# Building a Trader's Routine
The market doesn't care if you're tired, distracted, or not feeling it. But you should. A consistent routine is the difference between traders who last 5 years and those who last 5 months.
Why Routine Matters for Systematic Trading
If you trade a fully automated system, you might think routine doesn't apply. The model does the work. You just check it.
Wrong.
The routine is for you, not the model. When things go wrong — and they will — your routine determines whether you respond correctly or emotionally. A routine removes the decision-making from your response. You do what the routine says, not what your gut says.
The Pre-Market Routine (6:30-7:30 AM MT)
6:30 — Market Check
Don't look at charts yet. Check:
- Overnight session: Did ES move significantly in the electronic hours (6 PM – 6 AM)?
- Global markets: How did Asia and Europe trade?
- Key levels: Where did ES open? Any gaps?
This takes 5 minutes. The purpose is context, not decisions.
6:45 — Economic Calendar
Check today's releases:
- Any Tier 1 data (NFP, CPI, FOMC)? → Alert: prepare for volatility
- Any Tier 2 data (ISM, Retail Sales)? → Note: may cause movement
- Anything else? → Ignore
Make a note of release times. Set a phone alert 30 minutes before.
7:00 — System Check
- Is the model server running? (Ping or dashboard check)
- Is the data feed active? (Last data timestamp < 5 seconds ago)
- Are there any alerts from the watchdog?
- Are all systems communicating? (VPS ↔ NinjaTrader ↔ API)
If any system is down, fix it before the market opens. Don't trade with a broken infrastructure.
7:15 — Model Status
- What did the model output overnight? (Overnight signals from the E-mini session)
- Are there pending signals for the open?
- What is the model's current confidence mode?
No changes to the model at this point. Just read the output.
7:25 — Risk Check
- Current P&L for the day/week/month
- Current drawdown from peak
- Position limits: how many contracts are we allowed to trade today?
- Are we in any drawdown reduction phase?
If we're in a drawdown phase (Tier 2+), confirm reduced position sizes. Don't decide at the open — confirm.
7:30 — Market Open Prep
- RTH opens at 7:30 AM MT (9:30 AM ET)
- First 15 minutes: observe, don't trade
- Set price alerts at the opening range high and low
- Wait for the opening range to print
During Market Hours (7:30 AM – 2:15 PM MT)
The System Runs
For the most part, the model trades itself. Your job is:
- Monitor systems: Check the dashboard every 30 minutes (30-second glance)
- Respond to alerts: System failure, position mismatch, risk limit hit
- Don't interfere: No manual trades. No override signals. No "just this one because it looks good."
Alert Response Protocol
If the watchdog alerts:
- Read the alert message
- Assess severity: can the next bar wait, or is this urgent?
- If urgent: flat all positions immediately
- If non-urgent: finish the current bar cycle, then address
If the model produces a suspicious signal:
- Check the signal file. Is the format correct?
- Check the model logs. Was the inference normal?
- Check recent features. Are they in normal ranges?
- If anything is abnormal: don't trade the signal. Investigate.
If the broker disconnects:
- NinjaTrader will attempt auto-reconnect
- After 5 seconds: log the event
- After 30 seconds: flat any positions (via backup exit strategy)
- Don't reconnect manually until you know the cause
The Midday Review (10:30 AM MT / 12:30 PM ET)
After the first hour of confirmed action:
- How did the opening behave? (Gap hold, fill, or flat?)
- Is the model trading in line with expectations?
- Are there any alerts or flags?
- Update the session log: 3-4 sentences on what happened
The lunch lull is coming (12:00-1:30 ET). The model reduces position sizes during this period. Don't override.
The Close (2:15 PM MT / 4:15 PM ET)
Position Reconciliation
- What positions does the model think it has?
- What positions does the broker show?
- Are they the same? If not, identify the mismatch.
Daily Log
Write the session log:
- Trades taken (with timestamps)
- Trades skipped (and why)
- Gating failures (with reasons)
- System status
- Anything unusual
P&L Update
- Did we make or lose money today?
- Running P&L for the month
- Current drawdown from peak
- Update the performance spreadsheet
The Evening Review (5:00-6:00 PM MT)
Performance Review
Not every trade. Not every signal. The big picture:
- Was today's model behavior consistent with expectations?
- Did we hit any risk limits?
- Is the 5-day moving average P&L positive or negative?
- Is any new pattern emerging?
System Maintenance
- Check for updates (NinjaTrader, OS, Python packages)
- Verify data pipeline is ingesting correctly
- Check disk space on the VPS
- Review watchdog logs for anomalies
Plan for Tomorrow
- Note any economic releases
- Set alerts if needed
- Confirm the model is scheduled to run at market open
- Done for the day
The Weekly Review (Saturday)
Performance Deep Dive
- Weekly P&L and monthly running total
- Trade count, win rate, average R:R
- Gating failure analysis: are we blocking too many or too few trades?
- Feature importance check: have any features lost predictive power?
Model Health
- Feature drift check: are today's feature distributions different from training data?
- Regime check: has the market regime changed?
- Walk-forward review: is the model's performance in line with OOS expectations?
Strategy Adjustments
- Do NOT change the model based on one week
- DO note hypotheses for later testing
- If there's a clear data issue or execution issue, fix it
- Otherwise: let the model run
The Monthly Review
Full Performance Report
- Monthly P&L, win rate, Sharpe, max drawdown
- Comparison to prior months
- Comparison to model's backtest expectations
- Did the model pass the monthly acceptance criteria?
Model Retraining Decision
- Is it time for a scheduled retrain?
- Has the regime changed enough to require retraining?
- Are there enough new data points to make retraining worthwhile?
Capital Allocation Review
- Should we increase position sizes? (3+ months of consistent performance)
- Should we decrease? (drawdown thresholds hit)
- Is our account size still appropriate for the model's risk profile?
Goal Setting
- What do we want to accomplish next month?
- Any new features to test?
- Any infrastructure improvements?
- What's the one thing that, if improved, would have the biggest impact on performance?
The Golden Rule
Your routine is your insurance policy. When you're emotional, you don't follow the routine. When the routine breaks, you make bad decisions.
Write the routine down. Follow it. Review it when things go wrong.
If you're losing money, the first thing to check is: did we follow the routine? If not, fix the discipline first, then check the strategy. Discipline issues masquerade as strategy issues all the time.
If you followed the routine and are still losing money: the strategy needs work. But at least the routine didn't make it worse.