Как устроены данные
Monthly total returns (dividends and coupons included), nominal, USD, to 2026-09. Sources are open: Kenneth French Data Library, Aswath Damodaran, the Federal Reserve, BLS, Nareit, World Bank, Yahoo Finance. Before the first ETFs, bonds are built from yields; stand-ins are marked in every result.
| Class | From | Source | Caveats |
|---|---|---|---|
| Крупные компании США (S&P 500) | 1926-07 | Fama/French research factors: CRSP value-weighted US market = Mkt-RF + RF (1926-07 to French's last month); ^SP500TR (S&P 500 Total Return index), only to fill the months after French's last month CRSP total US market instead of the S&P 500 before 1988 (monthly S&P 500 total returns before 1988 are not freely available at month-end precision). Over 1988-2026 the two have correlation ~0.99 and annual return difference ~0.0-0.1% (see overlap). | 1926-1956 the S&P index had 90 stocks; CRSP covers the whole market, so single years can differ by a few points from the S&P figures (see Damodaran table). Pre-1926 (1871-1926) monthly series from Shiller is stored separately in assets_monthly_pre1926.csv: Shiller uses monthly AVERAGE prices, so month returns are smoothed (corr with CRSP only ~0.7 in 1926-1950, vol ~16% too low); fine for annual and longer horizons only. |
| Малые компании США | 1926-07 | French 6 portfolios (2x3 size/book-to-market), value-weighted: simple average of the three SMALL portfolios (Lo BM, neutral, Hi BM); IWM (Russell 2000), only for months after French's last month Small = below the NYSE median market cap (includes micro caps), equal-weighting the three value groups. It is a broader/smaller universe than the Russell 2000. | Over 2000-2026 this series beat IWM by roughly 1.9% a year (micro-cap/value tilt, no costs); do not read it as the return of an investable fund. |
| Крупные «стоимостные» акции США | 1926-07 | French 6 portfolios: BIG HiBM (large cap, top 30% book-to-market, value-weighted); VTV, tail only Fama-French BIG value portfolio. | Much more extreme value tilt than VTV: vol ratio ~1.4 in the overlap, ~1.3% a year higher return, corr 0.93. |
| Крупные акции роста США | 1926-07 | French 6 portfolios: BIG LoBM (large cap, lowest 30% book-to-market, value-weighted); VUG, tail only Fama-French BIG growth portfolio. | Book-to-market growth is not identical to index-provider growth (e.g. Russell/CRSP growth); corr with VUG 0.99. |
| Малые «стоимостные» акции США | 1926-07 | French 6 portfolios: SMALL HiBM (small cap, top 30% book-to-market, value-weighted); VBR (CRSP small-cap value), tail only Fama-French SMALL value portfolio. | Includes micro caps and has no costs; tracking error vs VBR/IJS 5-6% a year. |
| Развитые страны вне США | 1990-07 | French Developed ex-US market (Mkt-RF + RF), USD, value-weighted, from 1990-07; EFA, tail only Fama-French developed ex-US market portfolio (not the MSCI EAFE index). | Not found free before 1990-07: msci.com index-level download returned an error from this machine, no other free monthly source with total return. Beats EFA by ~1.1% a year in 2001-2026 (index vs fund, different universe/weights). |
| Развивающиеся страны | 1989-07 | French Emerging Markets market (Mkt-RF + RF), USD, from 1989-07; EEM, tail only Fama-French emerging market portfolio (not MSCI EM). | The very first years cover only a handful of countries (small, thin market sample); treat 1989-1994 with care. Beats EEM by ~1.6% a year in 2003-2026 (no fund costs, different weights). |
| Недвижимость (REIT) | 1972-01 | FTSE Nareit All Equity REITs total return, monthly, from 1972-01 (reit.com, official monthly workbook); VNQ, tail only None; equity REITs only (mortgage REITs excluded). | Before 1972 no REIT index exists in free data; the sector was tiny before 1990 (few listed equity REITs in 1970s). |
| Короткие гособлигации | 1961-07 | Gurkaynak-Sack-Wright (Federal Reserve) par yield, 2-year, daily from 1961-06, month-end; converted to returns of a constant 2-year par bond; H.15 constant-maturity 2-year yield (1976+) used only as cross-check/fill Constant-maturity 2-year par bond built from yields, standing in for the 1-3 year Treasury index (SHY). | Model return: assumes constant maturity and ignores roll-down, bid/ask, and the yield difference between old and on-the-run issues. Before 1961-06 no Treasury curve data free at monthly frequency; Damodaran has only the 10-year; the class therefore starts 1961-07. |
