Bucket strategy or total return: does a cash bucket protect a retirement portfolio?
Once both portfolios start with the same cash, the bucket does not protect the retirement in these runs: it runs out more often, and what it changes is the shape of the outcome — a somewhat higher median in good markets, a worse tail in bad ones. For a 65-year-old with $1,000,000 spending $40,000 a year (4.0%) on a 60/40 plan, a 3-year cash reserve refilled after non-losing years reached 95 with money left in 85.2% of 10,000 paths, against 87.0% for a total-return portfolio holding the same 12.0% in cash and rebalanced every year. On the same market paths the bucket arm ended $10,705 more at the median (10th–90th percentile of the per-path difference −$40,903 to +$183,393) and finished higher in 57.4% of paths. It reached 95 with money left on 21 paths the total-return portfolio lost, and ran out on 198 paths the total-return portfolio kept. Across all 36 cells the bucket arm's success rate is never higher than its matched total-return portfolio's (lower in 36, equal in 0, by −0.5 pp to −5.9 pp); its paired median at 95 is positive in 26, exactly $0 in 10 and negative in 0.
Key numbers
| Figure | Bucket, 3-yr reserve | Total return, same starting cash | Paired, path by path |
|---|---|---|---|
| Success to 95 ($40,000/yr, 60/40, refill after non-losing year) | 85.2% | 87.0% | −1.8 pp; only bucket survives in 21 paths, only total return in 198 |
| Median wealth at 95 (today’s dollars) | $629,938 | $634,170 | +$10,705 median of the per-path difference; bucket higher in 57.4% of paths |
| 10th-percentile wealth at 95 | $0 | $0 | per-path difference p10 −$40,903, p90 +$183,393 |
| Same cell, refill when below half | 85.0% success; median $636,169 | 87.0%; $634,170 | +$11,710 median; higher in 56.9% |
| All 36 cells: paired median difference at 95 | $0 ($40,000/yr, 60/40, 5-yr, non-losing year) to +$28,803 ($35,000/yr, 80/20, 2-yr, non-losing year) | ||
| All 36 cells: share of paths the bucket ends higher | 32.5% to 73.0%; below one half in 10 of 36 cells | ||
| All 36 cells: success-rate difference, bucket minus total return | −5.9 pp ($45,000/yr, 60/40, 5-yr, non-losing year) to −0.5 pp ($35,000/yr, 60/40, 2-yr, non-losing year) | ||
Run the $1M / $40,000 / 65-to-95 plan in the free planner →
Opens the Monte Carlo planner prefilled with this study’s plan — retire at 65, plan to 95, $1,000,000, $40,000 a year, 45/15/40 US equity / international / bonds. The link sets the plan, not the rule: choose Bucket (cash reserve) under Withdrawal Rule and set the reserve in years to see the bucket arm for your own numbers.
What is being compared
The bucket arm is the engine’s two-bucket rule. At retirement it carves a cash reserve of N years of planned spending out of the portfolio and invests it in the roster’s cash asset; everything else stays in the plan mix as the growth bucket, rebalanced to that mix every year. Each month’s spending comes out of the reserve while it lasts, then pro rata from the growth bucket. At the end of every retirement year the reserve is topped back up to N years of that year’s spending under one of two rules: after a non-losing year (only if the growth bucket’s nominal return for the year was zero or better — after a losing year the reserve is left to run down, which is the whole point of the strategy) or when below half (whenever the reserve has fallen below half its target, whatever the year’s return). Spending itself is the same constant-real amount in both arms; the rule only changes where the money comes from.
The total-return arm is the plain constant-real rule the site’s planner runs by default, with one adjustment that makes the comparison fair: it holds the same cash. A 3-year reserve on $40,000 of spending is $120,000, 12.0% of the portfolio, so the baseline starts as 60/40 scaled to 88.0% plus 12.0% cash — 39.6% / 13.2% / 35.2% US equity / international / bonds and 12.0% cash — and is rebalanced back to that mix once a year. Every cell’s baseline is built the same way from its own reserve size and spending, so the two arms of every pair begin from an identical $1,000,000 with an identical cash share. What differs afterwards is exactly the bucket strategy: withdrawals from cash first, refills by rule, and a cash share that drifts instead of being rebalanced.
Both arms are run on the same 10,000 market paths, so every statistic below can be taken path by path. The headline for each cell is the paired difference: the bucket arm’s wealth at 95 minus the total-return arm’s on the same path, summarised by its median, its 10th–90th percentile band and the share of paths on which the bucket ends higher. That is the figure a single retirement could actually experience; the gap between two separately-sorted medians is not.
