How factor ETFs are built: MSCI and S&P methodologies, screening versus tilting and factor purity
Read a factor index methodology, tell a screened index from a tilted one, and measure how much factor exposure a product really delivers.
What you will learn
- How the main index providers turn a factor definition into a list of stocks and weights.
- The difference between a screened index and a tilted index, and what each does to concentration and purity.
- How to read the key parameters of a methodology document: variables, scoring, selection count, caps, rebalancing.
- Two simple measures of real factor exposure: the weighted factor score and active share.
Explanation
A factor exchange-traded fund (ETF) does not hold "the factor" from lessons 1–6; it holds an index built by a provider such as MSCI or S&P Dow Jones Indices, and the index rules decide how much of the academic factor reaches your account. Market-cap index construction was covered in the course "ETF Investing"; here we look only at the factor layer on top.
The common recipe
Most single-factor indices follow five steps: a parent index (for example MSCI World or the S&P 500) defines the universe; each stock gets a raw score from one or more variables; raw scores become z-scores (value minus the average, divided by the standard deviation across the universe, cut off at plus or minus 3); a selection rule keeps the best-scoring stocks; a weighting rule sets their weights. Examples from the published methodologies as of 2026 (providers update them, so check the current documents):
| Index family | Variables | Weighting | Notable rules |
|---|---|---|---|
| MSCI Enhanced Value | Forward price-to-earnings, enterprise value to operating cash flow, price-to-book | Value score × market cap, then sector weights reset to match the parent | Fixed number of stocks, semi-annual review (May/November) |
| MSCI Momentum | 6- and 12-month price return minus risk-free rate, skipping the latest month, divided by 3-year volatility | Momentum score × market cap, 5 % issuer cap in broad indices | Extra ad-hoc rebalance if market volatility jumps |
| MSCI Quality | Return on equity, debt-to-equity, variability of earnings growth over five years | Quality score × market cap, 5 % issuer cap | Not sector-neutral in the standard version |
| S&P 500 Quality | Return on equity, balance-sheet accruals ratio, financial leverage ratio | Quality score × float-adjusted market cap | Top 20 % of the S&P 500, 100 stocks |
| MSCI Minimum Volatility | Risk-model optimisation, not a score | Optimiser output | Stock cap 1.5 %, sector and country deviation limits, one-way turnover limit |
The factor score used for weighting is usually 1 + z for positive z-scores and 1 / (1 − z) for negative ones, so a stock with z = 2 gets three times its cap weight and a stock with z = −1 gets half of it.
Screening versus tilting
Screening keeps only a subset (the top 20 % or a fixed count) and drops the rest. Tilting keeps most or all of the parent and re-weights it by score. Screening gives stronger factor exposure and a higher active share (half the sum of absolute weight differences against the parent), but more concentration and turnover. Tilting is cheaper and smoother but delivers less factor per euro. Many products are hybrids: a fixed-count screen with score-times-cap weighting.
Factor purity
A value screen on price-to-book alone loads on banks, utilities and cyclicals: a sector bet, not only a value bet. Providers fight this with sector neutrality (scores standardised within each sector, sector weights reset to the parent) and issuer caps. Without such rules a factor index can become very concentrated; a 2026 MSCI factsheet for a US value index showed one stock above 20 % of the index.
Measuring real exposure
Two quick checks from a holdings file: the weighted factor score (each holding's z-score times its weight, added up; the parent is near zero by construction) and active share against the parent (below about 20 % the product behaves like the market; above 60 % it is a concentrated bet). A regression check follows in lesson 9.
Worked example
Hypothetical example: a parent index with six stocks. Value z-scores are given. We build (a) a tilt of all six stocks and (b) a screen that keeps stocks with z ≥ 1 and cap-weights them. Fees, spreads and taxes are ignored.
| Stock | Parent weight | z | Score | Weight × score | Tilt weight | Screen weight |
|---|---|---|---|---|---|---|
| A | 30 % | −1.0 | 0.500 | 0.150 | 10.6 % | 0 |
| B | 25 % | 0.5 | 1.500 | 0.375 | 26.4 % | 0 |
| C | 15 % | 1.5 | 2.500 | 0.375 | 26.4 % | 45.5 % |
| D | 12 % | −0.5 | 0.667 | 0.080 | 5.6 % | 0 |
| E | 10 % | 1.0 | 2.000 | 0.200 | 14.1 % | 30.3 % |
| F | 8 % | 2.0 | 3.000 | 0.240 | 16.9 % | 24.2 % |
Step 1 – scores: A has z = −1, so score = 1 / (1 + 1) = 0.5; F has z = 2, so score = 3.
Step 2 – tilt weights: the products sum to 1.42; A's tilt weight = 0.150 / 1.42 = 10.6 %.
Step 3 – screen weights: C, E and F together have 33 % of the parent; C = 15 / 33 = 45.5 %.
Step 4 – weighted z: parent 0.25, tilt 0.87, screen 1.47. Step 5 – active share: tilt 25.8 %, screen 67.0 %.
Reading: the screen delivers almost twice the factor exposure of the tilt, but with three stocks and one position near 46 %.
Common mistake
Practice activity
On paper: a factsheet shows a momentum ETF with 120 holdings out of a 1,500-stock parent, top-10 weight 38 %, and the methodology says "momentum score × market cap, 5 % issuer cap, semi-annual review plus conditional rebalancing". (a) Is this a screen, a tilt or a hybrid? (b) Why does a momentum index need a conditional rebalance when a value index does not? (c) The parent's weighted momentum z is 0 and the ETF's is 1.1; the fee is 0.25 % against 0.12 % for the parent ETF. Write one sentence on what else you must know before deciding.
Show solution
(a) Hybrid: a fixed-count screen (120 of 1,500) with score-weighting. (b) Momentum signals change fast; after a volatility jump the previous winners can become losers (the momentum crash of lesson 3), so the provider allows an extra rebalance on 6-month momentum. Value scores move slowly. (c) You need the turnover and realised tracking difference of the ETF, because momentum turnover can cost more than the 0.13-point fee gap.
Limitations: active share and weighted z describe the holdings on one day; they say nothing about how the exposure will look after the next review.
Key takeaways
- Factor indices follow a recipe: parent universe, variables, z-scores, selection, weighting, caps and review dates.
- Screening gives more factor and more concentration; tilting gives less of both. Many products are hybrids.
- Sector neutrality and issuer caps improve purity and limit accidental bets.
- Check the weighted factor score and active share, not the product name.
Next step
In lesson 8 we combine several factors and compare mixing single-factor funds with integrated multi-factor construction.
Self-check
Answer the questions, then check your answers. You can retry as often as you like.
Automatic self-check for your own revision. Your answers are checked in your browser and are not seen or assessed by us. It is not an exam and not a qualification.
Your progress
“Complete” only means you marked the lesson as done. It is not a test of your understanding. Progress and notes are saved only in this browser on this device — not in an account.