Monitoring and Alerts: How to Notice When Something's Wrong Before It Costs You

In trading, problems compound when unnoticed. A position that goes wrong silently becomes a much bigger problem than a position that goes wrong with immediate alerting. Effective monitoring catches issues early, when they're recoverable. Designing monitoring well, the right alerts, the right channels, the right thresholds, is the difference between catching problems and being caught by them.

What needs monitoring

For active trading (manual or automated):

1. Position status. Open positions, unrealized PnL, distance to liquidation (for leveraged), distance to stops. The fundamentals of "where am I right now."

2. Strategy performance. Recent trade outcomes, rolling expectancy, drawdown from recent high. The "is the strategy still working" view.

3. Account state. Balances, available margin, withdrawal status. The "do I have what I think I have" view.

4. Execution quality. Slippage, fill rates, fee accumulation. The "is my execution clean" view.

5. System health (for automated). Bot is running, API is responsive, data feeds are current, recent errors. The "is the infrastructure working" view.

6. External events. Major news, regulatory events, exchange announcements, network issues. The "is the world doing something relevant" view.

Each category has different update cadences and alert thresholds.

Alerts vs dashboards, the distinction

Dashboards. Visual displays you check periodically. You look at them; they show current state. Appropriate for: routine monitoring, regular reviews, pattern recognition over time.

Alerts. Notifications pushed to you when specific conditions are met. They interrupt your attention. Appropriate for: things that need immediate response.

The two complement each other. Dashboards for "checking in"; alerts for "things you need to know right now."

A common mistake is using alerts for everything, producing alert fatigue. Or using only dashboards, missing critical events. The right system uses both.

Designing alerts that actually help

The principles:

1. Alerts should require action. Every alert should be: "this happened; you should do X." Alerts that are "informational only" become noise. If there's no action, it's not an alert, it's a dashboard item.

2. Alerts should be timely. Alerts about events from yesterday don't help you now. Real-time alerts for time-sensitive events; delayed alerts only for non-urgent matters.

3. Alerts should be reliable. False alerts (where you check and nothing's wrong) train you to ignore alerts. False-negative alerts (missing real events) defeat the purpose. Tune thresholds to minimize both.

4. Alerts should be appropriately routed. Critical alerts go to channels that interrupt sleep (SMS, phone calls). Important alerts go to channels that interrupt work (push notifications, Slack). Informational alerts go to passive channels (email summaries).

5. Alerts should have clear signal. "Position open with -10% PnL" is clearer than "Significant adverse movement detected." The alert should tell you exactly what happened.

Critical alerts (interrupt-everything)

Events worth waking you up:

• Position approaching liquidation. "BTC long is at 90% of liquidation distance."
• Account margin critical. "Available margin < 5% of total account."
• Bot stopped or errored. "Trading bot has not responded in 5 minutes."
• Major unauthorized activity. "Withdrawal initiated from your account."
• Exchange-wide event. "Exchange API showing major outage."

These events demand immediate response. Lost time costs real money or compounds danger.

Important alerts (interrupt-work)

Events worth notifying within minutes:

• Stop or take-profit fired. "Stop hit on BTC long at $65,200."
• Position size deviates from plan. "Bot position size 50% above target."
• Daily loss limit hit. "Day's losses now exceed daily limit."
• Strategy expectancy diverging. "Last 30 trades expectancy: -0.2R (vs expected +0.15R)."
• Significant news event. "Major regulatory announcement on crypto."

These need attention but not within seconds.

Informational alerts (review periodically)

Events worth knowing about but not urgent:

• Daily PnL summary.
• Weekly strategy review trigger.
• Funding rate above/below thresholds.
• Position open beyond expected duration.
• Bot completing routine maintenance.

These accumulate in a low-priority channel for periodic review.

A common mistake: too many alerts

A trader sets up alerts for everything. After a few days, alerts fire constantly. They start ignoring all alerts. When a real critical alert fires, they miss it because they've trained themselves to ignore.

The fix: aggressive alert pruning. Each alert needs justification: what action will this trigger? If the answer is "I'll just look at it," the alert is informational and shouldn't be a real-time alert. Routine info goes to email summary; only action- required events should interrupt.

