A value fund has beaten its benchmark by 3% a year, but its rolling value loading has fallen from 0.45 to minus 0.10 while its momentum loading rose from zero to 0.40. Is it still a value fund, what drove the returns, and what do you do?

No: on its loadings Nirvaan is now a momentum fund, and about 86% of its outperformance came from that momentum tilt. Momentum exposure times a strong momentum factor explains about 2.6 of the 3.1 points a year; value added about 0.1 and stock-picking about 0.3. Put the fund on watch, ask the manager what changed, and replace it as the value sleeve if the drift is deliberate.

A factor loadingHow strongly a fund returns move with a factor, such as cheap stocks beating expensive ones; 0.45 means the fund captures about 45% of that factor move. is how much of a style's return the fund picks up. A value loading of 0.45 means the fund rose about 0.45% for every 1% that cheap stocks beat expensive ones. Minus 0.10 means it now tends to lose a little when value wins, and a momentum loading of 0.40 means it rides recent winners. It is like hiring a cook for his vegetarian thali and finding, three years later, that he serves mostly grilled chicken: the diners are happy, but it is not the meal you ordered.

Multiply each year's average loading by that year's factor return, and whatever is left is stock selection. Momentum contributed about 0.2, 2.6 and 5.0 points in the three years, an average of 2.6 points a year, against 0.1 for value and 0.3 left for stock-picking. The outperformance grew exactly as the style drifted into a factor that was paying. Notice year three: with a slightly negative value loading, the fund even gained a little from value doing badly.

Because the fund was hired to do a job inside a portfolio, not to win on its own. The programme probably already holds growth or momentum managers; with Nirvaan drifting, it now holds momentum twice and value not at all. The risk shows up when momentum reverses: in a year when momentum loses 15%, a 0.40 loading costs Nirvaan about 6 points, just when the value sleeve was meant to cushion the portfolio. Good recent returns from a style that has been paying are not evidence of skill; they are evidence of exposure.

Three steps. First, check the model against the holdings: compare the portfolio's price to book and price to earnings with the value index, because a regression over 36 months lags and can be noisy. Second, meet the manager and ask directly whether the process changed, and why. A deliberate move into momentum is a mandate change, whatever the fund is called; confirm too whether the fund's own stated category or scheme documents constrain its style. Third, if the drift is confirmed and not reversed, replace Nirvaan as the value sleeve, in stages, with a manager whose loadings match the job. The limitation: factor models explain part of a fund's return, and a different model could split it differently, so treat the numbers as a strong signal, not a verdict.