A portfolio manager runs ten accounts in one strategy and advertises the best three, which averaged 24%. All ten averaged 13% against a 14% benchmark. What should a fair composite show, and why?

A fair composite shows about 12.0%, every account weighted by its assets, 2.0 points behind the 14% benchmark. The advertised 24% is three hand-picked accounts, and even the 13% simple average overstates the result, because the best accounts were among the smallest. The composite should also disclose the number of accounts and the spread of returns, 3% to 26%, which says the accounts were not really run alike.

Because the choice of which accounts to show was made after the results were known. Any manager with ten accounts has a best three; showing them measures luck in the spread, not skill in the strategy. A coaching class that advertises only its three toppers says nothing about what an average student gains. Performance is what the strategy did with all the money it ran, including the accounts that did badly and the ones that closed during the year.

The rules come from the Global Investment Performance StandardsA voluntary global standard for how investment managers calculate and present performance, maintained by CFA Institute. Firms claim compliance for the whole firm, not for one strategy., the voluntary standard many managers follow, and they are simple to state. Every actual, fee-paying, discretionary account run to the strategy goes in, returns are weighted by each account's assets, and accounts that closed stay in the history for the periods they were open. For Nakshatra that gives 12.0%. Indian portfolio managers also report performance to a format set by SEBI; confirm the current circular before quoting any rule.

Because the best accounts were small. The three advertised accounts hold Rs 7 crore of the Rs 40 crore, and the biggest account, Rs 8 crore, returned 12%. A simple average lets each Rs 2 crore account count as much as the Rs 8 crore one, so it tells you what a typical account did, not what the money did. Asset weighting multiplies each return by the account's share of assets and adds them up: 12.0%.

The number of accounts, the composite's assets, returns net and gross of fees, and a measure of dispersion. A range of 3% to 26% inside one strategy is itself a finding: accounts run to the same model should not differ by 23 points. It suggests client restrictions, different entry dates or discretion used unevenly, and a client choosing this manager is entitled to know which account they are likely to resemble. Say the limit too: one year of ten accounts is far too little to judge skill in either direction.