Results: 60/40 plan
45/15/40 US equity / international / bonds. Each row is one bucket configuration against its own total-return baseline. Dollar columns are wealth at 95 in today’s dollars; the last column counts paths in which only one arm reached 95 with money left.
| Spending | Reserve | Refill rule | Success: bucket | Success: total return | Δ success | Median at 95: bucket | Median at 95: total return | p10 at 95: bucket | p10 at 95: total return | Paired median Δ | Paired p10 … p90 | Bucket ends higher | Only bucket survives / only total return survives |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| $35,000 (3.5%) | 2 yrs (7.0% cash) | non-losing year | 94.1% | 94.6% | −0.5 pp | $981,839 | $957,458 | $127,371 | $140,963 | +$25,983 | −$16,087 … +$178,728 | 73.0% | 8 / 61 |
| $35,000 (3.5%) | 2 yrs (7.0% cash) | below half | 94.0% | 94.6% | −0.6 pp | $983,208 | $957,458 | $124,032 | $140,963 | +$26,279 | −$17,859 … +$199,536 | 71.1% | 3 / 66 |
| $35,000 (3.5%) | 3 yrs (10.5% cash) | non-losing year | 94.1% | 94.8% | −0.8 pp | $925,580 | $913,839 | $117,519 | $142,196 | +$19,478 | −$32,524 … +$201,670 | 65.8% | 14 / 89 |
| $35,000 (3.5%) | 3 yrs (10.5% cash) | below half | 93.9% | 94.8% | −0.9 pp | $931,451 | $913,839 | $111,252 | $142,196 | +$20,258 | −$33,991 … +$235,447 | 63.9% | 7 / 99 |
| $35,000 (3.5%) | 5 yrs (17.5% cash) | non-losing year | 93.8% | 95.3% | −1.5 pp | $812,089 | $830,549 | $95,203 | $143,190 | +$3,571 | −$73,136 … +$219,191 | 52.3% | 16 / 164 |
| $35,000 (3.5%) | 5 yrs (17.5% cash) | below half | 93.6% | 95.3% | −1.6 pp | $842,929 | $830,549 | $97,134 | $143,190 | +$22,184 | −$58,658 … +$322,046 | 61.1% | 10 / 173 |
| $40,000 (4.0%) | 2 yrs (8.0% cash) | non-losing year | 86.2% | 87.0% | −0.8 pp | $691,300 | $678,487 | $0 | $0 | +$21,051 | −$18,719 … +$173,196 | 66.2% | 19 / 97 |
| $40,000 (4.0%) | 2 yrs (8.0% cash) | below half | 86.1% | 87.0% | −0.9 pp | $698,609 | $678,487 | $0 | $0 | +$21,765 | −$20,705 … +$198,312 | 65.1% | 11 / 98 |
| $40,000 (4.0%) | 3 yrs (12.0% cash) | non-losing year | 85.2% | 87.0% | −1.8 pp | $629,938 | $634,170 | $0 | $0 | +$10,705 | −$40,903 … +$183,393 | 57.4% | 21 / 198 |
| $40,000 (4.0%) | 3 yrs (12.0% cash) | below half | 85.0% | 87.0% | −2.0 pp | $636,169 | $634,170 | $0 | $0 | +$11,710 | −$41,931 … +$220,990 | 56.9% | 10 / 205 |
| $40,000 (4.0%) | 5 yrs (20.0% cash) | non-losing year | 83.1% | 86.8% | −3.7 pp | $506,199 | $546,771 | $0 | $0 | $0 | −$96,339 … +$179,711 | 42.5% | 19 / 393 |
| $40,000 (4.0%) | 5 yrs (20.0% cash) | below half | 83.2% | 86.8% | −3.6 pp | $532,993 | $546,771 | $0 | $0 | +$7,334 | −$73,525 … +$290,719 | 53.3% | 11 / 374 |
| $45,000 (4.5%) | 2 yrs (9.0% cash) | non-losing year | 73.2% | 74.2% | −1.0 pp | $413,060 | $404,336 | $0 | $0 | +$13,017 | −$15,337 … +$159,697 | 57.3% | 33 / 134 |
| $45,000 (4.5%) | 2 yrs (9.0% cash) | below half | 73.1% | 74.2% | −1.1 pp | $411,889 | $404,336 | $0 | $0 | +$13,934 | −$15,637 … +$189,824 | 57.5% | 30 / 137 |
| $45,000 (4.5%) | 3 yrs (13.5% cash) | non-losing year | 70.8% | 73.3% | −2.5 pp | $348,053 | $359,930 | $0 | $0 | $0 | −$42,592 … +$160,855 | 47.8% | 28 / 276 |
| $45,000 (4.5%) | 3 yrs (13.5% cash) | below half | 70.8% | 73.3% | −2.5 pp | $348,101 | $359,930 | $0 | $0 | $0 | −$39,156 … +$202,929 | 48.9% | 24 / 278 |
| $45,000 (4.5%) | 5 yrs (22.5% cash) | non-losing year | 65.2% | 71.2% | −5.9 pp | $204,208 | $275,136 | $0 | $0 | $0 | −$104,732 … +$130,934 | 32.5% | 26 / 620 |
| $45,000 (4.5%) | 5 yrs (22.5% cash) | below half | 66.7% | 71.2% | −4.5 pp | $241,487 | $275,136 | $0 | $0 | $0 | −$67,147 … +$249,226 | 44.3% | 29 / 480 |
Sensitivity: 80/20 plan
60/20/20 US equity / international / bonds. Same construction, a more equity-heavy growth bucket.
