It is quite common today to find firms revalue their portfolios for large cash flows. But what do we mean by “revalue”? Is “repricing” enough?

Think about this scenario:

• On January 31 the portfolio holds 20 different stocks, along with some cash.
• On February 15, a purchase is made for 500 shares of Dell (which remains, at least at the moment, a public company).
• On February 16 a large cash flow occurs and the portfolio is revalued. Since it’s a trade date system, the Dell position is included.
• On February 18, the trade is cancelled, because it’s determined that it was done for the wrong client. And so, the position is reversed.
• On February 28, the portfolio is revalued with closing prices.
• On March 5, the portfolio is reconciled with the custodian’s month-end positions and found to tie out 100%.

UNLESS the firm went back to the February 15 revaluation and backed out the Dell stock, that revaluation is wrong.

When we get involved with designing performance systems for clients, the reliability of the daily valuations is often an issue we address. In most cases I am not a fan of storing daily valuations, but rather to back into them, when necessary. If you’re using a method that revalues for large flows, you should ensure that this is more than just repricing; otherwise, some errors may be creeping in (here, you do all this additional work for a more accurate result, only to have an avoidable error appear). At a minimum it’s worth asking, what happens when a cancel/correct occurs?