The Verification Venue · a windfall, deployed two ways
The Money Already in Your Hand
You just got a lump you already hold: an inheritance, a bonus, a rollover, the proceeds of a sale. Every instinct says ease it in slowly to be safe. Across 150 years of real market history, putting it all in today beat easing it in about two out of three times. Not because you are lucky, but because markets drift up, and money sitting in cash while you dribble it in is time out of the market you do not get back.
The honest verdict is a win-rate, not a slogan, so this page makes you turn the dial yourself. Race a windfall two ways across the real S&P total-return record, and you will find three things no thin explainer tells you at once: lump sum wins often, it wins small, and the exact fraction is a number you are choosing when you set the horizon, the allocation, and the spread. The third of windows dollar-cost averaging (DCA) wins are the crashes, which is the honest case for it.
The size of the edge
--
when lump won it won by --; when DCA won it won by --
The worst window for lump (regret)
--
the crash DCA is built for
--
Cosmetic scale only. Every ratio and the win-rate are the same at $10k or $10M (the check proves it).
Equal monthly buys the windfall is eased in over. Longer spread means more time in cash, so lower expected return.
Total years the money rides after t0. 1yr reproduces the 2023 study; 10yr the 2012 study. Holding longer changes the size of the gap, not who won.
Blend toward a 10-year Treasury sleeve. 100% stocks is the primary number, reproducible from the single S&P file; the bond sleeve is a secondary proxy.
The honest default is the 3-month T-bill (10-year yield before T-bills existed). 0% exaggerates lump's edge, so it is here to be caught, not believed.
The span of historical start-months the race rolls across. The window must fit inside the data, so the latest start shrinks as you lengthen the hold.
On the defaults above (a $100,000 windfall, eased in over 12 months, then held 10 years, 100% stocks, cash earning the short rate, every start month from 1871 on) lump sum ends ahead in about 64% of windows. Hit As Vanguard 2012 and slide allocation to a 60/40 blend to land on their published 67%. Then drag the spread out to 36 months and watch lump win more surely, and watch the average edge grow. That is the whole argument, live.
Why putting it all in usually wins
The mechanism is not complicated, and it is the honest general statement the exact percentage is downstream of: stocks and bonds beat cash most of the time. Vanguard measured it directly over 1976 to 2022: US stocks beat the 3-month Treasury bill 76% of the time, and bonds beat it 68% of the time. So every month your windfall spends sitting in cash while you ease in is a month you probably gave up market return. Dollar-cost averaging a lump you already hold is, by construction, a decision to hold some of it in cash for a while. That is the drag. Lump sum wins whenever the market rose during the spread-in, which is most of the time; DCA wins in the minority of windows where the market fell while you were still easing in, so your later, cheaper buys paid off.
Three turns that make DCA defensible, not just wrong
1. Lump wins often, but small, and DCA wins the crashes
The win-rate is real, but read the magnitude before you act on it. In Vanguard's US 60/40 case a lump sum ended just 2.3% higher on average over a decade ($2,450,264 versus $2,395,824 on a $1,000,000 windfall). Our engine reproduces that: on the same frame the average edge is about 2%. And the third of windows DCA wins are not scattered randomly: they cluster on the crashes, because that is exactly when easing in beats going all in. Scrub the race to the worst window (the button does it for you) and you will see the DCA staircase, still half in cash, sail over the lump line as the market craters. That is a genuine reduction in worst-case regret bought at the price of a modest expected shortfall. It is insurance, not a mistake. Call it what it is.
2. The exact win-rate is a dial, so never trust a single figure
There is no one true percentage, and a page that quotes one is hiding the machinery. Move the dials above and watch it swing: about 67% is Vanguard's US 60/40, 12-month, 10-year figure; drag the spread to 36 months and lump wins even more reliably (Vanguard reported about 90% for that case; our single-index engine climbs to about 70%, and the gap is named in the check below); shorten to Morgan Stanley's seven-year periods and the edge falls to about 56%; take Vanguard's 2023 one-year global frame and it is about 68%. Same underlying truth (lump usually wins) with a very different number depending on horizon, allocation, and averaging window. So the honest answer hands you the dial, not a decimal.
