@ -2936,6 +2941,9 @@ def decode_octopus_slot(self, car_n, slot, raw=False):
         if start_minutes == end_minutes:
             return 0, 0, 0, source, location
 
+        if boundaries_only:
+            return start_minutes, end_minutes, 0, source, location
+
         cap_minutes = end_minutes - start_minutes
 
         # The load expected is stored in chargeKwh for the period in use
@@ -3021,6 +3029,98 @@ def get_octopus_slot_max(self):
             return OCTOPUS_SLOT_MAX_CAPPED
         return OCTOPUS_SLOT_MAX_DEFAULT
 
+    def build_dispatch_timeline(self, car_n, completed, started, planned, window_before_hours=4, window_after_hours=24, step=30):
+        """
+        Build a fixed-width, one-character-per-block dispatch status string for the diagnostic
+        timeline log (#4516 Stage 1 - not yet used for any rate/plan decision, purely observational).
+
+        Each character covers `step` minutes, offset from now, spanning
+        [-window_before_hours, +window_after_hours). '.' = nothing known, 'P' = planned
+        (provisional), 'S' = started, 'C' = completed. Where lists disagree on the same block, the
+        most-confirmed status wins (completed > started > planned) - reflects Octopus's own view
+        having moved on, not a genuine simultaneous claim.
+
+        The letter is lowercased where Predbat's own plan charges in that block, so 'p' is a
+        provisional slot Predbat is relying on and 'P' one it is not. That distinction is the
+        point: a slot that disappears having never been planned against costs nothing, while one
+        Predbat committed an import to is a plan that will not happen.
+
+        Stacking consecutive lines (a heartbeat one per 30-minute boundary, plus one on any
+        earlier change - see the caller in fetch.py) in a monospace log viewer reveals dispatch
+        lifecycle as diagonal stripes: a specific dispatch drifts one column per line as `now`
+        advances, so a rescinded slot shows as a stripe that stops before reaching the `now`
+        column, while a genuinely-delivered one runs through into 'C'.
+        """
+        total_before = window_before_hours * 60
+        total_after = window_after_hours * 60
+        num_blocks = (total_before + total_after) // step
+        status = ["."] * num_blocks
+
+        def mark(slots, char):
+            for slot in slots or []:
+                start_minutes, end_minutes, _, _, _ = self.decode_octopus_slot(car_n, slot, raw=True, boundaries_only=True)
+                if start_minutes == end_minutes:
+                    continue
+                start_offset = start_minutes - self.minutes_now
+                end_offset = end_minutes - self.minutes_now
+                block_start = max(0, (start_offset + total_before) // step)
+                block_end = min(num_blocks, -(-(end_offset + total_before) // step))  # ceil division
+                for block in range(int(block_start), int(block_end)):
+                    status[block] = char
+
+        mark(planned, "P")
+        mark(started, "S")
+        mark(completed, "C")
+
+        # Lowercase where Predbat's own plan plans to import - so a stripe shows not just the
+        # dispatch's lifecycle but whether Predbat had committed a charge window to it. A slot
+        # that reads 'p' and then vanishes before reaching the now column is one Predbat planned
+        # around and lost, which is the case a plain P/S/C timeline cannot distinguish from a
+        # slot nothing depended on.
+        #
+        # This runs in fetch, before the planner, so the windows are the *previous* cycle's plan -
+        # one 5-minute cycle stale against 30-minute blocks. That is deliberate: what Predbat
+        # believed while the slot was still live is the more useful thing to record.
+        for window_n, window in enumerate(self.charge_window_best or []):
+            if window_n < len(self.charge_limit_best or []) and self.charge_limit_best[window_n] <= 0:
+                continue
+            start_offset = window.get("start", 0) - self.minutes_now
+            end_offset = window.get("end", 0) - self.minutes_now
+            block_start = max(0, (start_offset + total_before) // step)
+            block_end = min(num_blocks, -(-(end_offset + total_before) // step))
+            for block in range(int(block_start), int(block_end)):
+                if status[block] in ("P", "S", "C"):
+                    status[block] = status[block].lower()
+
+        return "".join(status)
+
+    def dispatch_timeline_should_log(self, car_n, timeline):
+        """
+        Decide whether fetch.py should log a dispatch-timeline diagnostic line this cycle, and
+        what marker to append if so.
+
+        A 30-minute heartbeat always logs - dispatch decisions are 30-min-granular anyway, so log
+        volume stays sane at that cadence - but a change from the last logged timeline for this
+        car also logs immediately outside that boundary, marked with a trailing ' *' (#4948
+        review). Without this, a provisional slot that appears and disappears entirely within one
+        half-hour window would never appear in this log at all, however briefly Predbat's own
+        plan relied on it - exactly the pattern this diagnostic exists to catch.
+
+        The marker is a suffix, appended by the caller after build_dispatch_timeline()'s
+        fixed-width string, rather than a prefix - so heartbeat and change-triggered lines still
+        stack column-for-column in a monospace viewer. That alignment is the whole point of the
+        render: consecutive lines are meant to line up into diagonal stripes.
+
+        Records the new timeline as "last logged" as a side effect whenever this returns True, so
+        a caller must log with the returned marker exactly when told to, and no other times.
+        """
+        is_heartbeat = self.minutes_now % 30 == 0
+        changed = timeline != self.dispatch_timeline_last.get(car_n)
+        if not (is_heartbeat or changed):
+            return False, ""
+        self.dispatch_timeline_last[car_n] = timeline
+        return True, "" if is_heartbeat else " *"
+
     def load_octopus_slots(self, car_n, octopus_slots, octopus_intelligent_consider_full):
         """
         Turn octopus slots into charging plan
 
@@ -497,6 +497,7 @@ def reset(self):
         self.octopus_intelligent_consider_full = False
         self.notify_devices = ["notify"]
         self.octopus_url_cache = {}
+        self.dispatch_timeline_last = {}
         self.ge_url_cache = {}
         self.github_url_cache = {}
         self.load_minutes = {}