This is from the text in Example 5 in the book
A broker with discretion on timing can also try to improve performance relative to a VWAP benchmark, because VWAP is partly determined at any point into the day. For example, if a buy order is received near the end of the day and the stock’s ask price exceeds the VWAP up to that point, the broker might try to move the order into the next day, when he will be benchmarked against a fresh VWAP.
In the solution to the same:
The CIO’s statement is correct. In contrast to the VWAP, which is partly deter- mined as the trading day progresses, the opening price is known with certainty at any point into the trading day, making it easier to game.
=====
Later in the white text immediately following :
To address the possibility of gaming VWAP, VWAP could be measured over mul- tiple days (spanning the time frame over which the order is executed), because traders would often be expected to try to execute trades within a day. However, the cost of measuring VWAP over a longer time frame is less precision in estimating trading costs.
=======
and in the page just preceding
Most traders measure implicit costs (i.e., costs excluding commissions) with ref- erence to some price benchmark or reference point. We have already mentioned one price benchmark: the time-of-trade midquote (quotation midpoint), which is used to calculate the effective spread. When such precise information is lacking, the price benchmark is sometimes taken to be the volume-weighted average price (VWAP). The VWAP of a security is the average price at which the security traded during the day, where each trade price is weighted by the fraction of the day’s volume associated with the trade. The VWAP is an appealing price benchmark because it allows the fund sponsor to identify when it transacted at a higher or lower price than the security’s average trade price during the day. For example, if a buy order for 500 shares was exe- cuted at €157.25 and the VWAP for the stock for the day was €156.00, the estimated implicit cost of the order would be 500 × (€157.25 – €156.00) = €625.15 If explicit costs were €25, the total estimated cost would be €650. Alternative price benchmarks include the opening and closing prices for a security, which use less information about prices and are less satisfactory. Although VWAP involves a data-intensive calculation, a number of vendors supply it.16 VWAP is less informative for trades that represent a large fraction of volume. In the extreme, if a single trading desk were responsible for all the buys in a security during a day, that desk’s average price would equal VWAP and thus appear to be good, however high the prices paid. Another limitation of VWAP (and of the effective spread) is that a broker with sufficient discretion can try to “game” this measure. (To game a cost measure is to take advantage of a weakness in the measure, so that the value of the measure may be misleading.) Furthermore, VWAP is partly determined at any point in the day; by using weights based on volume to that point in the day, a trader can estimate the final value of VWAP. The accuracy of such an estimate would tend to increase as the close of trading approaches. By comparing the current price to that estimate, the trader can judge the chances of doing better than VWAP.
==== Gaming of VWAP —> pretty important part of the LOSs if you have not read this before.