A trade can open underwater even when price has barely moved because the closing spread makes the execution price worse than the price used to review the signal. Approving direction without checking bid, ask, order type, and expected slippage leaves the final cost unreviewed.
On August 1, 2012, Knight Capital began routing orders after deploying new software across its systems. One server had not received the updated code. When trading started, that server activated dormant functionality and sent unintended orders into the market.
The failure lasted about 45 minutes. Knight Capital lost more than $460 million.
The U.S. Securities and Exchange Commission documented the incident in its 2013 order against Knight Capital Americas. The scale was exceptional, but the mechanism was ordinary: the instruction looked ready while one part of the execution path remained unchecked.
A Friday closing spread creates the same class of problem at a smaller scale. The signal can be coherent. The direction can be reasonable. The risk calculation can still be wrong because the executable price changed before the underlying market moved.
The signal price and the fill price answer different questions
Consider an illustrative setup in a stock quoted at $50.00 when the signal enters the approval queue. The proposed stop sits at $49.50, so the apparent risk is $0.50 per share.
Near Friday’s close, the market shows a bid of $49.94 and an ask of $50.12. A buy order filled near the ask starts roughly $0.12 above the signal’s reference price. The position appears down immediately when marked against the bid, even if the midpoint has barely changed.
That opening loss did not come from a failed directional thesis. It came from crossing the spread.
The distinction matters for position sizing. At the signal price, the distance to the stop was $0.50. At a $50.12 fill, it becomes $0.62. A position sized from the stale reference price carries 24 percent more per-share risk than the review screen suggested.
Those figures are illustrations, not a prediction of typical spreads or fills. The practical point holds at any scale: execution cost changes the amount at risk.
A trader who checks only the chart may see a clean setup. A trader who checks the order sees another decision:
- What are the current bid and ask?
- How wide is the spread relative to the planned stop distance?
- Is the order allowed to fill at any available price?
- Does the position size still fit the risk budget at a realistic fill?
- Would waiting for another session invalidate the setup, or merely delay it?
Direction is one field in the review. Execution belongs in the same review.
Friday’s close changes the approval decision
A queued signal records an analysis at a point in time. It does not freeze the order book.
By late Friday, the quoted spread may be wider than it was when the signal was generated. Available size may also differ across price levels. A market order prioritizes getting filled, so the final price can depart from the reference used in the risk calculation.
This creates three separate checks.
First, recalculate risk from a realistic entry. For a long position, that usually means examining the ask and considering how far the order could move through available liquidity. For a short position, inspect the bid and the constraints that apply to the order.
Second, compare the spread with the stop distance. A $0.10 spread deserves different treatment when the planned risk is $0.25 per share than when it is $2.00. The absolute spread matters less than its share of the risk budget.
Third, decide whether the order type expresses the actual intent. A limit order can set a maximum purchase price or minimum sale price, but it may remain unfilled. A market order raises fill probability while surrendering price control. Neither choice removes uncertainty.
This is why an approval gate matters. It creates a pause between analysis and execution, where the trader can reject, resize, reprice, or defer the order. The gate has value only when the review covers the conditions that determine the fill.
For another example of timing changing planned exposure, see Lena’s stale signal, where the new entry tripled her planned risk.
Add execution cost to the risk calculation
A practical pre-trade estimate can include four components:
Planned risk = position size × (expected entry price − stop price), plus estimated spread and slippage costs.
The exact formula varies with direction, fees, and instrument. Its purpose is simple: stop treating the signal price as a promised fill.
Record the estimate in the trading journal before approval. After execution, record the actual fill and compare it with the reference price. Over time, this creates a visible measure of execution quality by instrument, session, order type, and day of the week.
That record can answer useful questions. Do Friday approvals produce larger deviations? Are market orders repeatedly consuming too much of the planned risk? Does a particular instrument become difficult to trade when quoted size falls?
A backtest may not capture these costs with enough realism. Historical candles often show traded prices without reconstructing the exact spread and available liquidity that a live order faced. Testing survival therefore requires assumptions about execution alongside return and drawdown. The same discipline appears in this guide to testing max drawdown rather than returns alone.
Knight Capital’s failure was larger and technically different. Its lesson still survives the comparison: reviewing the intended instruction cannot compensate for an unchecked execution state.
Before approving the next late-session signal, write down three prices: the signal reference, the current executable quote, and the worst acceptable fill. If the position no longer fits the risk budget at the third price, resize it, reprice it, or reject it.
Educational content, not financial advice.
Comments
No comments yet.