Budget exceeded

run_wall_clock

warns at 2 s, stops at 30 s per calculation

Too much history, or heavy work on every bar.

Declare less history. Move heavy work into a higher-timeframe handler that runs once per bucket.

handler_wall_clock

5 s in one onBar or handler call

A loop over every price tick on every bar.

Step by the instrument's tick size or a row-size setting, not 0.01.

loop_iterations

350,000 per call; 2 million to 50 million per run, scaled by bar count

Nested loops, or a fine price step on a high-priced symbol such as BTC.

Process each higher-timeframe bucket once. The lint warns early with workload/fine-price-step.

ctx_calls

500,000 per call

Painting or accumulating inside a loop.

Compute in the loop, then call ctx.plot or ctx.accum once.

indicator_points

warns at 500,000, stops at 2,000,000

Many outputs on a long history.

Plot on transitions only, or drop outputs.

emitted_events

warns at 25,000, stops at 100,000

A condition that is true on most bars.

Tighten the condition, or declare oncePer on the event.

accum_slots

warns at 256, stops at 1,024

A new ctx.accum id per bar or per level.

Reuse a fixed set of ids and keep collections inside one slot.

lifecycle_instances

warns at 2,500, stops at 10,000

Every bar adds a range or line that never ends.

Declare expires, or use mode: "replace" for rolling zones.

layer_rows

warns at 5,000, stops at 20,000 rows per layer

One row per bar.

Key one row per session or per zone, and seal and remove old rows.

layer_bytes

warns at 4 MB, stops at 32 MB per layer

Fine bins or grid steps.

Coarsen the step or shrink float32array fields.

warmup_replay

warns at 2.5 s, stops at 10 s

A large warmup with arrays built in the handler.

Lower the warmup, or keep running sums instead of arrays.

script_execution_timeout

15 s for one event; also onFinish and startup

Do less in onEvent: shorter ctx.candles.forward windows, fewer dependency reads, return early for events you skip.

too_many_event_timeouts

10 event timeouts in one run

Same as above. One slow branch usually causes all of them.

budget_wall_clock

10 minutes for the whole run

Use a shorter date range or fewer declared definitions and indicators.

memory_limit

the worker's memory

Use a shorter range and collect smaller values.

budget_total_events

100,000 events

Use a shorter range or a more selective definition.

budget_emitted_events

100,000 events from one definition

Tighten the definition's condition.

budget_collected_rows

100,000 rows per key, 500,000 in total

Collect aggregates in onEvent, not raw series.

budget_forward_bars

5,000 bars in one forward scan

Shorten the forward window.

budget_payload

the published results' size

Publish fewer or smaller results; never publish a full series.

budget_annotations

10,000 chart annotations

Mark fewer events.

budget_call_stack

1,024 nested calls

Replace recursion with a loop.