Target-Close Algorithms

Many institutional orders are explicitly benchmarked against the closing price — index funds rebalancing to match a benchmark's close, for instance — so the execution goal isn't to minimize impact across the whole day but to land as close as possible to whatever price the closing auction actually prints. A target-close algorithm manages this by scheduling participation to finish the bulk of the order before the close, then routing a calibrated slice directly into the closing auction itself, rather than trying to execute continuously at a steady rate like a VWAP algorithm would.

The sizing of that final auction slice is the key design choice: too large and the order itself becomes a meaningful fraction of auction volume, moving the very price it's trying to match; too small and too much of the order gets filled earlier in the continuous session at prices that may drift away from the eventual close, creating tracking error against the benchmark. Algorithms typically size the auction participation using historical closing-auction volume patterns and current order-book imbalance signals to estimate how large a slice the auction can absorb without excessive price impact.

Worked example. An order to buy 200,000 shares of a stock whose closing auction typically prints 800,000 shares might route 40,000-60,000 shares (5-7.5% of typical auction volume) directly into the close, having already worked the remaining 140,000-160,000 shares earlier in the day using a schedule tied to historical intraday volume curves.