| Spending | Reserve | Refill rule | Success: bucket | Success: total return | Δ success | Median at 95: bucket | Median at 95: total return | p10 at 95: bucket | p10 at 95: total return | Paired median Δ | Paired p10 … p90 | Bucket ends higher | Only bucket survives / only total return survives |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| $35,000 (3.5%) | 2 yrs (7.0% cash) | non-losing year | 91.0% | 91.6% | −0.6 pp | $1,127,919 | $1,103,638 | $33,094 | $53,876 | +$28,803 | −$23,292 … +$300,524 | 69.0% | 8 / 70 |
| $35,000 (3.5%) | 2 yrs (7.0% cash) | below half | 90.9% | 91.6% | −0.7 pp | $1,125,945 | $1,103,638 | $30,653 | $53,876 | +$25,898 | −$26,903 … +$323,763 | 65.6% | 9 / 77 |
| $35,000 (3.5%) | 3 yrs (10.5% cash) | non-losing year | 90.9% | 91.9% | −1.0 pp | $1,066,039 | $1,058,115 | $27,870 | $61,395 | +$21,718 | −$43,256 … +$356,351 | 62.9% | 11 / 113 |
| $35,000 (3.5%) | 3 yrs (10.5% cash) | below half | 90.6% | 91.9% | −1.2 pp | $1,064,963 | $1,058,115 | $21,108 | $61,395 | +$17,640 | −$48,094 … +$385,284 | 58.9% | 6 / 131 |
| $35,000 (3.5%) | 5 yrs (17.5% cash) | non-losing year | 90.6% | 92.4% | −1.8 pp | $935,246 | $962,604 | $15,124 | $74,345 | +$2,464 | −$91,620 … +$410,371 | 51.1% | 13 / 195 |
| $35,000 (3.5%) | 5 yrs (17.5% cash) | below half | 90.5% | 92.4% | −1.9 pp | $963,956 | $962,604 | $14,012 | $74,345 | +$18,505 | −$81,902 … +$543,608 | 57.2% | 11 / 199 |
| $40,000 (4.0%) | 2 yrs (8.0% cash) | non-losing year | 83.1% | 84.0% | −0.9 pp | $825,877 | $809,666 | $0 | $0 | +$24,070 | −$24,329 … +$300,452 | 63.8% | 23 / 111 |
| $40,000 (4.0%) | 2 yrs (8.0% cash) | below half | 83.0% | 84.0% | −1.1 pp | $820,384 | $809,666 | $0 | $0 | +$20,937 | −$30,268 … +$325,350 | 60.9% | 16 / 122 |
| $40,000 (4.0%) | 3 yrs (12.0% cash) | non-losing year | 82.5% | 84.2% | −1.7 pp | $756,660 | $761,362 | $0 | $0 | +$12,518 | −$50,550 … +$339,629 | 56.4% | 21 / 189 |
| $40,000 (4.0%) | 3 yrs (12.0% cash) | below half | 82.1% | 84.2% | −2.1 pp | $753,029 | $761,362 | $0 | $0 | +$8,623 | −$57,840 … +$376,655 | 53.8% | 8 / 213 |
| $40,000 (4.0%) | 5 yrs (20.0% cash) | non-losing year | 80.5% | 84.2% | −3.8 pp | $608,480 | $661,095 | $0 | $0 | $0 | −$117,357 … +$357,433 | 44.5% | 20 / 397 |
| $40,000 (4.0%) | 5 yrs (20.0% cash) | below half | 80.4% | 84.2% | −3.9 pp | $632,587 | $661,095 | $0 | $0 | +$2,792 | −$96,962 … +$502,563 | 50.9% | 10 / 395 |
| $45,000 (4.5%) | 2 yrs (9.0% cash) | non-losing year | 72.0% | 73.3% | −1.3 pp | $525,146 | $512,827 | $0 | $0 | +$16,008 | −$19,402 … +$289,236 | 56.8% | 29 / 155 |
| $45,000 (4.5%) | 2 yrs (9.0% cash) | below half | 71.9% | 73.3% | −1.4 pp | $516,568 | $512,827 | $0 | $0 | +$12,355 | −$23,783 … +$318,071 | 54.9% | 20 / 162 |
| $45,000 (4.5%) | 3 yrs (13.5% cash) | non-losing year | 70.3% | 72.9% | −2.5 pp | $447,739 | $464,148 | $0 | $0 | $0 | −$50,225 … +$305,666 | 49.3% | 30 / 283 |
| $45,000 (4.5%) | 3 yrs (13.5% cash) | below half | 69.9% | 72.9% | −3.0 pp | $446,176 | $464,148 | $0 | $0 | $0 | −$54,683 … +$351,802 | 47.7% | 13 / 313 |
