Quantlake DCA Study: Method and Fund List
Every Quantlake DCA study runs on one published list: 90 US-listed ETFs and 27 US mutual funds. This page holds the simulation rules, the selection rules, the limits and every fund.
Methodology version 1.1 · data through 30 September 2026 · fund net assets on 26 October 2026 · updated 27 October 2026
If a figure in a Quantlake study looks too good or too bad, this page is where to check it: every fund the studies use, the rules that chose them, and how prices become the figures quoted.
The Quantlake DCA Study runs every test on one published fund list, so a figure in one article can be checked against the others and against the price history. Each article names the segment of the list it uses. This page records additions and deletions from the list, and its date updates when the list changes.
How often this is refreshed. The study is refreshed once a quarter, in the month after a quarter ends, once that quarter's consumer prices are published: the prices extend by three months, to the quarter's last trading day, the figures are updated from them, and the date at the top of each page moves. The model-portfolio tracker is updated each month, after the month's consumer prices are published. The method, the rules and the list of funds stay as described here. The fund net assets quoted here move with it. They describe what the list covers and are never used as a weight, so a quarter of staleness in them would not change a result.
Methodology at a glance
How a plan is simulated
Dollar-cost averaging means investing equal amounts at regular intervals, whatever the market is doing; the definition and its sources are in Does Dollar-Cost Averaging Work? Evidence from ETFs and Funds.
We apply that rule the same way in every fund.
- A plan. $1,000 invested in one fund on the first trading day of every month, from a given starting month, for a fixed number of years. There is one plan per starting month in each fund; only the model portfolios spread each payment across several funds.
- Purchases. ETFs are bought at the day's opening price. Mutual funds price once a day, after the close, so they are bought at that day's net asset value.
- Prices. End-of-day prices adjusted for splits and dividends, so dividends are reinvested; the opening price carries the same adjustment. Fund fees are already inside these prices and are not deducted again.
- Trading cost. A spread of 1 basis point on every purchase and sale, 10 cents on a $1,000 payment. The same rate is applied to every fund, at the level of the most heavily traded ETFs, so it understates the spread on the least traded ones.
- Cash. Money that waits, under a timing rule or while a lump sum is spread, earns the 3-month Treasury bill rate less 10 basis points a year, between the fees of two large Treasury-bill ETFs, 0.07% and 0.14% a year.
- Inflation. US consumer prices (CPI-U, seasonally adjusted, latest revision), each month's figure applied to every day of the month.
- Window. From each fund's first price to 30 September 2026. The model-portfolio tracker runs to the latest month end.
- Taxes are not modelled. Taxes on dividends and on sales are left out of every figure. Depending on the account and the country, they can change the final outcome.
Falls and market-wide episodes
A fall is a drop of 10% or more from a fund's high to its low. A fall ends when the price regains the high, and one still open at the end of the data is kept. The ETFs had 625 such falls with at least a year of prices before the high. Those falls bottomed in 29 different calendar years. The calendar-year count says little about how many separate crashes there were.
So the count uses market-wide episodes instead: stretches when most of the list fell together. Each day, each ETF is measured by its distance below its own highest close to date, and the ETFs are ranked from least to most. The median ETF is the one in the middle: half the list is down at least that far. A market-wide episode is a run of days when that median ETF stands 10% or more below its high. Two episodes less than three months apart count as one, and the sample uses only days with at least 10 ETFs priced. There were 15 since the end of 1998, 5 of them deeper than 20%, and 461 of the 625 fund falls bottomed inside one.
How a "typical" figure is computed
Every figure is computed for each fund first, from one plan per starting month, and "typical" means the median, with its population named: the median fund, the median fall, or the median start month. Each fund counts once, whatever its history length. Counting every plan once instead moved the headline figures by at most 1.1 points.
A fund's size today carries information about how it did: the funds that performed well drew money in. Weighting past plans by present assets would let the outcome pick the weights, so that weighting is excluded here.
How the universe was built, and what was cut from it
The universe was built in two steps. The first is the list Quantlake follows day to day: 112 US-listed ETFs, chosen for a clean split by asset class, index funds preferred and factor styles represented. It covers US and international shares across company size (large, mid and small) and style (growth, value, dividend, quality, momentum and low volatility), US sector and single-country funds, bonds across interest-rate and credit risk, commodities, real estate, multi-asset funds and alternatives. That step is a judgment about what a diversified list needs to hold, made long before this study.
The second step is mechanical and specific to this study. Three rules removed 22 of the 112, using a fund's type and history length, never its returns:
- By type. Volatility and inverse products and currency trackers, whose price follows a futures roll or an exchange rate.
- By mandate. Thematic funds, miners, a single-commodity oil fund and cash equivalents, since the studies build plans from diversified funds.
- By length. ETFs with less than the five years of prices a five-year plan needs.
That leaves our ETF universe: 90 US-listed funds, all but 3 (DBMF, FTLS, JEPI) tracking an index, with yearly fees from 0.03% to 1.46%. Where several ETFs follow the same index, the list holds one: SPY for the S&P 500, without VOO or IVV. Our mutual fund universe adds 27 funds with longer histories, on the same rules.
