What factors does your portfolio already hold?
Nobody chooses to hold a factor. A portfolio of shares picked one at a time has a size exposure, a value exposure and a momentum exposure whether or not anyone measured them, and the measurement is what turns "I hold twelve stocks I like" into something a risk report can read.
The regression, in words
Take the portfolio's monthly returns over three to five years. Take the same months' factor returns — market minus cash, small minus big, cheap minus expensive, winners minus losers. Ask, for each factor, how much of the portfolio's return moved with it. The answers are loadings: a value loading of 0.4 means the portfolio behaved like a portfolio holding 40% of a long-short value spread on top of everything else. A market loading is the beta from the performance course; the others are the same idea applied to each spread in turn.
What is left after all four is alpha, and the point of the exercise is how little of it there usually is. A manager who "picks cheap small companies" typically shows a positive size loading, a positive value loading and an alpha near zero: the skill was a rule, and the rule has a fund at a tenth of the fee.
Reading a loading
Three habits. Sign and size together: a value loading of −0.3 is a growth tilt, and a portfolio that never mentions growth can carry one, because the shares that seemed like good businesses were the expensive ones. The fit matters: a loading estimated on thirty-six months of a concentrated portfolio has a wide error band, and a loading that is not clearly different from zero is not a tilt. The universe matters: a loading measured against US factors tells you nothing about a portfolio of Japanese shares.
The tilt you did not choose
The commonest finding is a tilt nobody intended. A portfolio of "quality compounders" is a growth tilt and a quality tilt, and its underperformance in a value year is the tilt, not the picks. A dividend portfolio is a value tilt and a low-volatility tilt. A portfolio of the last five years' winners is a momentum tilt, and the spring-2009 lesson is waiting for it. Measuring the loadings does not say the tilt is wrong; it says the portfolio's next three years are partly written by a factor cycle its owner did not know they had bet on.
Without the regression
A rough read needs no statistics. Capitalisation-weight the portfolio's price-to-book and compare it with the market's: lower is a value tilt. Do the same with market capitalisation: lower is a size tilt. Do it with the past twelve-month return: higher is a momentum tilt. A holdings-based read is coarser than a returns-based one and available on any day from the holdings' published figures.
In the data
Measured on 2026-09-04, a two-name "portfolio" of AAPL.US and NVDA.US carried price-to-book ratios of about 44 and 23 against roughly 2.6 for XOM.US and 2.0 for VZ.US: the first pair is a growth tilt by any weighting, and anyone holding the two largest US companies holds it. The market's own tilt is in the ETF course's concentration lesson — 37.8% of an S&P 500 fund in ten names on the same day.
Try it now
- Take four holdings, read the price-to-book of each, and weight the four by rough position size. Four are below to start from: Apple, JPMorgan, NVIDIA and Exxon Mobil. To use holdings you own or would own instead, open one in the Terminal and change the symbol. Compare your weighted figure with the S&P 500's own price to book, the price-to-book row of SPY's portfolio valuation in the last table. Higher is a growth tilt; write the sentence.
Open AAPL.US — fundamentals in the EODHD Terminal
- Then the index your tilt is measured against — and the range at which the quarter of it that is ten names becomes visible:
Measure both over the same five years. The Nasdaq-100 fund is the S&P with a heavier growth tilt, and the gap between the two lines is what that tilt was worth in this window.