| $45,000 (4.5%) | 5 yrs (22.5% cash) | non-losing year | 66.0% | 71.4% | −5.4 pp | $284,013 | $371,350 | $0 | $0 | $0 | −$123,642 … +$286,060 | 37.0% | 23 / 566 |
| $45,000 (4.5%) | 5 yrs (22.5% cash) | below half | 66.5% | 71.4% | −4.9 pp | $318,878 | $371,350 | $0 | $0 | $0 | −$89,596 … +$449,667 | 44.1% | 19 / 510 |
The gap over time
The engine reports each path’s total wealth year by year, so the paired difference can be followed through retirement. The table shows, for the $40,000-a-year 60/40 plan, the median per-path gap (bucket minus total return, today’s dollars) at six ages and the share of paths on which the bucket arm is ahead at that age. The starting cash share is exact by construction; the engine’s public output does not carry the reserve’s balance afterwards, so the cash share’s later path is not in the dataset — what is shown is its consequence, the wealth gap it produces.
| Bucket configuration | Age 70 | Age 75 | Age 80 | Age 85 | Age 90 | Age 95 |
|---|---|---|---|---|---|---|
| 2 yrs, non-losing year (8.0% cash at 65) | +$4,081 72.7% higher | +$7,995 75.4% higher | +$11,728 75.2% higher | +$15,517 73.4% higher | +$18,771 70.5% higher | +$21,051 66.2% higher |
| 2 yrs, below half (8.0% cash at 65) | +$3,752 65.8% higher | +$7,712 70.2% higher | +$11,334 70.9% higher | +$15,048 70.6% higher | +$19,123 68.4% higher | +$21,765 65.1% higher |
| 3 yrs, non-losing year (12.0% cash at 65) | +$3,675 71.4% higher | +$7,095 73.0% higher | +$9,757 70.4% higher | +$11,943 66.9% higher | +$12,309 62.7% higher | +$10,705 57.4% higher |
| 3 yrs, below half (12.0% cash at 65) | +$3,320 64.3% higher | +$6,727 67.5% higher | +$9,176 66.3% higher | +$11,538 64.0% higher | +$12,055 60.2% higher | +$11,710 56.9% higher |
| 5 yrs, non-losing year (20.0% cash at 65) | +$2,615 67.1% higher | +$4,317 64.0% higher | +$4,384 58.2% higher | +$2,392 53.0% higher | $0 47.9% higher | $0 42.5% higher |
| 5 yrs, below half (20.0% cash at 65) | +$4,395 63.5% higher | +$8,977 65.7% higher | +$11,764 64.9% higher | +$12,855 61.2% higher | +$12,148 57.1% higher | +$7,334 53.3% higher |
What the bucket buys, and what it costs
What it buys
A place to take withdrawals from that is not the growth portfolio, and with it a different cash profile. After a losing year the reserve funds spending without selling equities or bonds at depressed prices; under the non-losing-year rule it is deliberately left to run down until the growth bucket has had a year that did not lose money. That mechanism carried 21 of 10,000 paths to 95 in the 3-year, $40,000, 60/40 cell that the matched total-return portfolio did not finish (3 to 33 paths per cell across the 36). The second thing it buys is less visible: because the reserve is a set number of years of spending rather than a constant share of wealth, a portfolio that grows carries proportionally less cash as the years pass. On the paths where markets are kind that is more equity exposure than the rebalanced baseline holds, and it is why the bucket arm ends higher at the median in 26 of 36 cells (+$10,705 in the anchor cell, up to +$28,803 at $35,000/yr on 80/20 with a 2-year reserve) and higher on more than half the paths in 26 of them.