A common mistake: not testing alerts

A trader sets up alerts. They believe they're working. A real event happens; the alert fails to fire. The trader didn't know because they hadn't tested.

The fix: test alerts on setup and periodically afterward. Trigger test conditions deliberately to verify alerts fire. Most alerting systems support test modes; use them.

A common mistake: relying on a single notification channel

A trader's alerts only go via email. They don't check email on weekends. Critical event happens Saturday morning; they miss it until Monday.

The fix: critical alerts use multiple channels. Email + SMS + push notification provides redundancy. The cost (slightly more noise from duplicate notifications) is worth the reliability.

A common mistake: alerting on lagging indicators

A trader's alert fires when "drawdown exceeds 20%." By the time the drawdown is 20%, the damage is done. The alert told them what already happened, not what was about to.

The fix: combine reactive alerts (the bad thing happened) with leading alerts (something's deteriorating, before the bad thing happens). E.g., alert at 10% drawdown for awareness; alert at 15% for action; alert at 20% for emergency review.

A common mistake: not maintaining alert thresholds

Thresholds set when the account was $10,000 are silly when the account is $100,000. The alert about "$500 unrealized loss" fires constantly at the larger account size.

The fix: review thresholds periodically. As account size changes, thresholds should scale. As strategy matures, thresholds should reflect current expected variance. Stale thresholds become noise.

A common mistake: alerts for things you can't act on at 3am

A trader gets a 3am alert about a position. They check the chart. They make a decision under fatigue. The decision is worse than no decision would have been.

The fix: pre-define what alerts justify middle-of-the-night action. Most don't. For events that don't justify night action, route alerts to channels that don't wake you (email batch). Save phone-call-level alerts for genuinely critical events.

A common mistake: monitoring without dashboards

A trader has alerts but no dashboards. They have no way to "check in" on their portfolio without specific alerts firing. The big-picture view is missing.

The fix: dashboard for routine monitoring + alerts for specific events. The two complement each other. Most exchanges have basic dashboards; for advanced needs, custom dashboards (Grafana, custom spreadsheets, dedicated apps) provide the broader view.

Tools for monitoring and alerting

Several practical options:

1. Exchange built-in alerts. Most exchanges offer price alerts, position alerts, filled-order notifications. Free, basic, sufficient for many traders.

2. TradingView alerts. Powerful technical-condition alerts. Can trigger webhooks for integration with other systems.

3. Telegram / Discord bots. Custom bots that consume exchange data and post to chat channels. Free, flexible, programmable.

4. PagerDuty / OpsGenie (overkill for retail). Enterprise alerting platforms with sophisticated routing. Worth it only at significant scale.

5. Hex37's notification system. For Hex37 users, the platform's notifications (fills, liquidations, alerts) cover the basics without external setup.

The right tool depends on what you're trading and how much customization you need. Start with built-in exchange alerts; add complexity as needed.

Mental model, monitoring as the smoke detector network of your trading

A house has smoke detectors in critical rooms. They don't fire constantly; they fire when there's actual smoke. When they fire, they're loud and clear. The system catches fires early enough to deal with them.

Trading monitoring works the same way. Critical detectors in critical places. Quiet most of the time. Loud when something matters. The goal isn't to constantly monitor, it's to be reliably notified when monitoring becomes essential.

A trader who's constantly checking the platform is either over-engaged (anxiety-driven) or has bad alerting (compensating for missing notifications). Good monitoring lets you ignore the system most of the time and trust that you'll be notified when it's relevant.

Why this matters for trading

Monitoring is the difference between catching problems early and being caught by them. Hex37's notification infrastructure handles many of the basic alerts (fills, liquidations, price alerts); the discipline of configuring the alerts appropriately and maintaining them as your trading evolves is what makes the system operational.

Takeaway

Monitoring catches problems early. Use dashboards for routine view; alerts for specific events. Critical alerts (liquidation, account margin, unauthorized activity) interrupt everything; important alerts (stops fired, daily loss limit) interrupt work; informational alerts (summaries) accumulate in passive channels. Each alert should require action. Test alerts on setup. Use multiple notification channels for critical events. Combine reactive and leading alerts. Maintain thresholds as account/strategy evolves. The right monitoring lets you ignore the system most of the time while reliably catching what matters.