| Гособлигации средней срочности (10 лет) | 1926-07 | Fed H.15 constant-maturity 10-year Treasury yield, daily from 1962, month-end -> constant 10-year par bond; Shiller GS10 (long-term US government bond yield, MONTHLY AVERAGE, from 1871) for 1926-07 to 1962-01 A 10-year par bond stands in for the 7-10 year Treasury index (IEF). The 10-year has ~1 year more duration than IEF (vol ratio 1.14 vs IEF). | Pre-1962 yields are monthly averages (smoothed month returns; annual returns fine) and before ~1953 the series is not a pure 10-year Treasury (Shiller's long-rate splice). Annual deviation from Damodaran (mean abs 1.0 pt, max 6.6 pt in 1982) comes mainly from different yield inputs: our formula applied to Damodaran's own year-end yields reproduces his returns to 0.35 pt on average (see checks.bond_formula_vs_damodaran_own_yields). |
| Длинные гособлигации (20+ лет) | 1962-02 | GSW par yield 25-year (1985-12+): constant 25-year par bond; H.15 CMT 20y and 30y yields, average = interpolated 25-year yield, 1977-02 to 1985-11; H.15 CMT 20-year yield (1962-02 to 1977-01): constant 20-year par bond 25-year par bond for the 20+ year Treasury index (TLT, avg maturity ~25y); a 20-year bond for 1962-1977 (rate sensitivity ~13% lower). | Starts 1962-02: no long Treasury yield series free at monthly frequency before 1962 (Damodaran has only 10-year). Constant-maturity model without roll-down; TLT has slightly higher duration in low-yield years. |
| Рынок облигаций США (весь) | 1987-01 | AGG (iShares Core US Aggregate Bond), 2003-10+; VBMFX (Vanguard Total Bond Market Index Investor), 1987-01 to 2003-09 VBMFX (mutual fund, net of ~0.2% expenses) before 2003. | Bloomberg US Aggregate index history before 1987 is not free; no honest substitute built, so class starts 1987-01. Yahoo does not provide earlier fund history than ~1985. |
| Корпоративные облигации | 1993-11 | LQD (iShares iBoxx IG corporate), 2002-08+; Average of VWESX (Vanguard Long-Term Investment-Grade) and VFICX (Vanguard Intermediate-Term Investment-Grade), 1993-11 to 2002-07 50/50 blend of a long and an intermediate investment-grade fund to approximate LQD's duration (~8.5y). | Funds carry expenses (0.1-0.2%). Moody's Aaa/Baa monthly yields (FRED) were not reachable; Damodaran's annual Baa series is used only as a check. Blend corr with LQD about 0.93-0.95; credit spread behaviour in 1990s may differ from iBoxx. |
| Облигации с защитой от инфляции (TIPS) | 2000-07 | TIP (iShares TIPS Bond), 2004-01+; VIPSX (Vanguard Inflation-Protected Securities), 2000-07 to 2003-12 VIPSX fund (net of expenses). | US TIPS were issued from 1997, but no free monthly total return series exists before the Vanguard fund (2000-06). |
| Наличные (казначейские векселя) | 1926-07 | French RF = 1-month Treasury bill return (Ibbotson/CRSP), 1926-07+; BIL, tail only none | Before 1926 no free monthly bill series. |
| Золото | 1968-04 | GLD (SPDR Gold Shares) month-end, 2004-12+; GC=F (COMEX gold futures, front month) month-end, 2000-09 to 2004-11; World Bank Pink Sheet gold price, monthly average, USD/oz, 1968-03 to 2000-08 Spot price return (gold pays no income). Pre-2000 uses monthly AVERAGE prices (not month-end) -> smoothed month returns; the gold price in the World Bank file is rounded to whole dollars (matters at $35-$45: +-1.3% monthly noise in 1968-1972). | Before March 1968 the official price was fixed (US$20.67 until Jan-1934, then $35 until the 1968 two-tier market; US residents could not hold gold until 1974): no free-market monthly series is used. A 'flat' 0% series for 1934-02..1968-02 would be the official price, but that is not a market return, so it is NOT included. GC=F front-month continuous contract includes small roll effects (2000-09 to 2004-11). Monthly averages vs month-end: correlation of the World Bank monthly-average returns with GLD month-end returns is only ~0.63 (tracking error 13% a year), so 1968-2000 monthly gold returns are smoothed; use annual or longer horizons for risk statistics. |
| Сырьё | 1984-02 | ^SPGSCI (S&P GSCI, excess-return index) month-end, from 1984-01 (Yahoo); plus French RF (T-bill return) as collateral return: total return = (1+ER)(1+RF)-1 Total return of fully collateralised GSCI futures = excess return compounded with T-bill return. | S&P GSCI is heavily weighted to energy (more than DBC or BCOM). DBC (2006+) is shown only as an overlap check; DBC has lower total return (2.8% vs 4.5% a year) and 1.2x lower vol. Starts 1984-02 (Yahoo history limit for the index); the S&P GSCI began in 1970 but free monthly levels were not found. The Yahoo ER series was assumed to be the excess-return index; the DBC overlap (annual return diff vs GSCI ER only 0.02%) is consistent with that. |
Расчёт по историческим данным. Прошлые результаты не предсказывают будущее. До 1990-х годов для классов, у которых ещё не было индекса, использованы заменители; комиссии, налоги и издержки фондов не учтены. Это математика, а не инвестиционная рекомендация.