Monitoring Trade Bot Performance

Once your bots are running, the work shifts from building to watching. Monitoring is not staring at charts all day — it is setting up the right alerts, checking the right numbers, and intervening only when the data justifies it.

What to Monitor

The bot dashboard tracks a large number of metrics. Focus on the ones that tell you whether the strategy is working as intended.

Core Health Metrics

Metric	What to watch for
Realized profit	Your actual net return after fees. If this is negative over a meaningful period, the strategy is not working.
Return on investment (ROI)	Profit relative to capital deployed. Compare across bots to see which strategies are worth allocating more capital to.
Win percentage	Should be consistent with your backtest baseline. A declining win rate over many trades suggests the market has shifted or the strategy is degrading.
Maximum drawdown	The largest peak-to-trough decline. If this exceeds your expected max drawdown from backtesting, stop and investigate.
Fee costs	Fees add up. If fee costs are high relative to gross profit, the strategy may be overtrading — too many entries and exits eroding the edge.

Operational Metrics

Metric	What to watch for
Failed orders / Cancelled orders	Occasional failures are normal. A spike in failed orders suggests an exchange issue, rate limiting, or insufficient margin.
Average open time	If orders are sitting unfilled for unusually long periods, your limit prices may be too far from the market.
Partially filled orders	Frequent partial fills on a market you expected to be liquid suggests the order size is too large for the order book depth.

Using Dashboards

Create a custom dashboard for monitoring. Add widgets for bot status, portfolio balance, and account overview. A well-set-up dashboard lets you check the health of all your bots in seconds without navigating into each one.

Keep a dashboard open during active trading sessions if you trade intraday timeframes. For longer timeframes (4-hour, daily), checking once or twice a day is usually sufficient.

Notifications as Early Warning

Configure notifications for the events you actually need to act on:

• Bot closed position — lets you know when a trade completes. Useful for strategies that take few positions so each one matters.
• Order failed — alerts you to execution problems that need attention.
• Security — always enable. Unauthorized access or API key changes should trigger immediate action.

Avoid enabling notifications for every order placed — on a high-frequency bot, this floods you with noise and buries the signal. Enable notifications for the exceptions, not the routine.

Logs

The Logs tab on each bot's dashboard records every action the bot takes: signals evaluated, orders placed and filled, errors encountered. Logs are your first stop when something looks wrong.

If a bot's performance drops unexpectedly:

1. Open the Logs tab
2. Scroll to the period where the drop started
3. Look for error messages, rejected orders, or unexpected behavior (e.g., the bot stopped placing orders entirely)
4. Check whether the bot is still evaluating signals correctly or if a data feed issue is preventing it from running

Logs also serve as an audit trail. If you ever need to reconcile a discrepancy between what you expected and what happened, the logs tell the story.

When to Intervene

Not every red number requires action. Distinguish between noise and a problem.

Intervene when:

• The bot has stopped trading entirely (no orders, no signals)
• Order failures are consistent and preventing the bot from operating
• Drawdown exceeds your risk tolerance
• Performance has degraded over a statistically meaningful sample of trades
• An exchange announces downtime or a delisting that affects your market

Do not intervene when:

• A single trade loses money (that is expected)
• A single day is red (markets have noise)
• The bot is trading less frequently than you expected — check the logs to understand why before assuming something is broken
• You are bored and want to "do something"

Over-intervention is a common mistake. If you manually override a bot every time you feel uneasy, you are no longer running an automated strategy — you are manually trading with extra steps. Trust the strategy you validated, and let it run its course.

Comparing Live vs Simulated Performance

If you run the same bot on both a simulated and a live account, compare their results periodically. The gap between sim and live ROI tells you how much slippage, partial fills, and real-world friction are costing you.

A small gap is normal. A large gap — sim showing 20% ROI while live shows 2% — means something is wrong. The most common causes: the market is less liquid than the sim assumed, your order size is too large for the order book, or exchange fees are higher than expected on that pair.

Use this comparison to size your positions realistically. If live underperforms sim by 5% ROI on average, factor that into your expectations.

Regular Review Cadence

Set a rhythm:

• Daily (2 minutes): Check dashboard for failed orders and unusual PnL swings
• Weekly (15 minutes): Review performance metrics, compare against backtest baselines, check logs for patterns
• Monthly (30 minutes): Full review. Compare live vs sim performance, assess whether the strategy still fits current market conditions, decide whether to adjust or replace

This cadence keeps you informed without making monitoring a full-time job. The bot does the trading — your job is to make sure the bot is doing what you built it to do.