What it costs
The tail. The same years-of-spending target works the other way when markets are unkind: a shrinking portfolio carries proportionally more cash, every refill after a loss sells growth assets to buy it, and the reserve never buys back into what fell — while the total-return baseline rebalances into the fallen assets once a year. So the bucket arm runs out more often in 36 of 36 cells (0 equal, none better), by −0.5 pp to −5.9 pp; its 10th-percentile wealth at 95 is lower in 12 cells and equal (at $0) in 24; and the paths only one arm survives run 8,143 to 609 in the total-return portfolio’s favour across all 36 cells (61 to 620 per cell against 3 to 33). In the anchor cell the per-path difference is −$40,903 at its 10th percentile against +$183,393 at its 90th. The larger the reserve, the larger both effects: at $40,000 on 60/40 with refills after non-losing years, the success gap is −0.8 pp for 2 years of cash, −1.8 pp for 3 and −3.7 pp for 5, while the paired median goes from +$21,051 to +$10,705 to $0 and the share of paths the bucket ends higher from 66.2% to 57.4% to 42.5%.
Where the difference is largest
The median advantage is largest where the plan is comfortable and the reserve small: 3.5% spending with two years of cash, where the reserve shrinks fastest as a share of a growing portfolio (+$25,983 at the median on 60/40, higher on 73.0% of paths). The cost is largest where the plan is fragile and the reserve large: 4.5% spending on 60/40 with a 5-year reserve (non-losing year) runs out 5.9 pp more often than its matched baseline and ends higher on only 32.5% of paths. At 4.5% spending the paired median is $0 in 8 of the 12 cells because both arms end at $0 on roughly a quarter of the paths, and the middle of the per-path distribution sits inside that tie.
When the refill rule matters
The two rules differ in what happens after a loss and in how much cash they hold on average. After a non-losing year refuses to refill while the growth bucket is recovering, so the reserve can run to empty in a long bear market; in every other year it tops the reserve back up to its full target, so it holds close to the full N years of cash most of the time. When below half refills only once the reserve is half gone, even in a losing year, which means selling growth assets after a loss — but between refills it lets the reserve drain to half, so on average it holds less cash. Path by path, that second difference is the one that shows: the below-half rule ends higher at the median in 11 of the 18 spending / mix / reserve combinations (the non-losing-year rule in 0, a $0 paired median in 7), by at most $29,716 for a 5-year reserve and at most $3,829 for a 2-year one; the below-half rule is ahead on 41.2% to 73.0% of paths. The success rates of the two rules never differ by more than 1.4 pp in any cell. Next to the reserve size, the refill rule is a second-order choice.
