Case 034Operational risk and loss eventsCore
A dealer keys a sell order for 10 lakh shares instead of 1,000. Half fills before it is cancelled and the firm buys the shares back higher. Compute the loss and name the controls that would have blocked the order.
1The situation
A dealer at Deshvel Securities means to sell 1,000 shares of a stock trading at Rs 250 for a client. He types 10,00,000 into the quantity field and sends a market order: about Rs 25 crore of stock instead of Rs 2.5 lakh.
The order sweeps the buy side of the book. By the time a supervisor cancels it, 5,00,000 shares have been sold at an average of Rs 242.5. Deshvel now has a short position it never wanted and buys the 5 lakh shares back over the next half hour at an average of Rs 251, as the price recovers. The client's genuine 1,000-share order is then re-entered separately.
2Your task
What did the error cost, why did the firm lose on both legs, and which pre-trade controls would have stopped the order before it reached the exchange?
Quick check
What is Deshvel's loss on the error?
Worked solution
Try it on paper, then open one step at a time.
30-second answerThe answer to give first
The error cost Rs 42.5 lakh: 5 lakh shares sold at Rs 242.5 and bought back at Rs 251. The firm lost on both legs because its own order pushed the price down while selling and the recovery pushed it up while buying back. Three ordinary pre-trade checks, a quantity cap, a value cap and a limit relative to daily volume, would each have rejected an order a thousand times the intended size before it reached the market.
Step 1Where exactly does the loss come from?
The unwanted trade is a round trip: sell shares Deshvel did not mean to sell, then buy them back. The loss is the gap between the two prices times the shares, Rs 8.5 times 5 lakh, which is Rs 42.5 lakh. Deshvel received Rs 12.125 crore on the sale and paid Rs 12.550 crore on the buyback. The value traded, over Rs 12 crore, is not the loss; it is the size of the mistake.
Step 2Why did the firm lose on both legs?
A market order to sell far more than the book can absorb walks down the bids: the first shares sell near 250, the last near 235. That is why the average fill is 242.5. Once the order stops, other traders see the price as too low and buy, and the price climbs back, so the buyback pays up. A large error order sets its own bad price going in and meets the market's correction coming out. A shopkeeper who dumps a whole season's stock on one morning sells cheap, and then pays full price to restock the shelves when regular customers arrive.
Step 3Which controls would have stopped it?
Pre-trade controlsAutomatic checks an order must pass inside the firm system before it is sent to the exchange, such as caps on quantity, value and price distance. sit between the dealer's keyboard and the exchange. A cap on quantity per order, a cap on value per order and a check against the stock's average daily volume would each have rejected this order outright. A price collar would have converted the market order into a limit near the last price, so even an oversized order could not sweep the book. Above a threshold, a second person confirms. A kill switch lets a supervisor stop all of a dealer's orders at once, faster than cancelling one by one.
Step 4What would you report as the root cause?
Not the dealer's typing. People mistype; controls exist because they do. The root cause is that the firm let a single keystroke send an order a thousand times the dealer's normal size. The report should record the loss, the missing or disabled limits, why they were missing, whether other dealers have the same gap, and the date by which caps sized to each dealer's normal activity will be in place. Exchanges run their own checks too, but a firm that relies on the exchange to catch its errors has handed away its own control.
Where candidates lose it
Candidates often quote the value sold, over Rs 12 crore, as the loss, or compute only the fall from 250 to 242.5 and forget that the buyback cost more than the starting price.
The other weak answer blames the dealer and proposes training. Training reduces errors a little; hard limits at order entry make this particular error impossible, which is the answer interviewers look for.
What the interviewer asks next
- How would you size per-order limits so they block errors without blocking genuine large client orders?
- Should Deshvel have bought back all at once or over half an hour? What is the trade-off?
- Who bears the loss, the dealer's desk, the client or the firm, and why?
Company names and figures are illustrative.
