Rebalancing vs Letting It Drift

Rebalancing is usually sold as a free bonus. On the long record it loses to doing nothing — because doing nothing quietly changes what you own.

Worked example — Rebalancing vs Letting It Drift

1.11% a year ahead	99.6%
Rebalanced	8.34%
Never touched	9.45%
Rebalanced volatility	12.06%
Untouched volatility	15.70%

These are the values the calculator shows when the page opens. Change any input above to recompute them.

Loading interactive simulation...

A bonus that only appears when neither asset is winning 🖖

The rebalancing bonus is real but conditional, and the condition is that the sleeves have similar long-run returns and move independently enough to be worth trading between. Split US large-cap against small-cap over the whole record and rebalancing gains 0.24 points a year; run a 60/40 through 2000 to 2012, when bonds beat stocks, and it gains 0.44. Run the same 60/40 across the whole century, when equities beat bonds decade after decade, and it loses 1.11 — because every year it sold the asset that was going to keep winning. No strategy that trims its best holding can beat a permanent trend, and no strategy that holds its risk steady is trying to.

Buy and hold names a rule, not a portfolio 🖖

The drift curve is the answer to what rebalancing is for. Left alone from 1928, a 60/40 ends the window 99.6% in equities — so the thing it beat by 1.11 points a year was not a 60/40 at all, it was an all-equity portfolio that got there by neglect. Switch to Large vs small caps, an even split of the money, and the drift runs the other way: untouched, the portfolio ends 15.0% in large caps, swallowed by the small-cap sleeve. On the lost decade it ends 43.8% in equities, because bonds won those years. One instruction, three destinations, each chosen by whichever sleeve happened to win.

Rebalancing does not always calm a portfolio down 🖖

On the long record it does: from 1928 the rebalanced 60/40 runs at 12.06% annual volatility against the drifting portfolio’s 15.70%. Click 60/40, 2000-2012 and it inverts — rebalanced 8.49%, drifting 6.07% — and the rebalanced portfolio earns more anyway, 4.57% against 4.13%. Drifting into bonds made the untouched portfolio calmer by accident, in the thirteen years when that happened to be the right place to drift. What rebalancing does reliably is hold the risk where you put it; whether that is also the low-risk answer depends on what the drift would have done instead.

Problem solved in full

60% US stocks and 40% 10-year Treasuries rebalanced every year 7 steps
60% US stocks and 40% 10-year Treasuries, 1928–2025, rebalanced every year against bought once and never touched. Work the untouched line out from two numbers — then show why the rebalanced one cannot be worked out at all.
1. Start with each sleeve on its own. Set First asset (%) to 100 and both return cards read 10.02%; set it to 0 and both read 4.53%. They have to agree at those two settings, because with one asset there is nothing to sell back to target — so this is a check on the tool as much as a reading off it.
  
  $g_{s} = 10.02%, g_{b} = 4.53%, n = 98$
2. Ninety-eight years turns a gap in rate into a gap in money. The gap is 5.49 points a year, and compounded that far it becomes a factor of about 150 — the number the rest of the problem is really about.
  
  $1.100 2^{98} = 11594, 1.045 3^{98} = 76.84, \frac{11 594}{76.84} = 150.9$
3. Buy and hold means nothing is ever bought or sold, so each sleeve simply compounds by itself. Track the 60 cents and the 40 cents separately and add them up at the end.
  
  $V_{H} = 0.60 \times 11594 + 0.40 \times 76.84 = 6956 + 30.74 = 6987$
4. Turn the total back into a rate and it is the Never touched card. Notice what never entered: the order of the years. Reshuffle each sleeve’s 98 returns into any order you like and the answer is unchanged — but only within a sleeve. Pool all 196 numbers and re-deal them into two fresh sleeves and it is a different portfolio.
  
  $g_{H} = 698 7^{1/98} - 1 = 9.45%$
5. Now ask what that portfolio is made of when it finishes, which is the figure the verdict prints. The line that won was not a 60/40 left alone. It was an all-equity portfolio that used to be a 60/40.
  
  $\frac{6 956}{6 987} = 99.6%$
6. Rebalancing cannot be done from two numbers at all. Each year’s return is the blend 0.6r_s + 0.4r_b, and a product of 98 blends is not any combination of the two sleeve totals — swap one year’s stock return for that same year’s bond return and the answer moves. Order is no more relevant here than it was above — a product of 98 factors does not care — but the pairing is: reaching 8.34% needs all 196 numbers matched to the years they actually happened in, which is why the tool computes it and this page cannot.
  
  $g_{R} = (t = 1 \prod 98 (1 + 0.6 r_{s, t} + 0.4 r_{b, t}))^{1/98} - 1 = 8.34%$
7. So the 1.11-point gap is not a penalty for rebalancing. It is the difference between a portfolio that ended 99.6% in equities and one held at 60% for 98 years, and the volatility cards price that difference from the other side: the untouched line carries 30% more year-to-year variation.
  
  $g_{H} - g_{R} = 9.45% - 8.34% = 1.11%, \frac{15.70%}{12.06%} = 1.30$
Answer

9.45% a year untouched against 8.34% rebalanced — and the untouched line ends 99.6% in equities. Drift on that scale is not a quirk of 60/40. Ask which starting weight would still leave the portfolio half in bonds after 98 years: w × 150.1 = 1 − w, so w = 1/151.1 = 0.66%. Any buy-and-hold portfolio beginning with more than two-thirds of one percent in stocks ends up majority stocks. Type 0.66 into First asset (%) and the verdict reads 49.9%, the threshold arriving on cue. Which means “never touched” is not a strategy with a risk level. It is a strategy that hands its risk level to whichever sleeve wins, and tells you which one 98 years later.

$w^{*} = \frac{1}{1 + 150.9} = 0.66%$

Learning path

Risk, measured

Leads to How sure can you be the warning that two strategies ending at different risk levels cannot be ranked by return alone.

References (2)

Where the "rebalancing bonus" comes from, and the arithmetic that says when it is there at all: S. Willenbrock, "Diversification Return, Portfolio Rebalancing, and the Commodity Return Puzzle." Financial Analysts Journal 67(4), 42–49, 2011.
Portfolio variance as wᵀΣw, the quantity a rebalanced portfolio holds still and a drifting one does not: H. Markowitz, "Portfolio Selection." The Journal of Finance 7(1), 77, 1952.

By the 42Math team · Updated 2026-08-19

Reviewed against published sources. Corrections welcome. Editorial policy · Corrections

All Statistics & Finance lessons →

Example problems

60/40, whole record - A 60/40 across 1928-2025, the case where leaving it alone wins on return and ends up almost entirely in equities.
60/40, 2000-2012 - 2000-2012, a window where bonds beat stocks and annual rebalancing comes out ahead.
Large vs small caps - US large-cap against small-cap: similar assets with similar long-run returns, which is the setting the rebalancing bonus needs.
Stocks and gold - Stocks and gold from 1972, two assets that move very differently.