| Spending | Mix | Reserve | Success: non-losing year | Success: below half | Median at 95: non-losing year | Median at 95: below half | Paired median, non-losing minus below-half | Non-losing rule higher |
|---|---|---|---|---|---|---|---|---|
| $35,000 (3.5%) | 60/40 | 2 yrs | 94.1% | 94.0% | $981,839 | $983,208 | −$3,829 | 39.3% |
| $35,000 (3.5%) | 60/40 | 3 yrs | 94.1% | 93.9% | $925,580 | $931,451 | −$5,719 | 37.2% |
| $35,000 (3.5%) | 60/40 | 5 yrs | 93.8% | 93.6% | $812,089 | $842,929 | −$29,716 | 21.2% |
| $35,000 (3.5%) | 80/20 | 2 yrs | 91.0% | 90.9% | $1,127,919 | $1,125,945 | $0 | 41.5% |
| $35,000 (3.5%) | 80/20 | 3 yrs | 90.9% | 90.6% | $1,066,039 | $1,064,963 | −$540 | 40.7% |
| $35,000 (3.5%) | 80/20 | 5 yrs | 90.6% | 90.5% | $935,246 | $963,956 | −$27,218 | 24.6% |
| $40,000 (4.0%) | 60/40 | 2 yrs | 86.2% | 86.1% | $691,300 | $698,609 | −$1,391 | 35.1% |
| $40,000 (4.0%) | 60/40 | 3 yrs | 85.2% | 85.0% | $629,938 | $636,169 | −$3,424 | 32.3% |
| $40,000 (4.0%) | 60/40 | 5 yrs | 83.1% | 83.2% | $506,199 | $532,993 | −$29,422 | 16.6% |
| $40,000 (4.0%) | 80/20 | 2 yrs | 83.1% | 83.0% | $825,877 | $820,384 | $0 | 37.3% |
| $40,000 (4.0%) | 80/20 | 3 yrs | 82.5% | 82.1% | $756,660 | $753,029 | $0 | 35.9% |
| $40,000 (4.0%) | 80/20 | 5 yrs | 80.5% | 80.4% | $608,480 | $632,587 | −$24,811 | 20.0% |
| $45,000 (4.5%) | 60/40 | 2 yrs | 73.2% | 73.1% | $413,060 | $411,889 | $0 | 28.6% |
| $45,000 (4.5%) | 60/40 | 3 yrs | 70.8% | 70.8% | $348,053 | $348,101 | $0 | 24.5% |
| $45,000 (4.5%) | 60/40 | 5 yrs | 65.2% | 66.7% | $204,208 | $241,487 | −$18,041 | 10.5% |
| $45,000 (4.5%) | 80/20 | 2 yrs | 72.0% | 71.9% | $525,146 | $516,568 | $0 | 31.5% |
| $45,000 (4.5%) | 80/20 | 3 yrs | 70.3% | 69.9% | $447,739 | $446,176 | $0 | 29.2% |
| $45,000 (4.5%) | 80/20 | 5 yrs | 66.0% | 66.5% | $284,013 | $318,878 | −$11,348 | 14.0% |
What it means
- Match the cash before you compare. Much of the apparent difference between a bucket strategy and a total-return portfolio is the cash allocation, not the withdrawal rule. With the cash matched, the 3-year bucket at $40,000 on 60/40 changes the success rate by −1.8 pp and the median outcome by +$10,705.
- A bucket is a glide path in disguise. A reserve set in years of spending is a shrinking cash share in a rising market and a growing one in a falling market — the reverse of a rebalanced mix. That is where the higher median comes from, and where the extra failures come from.
- It does not buy protection against the total-return alternative. The bucket arm ran out more often in every one of the 36 cells; the paths it alone rescued numbered 3 to 33 per cell against 61 to 620 the other way.
- Reserve size is the lever, not the refill rule. Two, three and five years of cash move the success gap from −0.8 pp to −3.7 pp in the anchor plan; the two refill rules never differ by more than 1.4 pp.
- The value of a bucket may be behavioural. These runs assume the total-return retiree rebalances into every downturn without flinching. If a visible cash reserve is what lets a person keep the plan through a bear market, that benefit is real and is not in the numbers here.
- A spending rule does more than a sourcing rule. Both arms here spend the same constant-real amount; the site’s withdrawal-strategy study shows what changing the amount does, which is a different order of effect.
CC-BY-4.0 — free for any use including republication and journalism, with attribution to QuantCalc Research.
See both rules for your own numbers
The free Monte Carlo planner runs the same engine, the same forecast source and the same bucket rule on your balance, allocation and spending — no signup. Open it with this study’s plan, then switch the Withdrawal Rule from the planner’s default to Bucket (cash reserve) and change the reserve in years.
Open the free planner →Methodology
Engine and plan. Every number comes from QuantCalc’s C Monte Carlo engine through its public simulate contract, running the bucket withdrawal rule and the constant-real rule under the JPMorgan LTCMA 2026 capital-market assumptions (as of 2025-11-01; five asset classes, expected nominal returns US Equity 6.7%, International Equity 7.4%, Bonds 4.8%, Real Estate 6.5%, Cash 3.2%) with 2.5% inflation. The retiree is 65 with $1,000,000, no further contributions, no Social Security or pension, and spends $35,000 / $40,000 / $45,000 a year in today’s dollars, raised with inflation, to 95. Plan mixes are 45/15/40 US equity / international / bonds and 60/20/20 US equity / international / bonds. Cash reserves of 2, 3, 5 years are crossed with both refill rules: 36 bucket arms and 18 total-return baselines.