3. Investing every paycheck is a different thing, and not a mistake
This is the single most common way the question is answered wrong. Putting a fixed slice of every paycheck into the market is also called dollar-cost averaging, and it is emphatically not what this study is about and not a mistake. You cannot lump-sum money you have not earned yet, so paycheck investing is simply investing-as-you-earn, the best thing to do with a stream of income. This whole page is about a lump you already hold and are choosing whether to deploy now or dribble in. Conflating the two turns a right answer into a wrong one. If your question is "should I invest each paycheck as it lands," the answer is yes, and it has nothing to do with the race above.
The two, side by side
| Lump sum (LSI) | Dollar-cost averaging (DCA) | |
|---|---|---|
| What you do with the windfall | Invest it all today | Ease it in over a fixed number of months |
| Money out of the market | None | The un-invested part sits in cash |
| How often it ended ahead (US, 60/40, 12mo, 10yr) | About 67% | About 33% |
| Average size of the edge over a decade | About 2.3% more | Behind on average |
| Which windows it wins | The rising markets (most of them) | The crashes (the minority) |
| What it really is | The return-maximizing choice on average | Regret insurance you pay a small premium for |
| Applies to a paycheck? | No, you have no lump to deploy | No, paycheck investing is a different thing and not a mistake |
The check: the win-rate recomputed in front of you
Nothing here is pre-baked. The page ships Robert Shiller's monthly S&P Composite series (price, dividends, CPI, 1871 to June 2024) plus the 3-month Treasury bill, reinvests dividends for a real total-return index, and derives the whole race, including the headline win-rate, from that array on every input change. For your current settings:
The offline gate reruns all of it, two independent ways (a ratio comparison and a full dollar-by-dollar simulation, which must agree), loading the same series.json: node research/dollar-cost-averaging-vs-lump-sum/verify-dollar-cost-averaging-vs-lump-sum.mjs.
Free choices and honest scope. Total return reinvests Shiller's monthly dividend (this matches Shiller's own published real-total-return column to within 0.05% a month, an independent check). The nominal vs real distinction does not change who won: both strategies end on the same date and are deflated by the same CPI, so the win-rate is identical either way; CPI is used only for real-dollar readouts. Sidelined cash earns the 3-month T-bill from 1934 on, and the bundled 10-year yield before T-bills traded (generous to DCA, which only understates lump's edge). The bond sleeve for allocations under 100% is a constant-maturity 10-year proxy built from the yield, not a specific bond fund. The named gap: Vanguard reported lump beating 36-month DCA in about 90% of US 10-year spans; our transparent single-index engine reproduces the direction (lump's edge rises as the spread lengthens) but reaches about 70%, not 90%. The difference is our Shiller-S&P-plus-Treasury proxy versus Vanguard's proprietary local 60/40 benchmark blend. We show our reproducible number and cite theirs rather than claiming a match. Past performance: this is one country's up-drifting history. In Vanguard's own words the result holds "if such trends continue"; the mechanism (assets beat cash most of the time) is the general statement, the exact percentage is historical.
What is exactly true here, and what is a model
Exactly true (the reproducible core). Over rolling historical windows on a real total-return series, investing a windfall all at once ended ahead of easing it in about two-thirds of the time, and the average edge is small. The mechanism is robust: because stocks and bonds beat cash across the sample, holding some of a windfall in cash while you ease in is expected opportunity cost. Lump wins the rising windows (most of them); DCA wins the falling windows (the crashes). Setting sidelined cash to 0% strictly raises lump's measured win-rate, so a page that quietly used 0% cash would be inflating the number. The dollar amount is cosmetic: the win-rate is identical at every windfall size.
A model, not a measurement (the exact decimals). The precise win-rate depends on the return series, the cash proxy, the bond proxy, and the exact DCA and holding conventions. Different honest choices move it a few points, which is the whole point of the dials. Our engine uses a single US equity index (Shiller S&P) where Vanguard used local stock and bond benchmarks, so our numbers land in the same neighborhood, not on the exact decimal.
Named simplifications. Monthly rebalancing for blended allocations; a constant-maturity 10-year Treasury proxy for the bond sleeve (income plus a duration-based price change, no specific fund); the 10-year yield as the pre-1934 cash stand-in; no taxes, no trading costs, no bid-ask, no fund fees; one country's history; and the future is assumed to rhyme with the past, which is an assumption, not a fact.