The ETFs held $5.3 trillion of investors' money on 26 October 2026, a median of $24 billion each, the smallest CPER at $679 million. The mutual funds held $2.8 trillion across their share classes, a median of $30 billion, the smallest FGOVX at $1.3 billion. Net assets are a snapshot taken on the date above and are never used as a weight.
Survivorship, and what these lists are made of
A fund that closed or merged is not in the data, so every fund here survived to 2026. That flatters the figures. No correction is applied.
The lists hold mainstream exposures: broad indices, sectors, countries, styles and the main bond markets, with 87 of the 90 ETFs tracking an index. Thematic funds, miners, a single-commodity oil fund, volatility and inverse products and currency trackers are excluded by rule, before any result is read. Each fund carries at least five years of prices, the median ETF 21 years, and the median ETF holds $24 billion.
The funds that close have a different profile. Of about 150 active ETFs shuttered in 2025, only six held more than $50 million at the start of that year and most held under $25 million, and the typical closed fund had existed for about 1.75 years (Morningstar, February 2026). Those figures cover active ETFs, where launches and closures are most frequent; the ETF list is almost entirely index funds. The bias is smaller here than in a list drawn from every fund ever launched. It is not zero, and nothing corrects for it.
What this method does not claim
- It is not every ETF ever launched, and closed funds are not in it.
- It does not model taxes, broker commissions, ticket minimums or platform fees.
- It does not weight results by fund size, so a $700 billion fund and a $1 billion fund count alike.
- It does not blend the S&P Composite record into the fund results; that record is reported on its own.
- "Typical" is a median across funds, falls or starting months.
- A share measured across past starting months is a frequency in this record. It gives no probability for one investor.
- It assumes the plan is never interrupted. A backtest does not lose its income, change its goals, need the money early or stop after a loss. These figures measure the rule as followed, and what departing from it cost is measured separately, on Does Dollar-Cost Averaging Work?.
Two returns: money-weighted and time-weighted
The money-weighted return (MWR) is the annual rate that turns each payment, counted from its own date, into the final value: the return on the investor's money, which depends on when the money went in. The time-weighted return (TWR) removes the payments and measures the fund or portfolio itself, the figure a fund reports.
A monthly plan in a fund that falls after most of the money has gone in has an MWR below the fund's TWR. In short, the MWR is what the investor's money earned, and the TWR is what the fund earned. The model-portfolio tracker quotes both side by side. Elsewhere, fund returns are time-weighted, and plans are measured against the money paid in.
The record since 1871 and the losing decades
Two descriptive studies sit beside the lists without joining them.
The long record runs the same monthly plan on Robert J. Shiller's S&P Composite data (shillerdata.com), January 1871 to June 2026. Each month's price is the average of that month's daily closes, and one twelfth of the annual dividend is reinvested each month. Before any result was read, the total return rebuilt from his columns was checked against his own real total-return series, to within one part in a million.
This record ends earlier than the fund data above. His file carries prices past June 2026 but no dividends for those months, and a total return cannot be built without them, so the series stops where his dividends stop. The fund figures on the study pages run to 30 September 2026.
The losing decades are the ten-year windows in which a fund on the list lost money, dividends included; they stand in, in part, for the funds that closed and are missing from the data. Both are quoted in Does Dollar-Cost Averaging Work? and in the crash study, and their plans' returns are money-weighted.
Data
Each daily update checks that every recent split and dividend has reached the adjusted series before it is accepted. The study reads a frozen copy of those prices, extended at each refresh by chaining the new returns onto the old history, so a revision of past prices is reviewed before it is adopted. Treasury bill rates and consumer prices come from FRED.
Pages that use these figures
- Does Dollar-Cost Averaging Work? Evidence from ETFs and Funds
- What a Market Crash Does to a Monthly Investment Plan
- Lump Sum vs Dollar-Cost Averaging: A Study on ETFs and Funds
- What If You Invested $1,000 a Month in an ETF Portfolio?
The full list
Every fund, by ticker, with its name and history below.
The 90 ETFs
Equity (58)
Fixed income (12)
Credit (7)
Commodities (5)
Multi-asset (3)
Real estate (3)
Alternatives (2)
The 27 long-history mutual funds
The first ETF starts in 1993, so every ETF plan misses the crashes before it. The mutual funds carry the same exposures further back. 7 of the 27 have prices from 1979, the start of the data; the rest start at their launch, half of them by 1990, and the youngest, VIPSX, in 2000. 24 of the 27 are Vanguard funds; 12 are actively managed, 11 of them among the 12 bond funds. Where several funds track the same index, the one with the longest history is kept.
Frequently asked questions
Related
- Does Dollar-Cost Averaging Work? Evidence from ETFs and Funds
- What a Market Crash Does to a Monthly Investment Plan
- Lump Sum vs Dollar-Cost Averaging: A Study on ETFs and Funds
- What If You Invested $1,000 a Month in an ETF Portfolio?
- Quantlake model portfolios: Classic and Smart
- Compound interest calculator with volatility, inflation and fees
Join our Community of Smarter Investors
