Automation is only trustworthy if it is accountable. The execution audit trail is the bot's complete, chronological record of what it did and why — every signal it evaluated, every order it placed, and every signal it refused. Learning to read it turns "the bot traded" into "the bot traded for these specific, checkable reasons."
What the audit trail records
The trail is append-only and time-ordered. For each moment of decision it captures the signal that was considered, the cost check that was run against it, the resulting action — an order or a rejection — and the outcome once the exchange responds. Nothing the bot does happens off the record.
Step by step
- Open the audit trail. Find the chronological log of every action, from signal evaluation through to order outcome. Read it top to bottom; sequence is part of the story.
- Trace a single signal. Follow one signal from generation, through its cost check, to either a placed order or a logged rejection. This end-to-end view is how you confirm the rules fired correctly.
- Read the rejections. Rejections record why a signal failed the cost bar. These are working-as-intended entries, not faults — a quiet day of rejections is the discipline doing its job.
- Reconcile orders to fills. Match each placed order to its fill and to any risk event, so every order has a beginning and an end you can account for.
Using the trail as a tool, not a receipt
The audit trail is most valuable read forward, not just looked up after something goes wrong. Reviewing it regularly tells you whether the bot is rejecting weak signals as expected, whether fills are landing near the prices the cost model assumed, and whether risk limits are intervening when they should. It is also the first place to look after any unexpected event.
The same log is where you start a recovery: see how to recover the bot after a kill switch. To understand the rejections you will read here, the signals pillar explains the cost-beating rule, and paper mode vs live mode covers how the same trail behaves before and after you go live.