Baseline construction. For a reserve of N years and spending S, share₀ = N × S ÷ $1,000,000. The total-return arm’s allocation is the plan mix × (1 − share₀) with share₀ added to the cash asset, rebalanced annually — the construction documented alongside the engine’s bucket rule. For 3 years at $40,000: share₀ = 12.0%, baseline US Equity 39.60%, International Equity 13.20%, Bonds 35.20%, Real Estate 0.00%, Cash 12.00%. The bucket arm carves the same 12.0% into its reserve at the first retirement month, so both arms hold identical cash at the start.
Paths and pairing. 10,000 quasi-Monte Carlo paths per arm from the engine’s Sobol sequence. The contract returns the yearly values of at most 30 paths per call, so each arm is run as 334 calls that tile the engine’s 10,000-row Sobol cache by seed (seeds 0, 30, …, 9990), which reproduces exactly the path set a single 10,000-path call uses; the two arms of a pair therefore see identical market draws and every difference is taken path by path. Reported values are the engine’s real (deflated) year-end balances; a path is a success if it never reaches $0 before 95, and a ruined path’s terminal value is $0.
Verification before publication. (1) With a 100%-cash plan mix the two arms are the same portfolio: their per-path yearly values agreed to a maximum relative gap of 5.3e-10. (2) A re-issued batch reproduced its 30 paths bit for bit. (3) In every batch of every arm, the count of paths ending above $0 equalled the engine’s own survivor count. (4) For all 54 arms, the success rate and the 10th/50th/90th-percentile terminal wealth of the batched paths matched a single 10,000-path call to within one cent. (5) Every baseline’s cash weight equals its cell’s share₀ and its weights sum to one. At render time the generator re-asserts these records, that each paired band brackets its median, that the higher / lower / tied shares sum to one, and that the survival crossovers reconcile with the two success counts.
Reproducibility. The generator’s run mode drives a local build of the engine and writes the raw results to a committed snapshot; its render mode writes this page, the CSV and the JSON from that snapshot in one pass, so the text and the data cannot disagree.
Assumptions and limitations
- Two buckets, not three. The engine’s rule is the classic bucket strategy reduced to a cash reserve plus one growth portfolio at the plan mix. A bond “middle bucket” is not modelled separately; bonds sit inside the growth mix in both arms.
- The reserve is sized on gross spending and refilled to that target; there is no Social Security or pension in these runs, so gross and net spending coincide.
- Cash earns the cash asset. The reserve earns the roster’s cash return (3.2% nominal, 1.0% volatility under the JPMorgan LTCMA 2026 assumptions), which is what the total-return baseline’s cash earns too. A retiree using a higher-yielding but riskier “cash” would get a different drag.
- Same spending in both arms. Both arms withdraw the same constant-real amount every month; no spending rule reacts to the portfolio. The comparison isolates where withdrawals come from, not how much is withdrawn.
- Parametric returns. Returns are lognormal with the stated means, volatilities and correlations; no regime switching, fat tails or historical block resampling. Deeper or longer bear markets than the model draws would give the reserve more to do.
- The cash share’s later path is not in the dataset. The public contract reports each path’s total wealth, not the reserve’s balance, so the starting cash share is exact and its evolution is described qualitatively; the wealth gap it causes is reported year by year instead.
- Pre-tax, no fees. No taxes, account types, fund expenses or trading costs on either side.
- Behaviour is not modelled. The total-return arm sells into every downturn as the rule requires; any value a visible reserve has in helping a person stay the course is outside these numbers.
- Not advice. Educational research on two ways of sourcing retirement withdrawals; not financial, tax or legal advice.
Frequently asked questions
Does a bucket strategy protect a retirement portfolio better than a total-return approach?
No, not once both approaches start with the same cash. For a 65-year-old with $1,000,000 spending $40,000 a year on a 60/40 plan, a 3-year cash reserve refilled after non-losing years survived to 95 in 85.2% of 10,000 simulated paths against 87.0% for a total-return portfolio holding the same 12.0% in cash and rebalanced every year. Path by path, the bucket arm ended $10,705 more at the median and finished higher in 57.4% of paths, but it ran out on 198 paths the total-return portfolio survived and survived only 21 the other way. Across all 36 combinations of spending, mix, reserve size and refill rule the bucket's success rate was lower in 36 (by -0.5 pp to -5.9 pp) and never higher, while its paired median ran from $0 to +$28,803: a better middle, a worse tail.
