A trade should be rejected when its planned loss would push total weekly risk above the limit set before the week began. A valid signal does not override the risk budget.
At 10:14 PM on Sunday, the order looks reasonable in isolation. The entry, stop, and position size all fit the setup. Then one line in the review changes the decision: if the stop is reached, the week’s total planned loss will exceed its limit.
Monday has not opened. Nothing has been sent to the broker. The cleanest decision available is still “reject.”
The limit has to govern the order
On August 1, 2012, Knight Capital deployed software connected to U.S. equity markets. Old code on one server remained active. When trading began, the system sent millions of unintended orders into the market.
Knight lost more than $460 million in about 45 minutes.
The U.S. Securities and Exchange Commission documented the incident in its 2013 administrative order. Knight had controls, procedures, and experienced people. What failed was the system around deployment and risk containment. The software could act at market speed before the firm stopped it.
A retail trader operates at a smaller scale, but the control problem has the same shape. Once an order executes, the question changes from “Should I take this risk?” to “How do I manage what is already live?”
That distinction matters on Sunday night. The weekly loss limit only protects the account if it can block an otherwise attractive order before execution. A limit that becomes negotiable whenever a setup looks promising is a note, not a control.
Planned loss belongs to the whole portfolio
Suppose the weekly risk limit is $100. Closed losses and open positions already account for $85 of that budget. A new order carries $25 of planned loss from entry to stop.
The new trade may satisfy every setup rule. The arithmetic still fails:
$85 already allocated + $25 proposed = $110 total planned loss.
The excess is $10. The trade does not become acceptable because the chart is clean, the signal arrived late, or Monday’s open feels important. Those details may affect the trade thesis. They do not create another $10 of risk capacity.
This is where traders often separate decisions that should remain connected. They examine the new order alone, then compare its potential reward with its $25 risk. The missing calculation is portfolio-level exposure: what happens if current positions and the new position all reach their stops?
Correlated positions make the review stricter. Three different tickers can still express one underlying bet. A technology stock, a semiconductor stock, and a broad growth fund may react to the same market move. Adding their stop distances does not capture every form of concentration, but it gives the approval decision a concrete starting point.
For a closer look at how losses reduce available room for later trades, see What Happens to Your Trading Capacity During a Drawdown?.
Rejection preserves Monday’s choices
Rejecting the order does not predict that it will lose. The trade could rally immediately after the open. Risk management cannot remove that discomfort.
The rejection protects a different outcome: the trader reaches Monday with the agreed limit intact.
That creates several possible next steps. The trader can wait for an entry that permits a smaller position, close or reduce an existing exposure for a reason already defined in the plan, or let the setup pass. Each choice respects the same constraint. Raising the weekly limit after seeing the signal would rewrite the plan at the exact moment it became inconvenient.
An approval gate makes this conflict visible. The AI can generate and queue a trade signal, including the proposed entry, stop, and size. A human still has to compare the order with current exposure and approve or reject it before anything executes. The gate creates time for judgment where an autonomous bot would create a fill.
The rejection should also enter the trading journal. Record the proposed loss, risk already used, projected weekly total, and rejection reason. “Exceeded weekly risk limit by $10” is more useful than “felt risky.” Over time, those records reveal whether signals tend to arrive after capacity is exhausted, whether position sizes are consistently too large, or whether the weekly limit conflicts with the strategy’s normal trade frequency.
Write the rejection rule before Sunday night
Knight Capital’s loss unfolded faster than a person could review each order. That is why controls must operate before exposure expands.
For an approval-gated trading process, define the weekly rule in arithmetic:
- Count realized losses for the current week.
- Add planned loss on open positions, based on their current stops.
- Add the proposed order’s planned loss.
- Reject the order if the total exceeds the preset weekly limit.
Document how gains affect the calculation, if at all. Decide whether correlated positions receive an additional cap. Define what happens when a stop moves. Make those choices before a live signal tests them.
At 10:14 PM, the trader does not need confidence about Monday. The trader needs one calculation and the discipline to accept its result.
Educational content, not financial advice.
Comments
No comments yet.