What does the cash bucket actually buy?
A different source for withdrawals after a bad year, and a cash share that drifts instead of being rebalanced. In the 3-year, $40,000, 60/40 case the bucket arm reached 95 with money left in 21 paths where the matched total-return portfolio ran out, while the reverse happened in 198 paths. Summed over all 36 cells the counts are 609 paths rescued by the bucket against 8,143 lost to it, out of 360,000 paired comparisons. Because the reserve is a set number of years of spending, a growing portfolio holds proportionally less cash and a shrinking one proportionally more; that is why the bucket ends higher at the median in 26 of 36 cells and runs out more often in all 36.
Why compare against a total-return portfolio that also holds cash?
Because a bucket portfolio holds cash the plan mix does not, and the comparison is only fair if the total-return side starts with the same amount. A 3-year reserve on $40,000 of spending is $120,000 of a $1,000,000 portfolio, 12.0%, so the total-return baseline holds the 60/40 mix scaled to 88.0% plus 12.0% cash and is rebalanced back to that mix every year. The two arms then differ only in where each withdrawal comes from and in how the cash share drifts. Comparing a bucket portfolio with a cash-free 60/40 portfolio would confound the cash allocation with the withdrawal rule.
Does the refill rule matter?
Less than the size of the reserve. Refilling only after a non-losing year and refilling whenever the reserve drops below half its target produced success rates that never differed by more than 1.4 pp in any cell. At the median the below-half rule ended higher in 11 of the 18 spending / mix / reserve combinations (the non-losing-year rule in 0), by at most $29,716 for a 5-year reserve, because letting the reserve drain to half means holding less cash on average. For the 3-year, $40,000, 60/40 case the two rules reached 95 with money left in 85.2% and 85.0% of paths respectively.
How big should the cash bucket be?
In these runs a larger reserve erodes the median advantage and widens the extra chance of running out. At $40,000 on 60/40 with refills after non-losing years, the paired median difference against the matched total-return portfolio was +$21,051 for a 2-year reserve, +$10,705 for 3 years and $0 for 5 years, with success rates of 86.2%, 85.2% and 83.1% against 87.0%, 87.0% and 86.8% for total-return portfolios holding the same starting cash. Each extra year of reserve is another year of spending whose share of the portfolio rises exactly when the portfolio is falling.
Are these numbers real dollars, and are taxes included?
All dollar figures are in today's dollars: the engine raises spending with 2.5% inflation and deflates every reported balance back to the retirement date. Nothing is after tax, there is no Social Security or pension income, and both arms spend the same constant-real amount every year; the study isolates the withdrawal-source rule and nothing else. Returns follow the JPMorgan LTCMA 2026 capital-market assumptions the site's planner uses, sampled with quasi-Monte Carlo, 10,000 paths per arm, identical draws for the two arms of every pair.
Related research
Changelog
- v2026.1 (2026-09-07) — initial release. 36 bucket cells (3.5% / 4.0% / 4.5% spending, 60/40 and 80/20, 2/3/5-year reserves, two refill rules) against cash-matched total-return baselines; 10,000 paired paths per arm; per-path paired statistics at 95 and by year.
Last updated 2026-09-07. Dataset license: CC-BY-4.0. QuantCalc is an independent retirement-planning research project. The JPMorgan LTCMA 2026 name identifies the published capital-market assumptions the engine uses; QuantCalc is not affiliated with, endorsed by, or sponsored by that firm, and all trademarks belong to their respective owners. Educational research, not financial, tax, or legal advice.
Cite this research study
QuantCalc Research (2026). Bucket Strategy vs Total Return: Does a Cash Bucket Protect a Retirement Portfolio? (2026). https://quantcalc.app/research/bucket-strategy-vs-total-return-2026/ (accessed <date>).
BibTeX
@misc{quantcalc2026bucketstrategyvstotalreturndoesacashbuck,
title = {Bucket Strategy vs Total Return: Does a Cash Bucket Protect a Retirement Portfolio? (2026)},
author = {{QuantCalc Research}},
year = {2026},
url = {https://quantcalc.app/research/bucket-strategy-vs-total-return-2026/},
note = {Accessed <date>}
}
Machine-readable citation metadata (schema.org identifier and citation fields) is embedded in this page's JSON-LD, at the stable identifier https://quantcalc.app/research/bucket-strategy-vs-total-return-2026/.