Selling What Went Up to Buy What Went Down

Why a Portfolio Drifts

An investor chooses a mix of targets say a certain share in stocks and the rest in bonds sized to match the risk he can bear. That mix doesn't stand still. Stocks and bonds grow at different rates and the one that grows faster starts to eat a bigger slice of the pie. If you let stocks run ahead of bonds for a few years stocks will outperform their target. No one decided to take more risk. The portfolio just went into it

Rebalancing It is the practice of periodically returning a portfolio to its target mix: selling what has grown beyond the target and buying what has fallen below it. It sounds like housekeeping. In reality it is one of the few investment disciplines that makes a reliable living

The Risk Control You're Actually Paying For

The main task of rebalancing is to control risk not increase returns. That order is important and is almost the opposite of what most people think when entering

Imagine an investor who wanted a balanced portfolio half stocks and half bonds and then experienced a multi-year stock market rally. Check the account today and stocks could make up two-thirds of it. They are taking on significantly more risk than they had committed to and if stocks fall sharply they lose more than budgeted. Rebalancing toward the target resets the level of risk they actually chose. That's the entire justification and it would still be valid even ifrebalancing will add zero additional return

Two Ways to Trigger a Rebalance

There are two honest ways to decide when to rebalance and textbooks argue about which is better more than real-world evidence probably warrants

Calendar rebalancing sets a fixed schedule once a year or once a quarter and the portfolio is rebalanced on that date no matter what it looks like when the date arrives. It's simple requires no value judgments and survives solely on a recurring reminder

Threshold rebalancing sometimes called tolerance band rebalancing ignores the calendar and instead looks at drift. Pick a band say five percentage points from target and rebalance only when an asset crosses that band whenever that happens

The compensation is not subtle once you accept it. Calendar rebalancing can be triggered on a date when the portfolio has barely drifted paying transaction costs and perhaps taxes to fix a problem that was never big enough to matter. It can also miss a large deviation that occurs just after the scheduled date and then remains unbalanced for months. Threshold rebalancing is only operated when the drift is large enough to be worth correcting but requires reallooking at your wallet or setting something up to keep an eye on it rather than just marking a date on a calendar

Academic comparisons of the two methods generally find that the difference in results is small less than the amount of debate the choice generates. What matters much more is choosing one and sticking with it rather than making a new emotional decision every time the market moves

What Actually Decides Which One Wins: Cost

If the two methods work similarly before costs costs are the tiebreakers in practice and this is the part that a simple case for rebalancing tends to miss

Every rebalancing trade in a taxable account is a taxable event if it involves selling something that has gained value. Sell the winners to fund the purchase of the laggards and the profit will be made right there even if the money never left the market. That tax bill is real money leaving the portfolio forever not a paper cost

Transaction costs also matter although they are lower than before. Bid-ask spreads and in the case of less liquid holdings the impact of the trade on the market itself reduce the profit. A rebalancing that reduces a one percent drift in a thinly traded fund may cost more in friction than it saves in risk control

This is why the threshold is not just a question of time but also of cost. A strict threshold which rebalances with each deviation of two percentage points generates more transactions more tax events and more cumulative costs for a risk control benefit that is only marginally better than a looser band of five or ten points. Widening the band is often the most valuable change an investor can make to a rebalancing policy because it directly reduces the number of taxable events.without giving up much risk control

Three practical steps reduce this cost without abandoning discipline. First rebalance tax-advantaged accounts an IRA or 401k where a sale triggers no current tax bill. Direct new contributions toward whatever is underweight rather than selling the winner which rebalances without triggering any sales. Use dividends and interest which appear as cash anyway to buy out the laggards instead of automatically reinvesting in what you paid them. Do these three things and oneportfolio can remain close to target for years with very few actual taxable sales

The Rebalancing Bonus Is Not Free Money

Sometimes people describe rebalancing as generating a bonus an additional return on top of what an investor would have earned by keeping the original combination intact. I want to be careful with that word because it makes the effect sound like a law of the markets rather than what it really is: a bet on how prices behave over the rebalancing horizon

Here's the mechanism. If the price of an asset bounces around a level rather than steadily moving away from it systematically trimming it after it rises and raising it after it falls captures a little bit of each swing. That's selling something high and buying something low over and over again with each back and forth generating a small profit relative to simply standing still. That effect is real and well documented over long periods of market history

But let's look at what the mechanism depends on: prices have to retrace at least partially during the period it is rebalancing. If on the other hand an asset is in a genuine multi-year uptrend - not a rally that then gives most of it back but a real and persistent outperformance - then trimming it every year to fund the laggards is not reaping a swing. It's trimming the winner to feed what it has been losing and the bond quietly becomes aballast.No one discovers what regime they were on until long after the fact

This is the part I think is left out when rebalancing is presented as a free lunch. It's not free and it's not guaranteed. It's a disciplined bet that over long periods and in a mix of genuinely different asset classes returns tend to reverse more than they tend to. That bet has a good long-term track record. It's still a bet

A Worked Example: Rebalancing a 60/40 Portfolio

Let's say an investor starts with $100,000 split 60/40 between stocks and bonds a common illustrative goal and each figure below is chosen just to make the math easy to verify. This equals $60,000 in stocks and $40,000 in bonds

Now let's say stocks have a good run and are up 50 percent while bonds are up 5 percent both examples. Stocks: 60,000 times 1.50 is $90,000. Bonds: 40,000 times 1.05 is $42,000. The portfolio is now worth 90,000 plus 42,000 or 132,000.dollars

Recalculate the weights of that new total. The stocks are 90,000 divided by 132,000 which is about 68.2 percent. The bonds are 42,000 divided by 132,000 about 31.8 percent. Check that those two add up to 100: 68.2 plus 31.8 is 100.0 so thepesos are internally consistent. Shareholding has fallen about 8.2 percentage points above its target of 60 percent. If the investor's threshold rule is set at 5 points the investor exceeds it and causes a rebalancing

To get back to 60/40 on the new total of 132,000 the target dollar amounts are 60 percent of that or 79,200 in stocks and 40 percent or 52,800 in bonds. Check: 79,200 plus 52,800 is 132,000 which matches the total exactly

The trade needed is the gap between where the portfolio is and where it should be. Stocks need to go from 90,000 to 79,200 a $10,800 sell. Bonds need to go from 42,000 to 52,800 a $10,800 buy. The put and buy are the same size because the money coming out of the stock is exactly what finances the bond purchase. That's all.the business

Now the cost. Let's say this is in a taxable account and the original $60,000 invested is the total cost basis with nothing bought or sold in between. The stock position grew from 60,000 to 90,000 so 30,000 of that $90,000 is profit which is one-third of the position. Selling $10,800 worth of stock generates a profit in that same proportionof one-third: 10,800 divided by three is $3,600 taxable gain. Using an illustrative long-term capital gains rate of 15 percent the tax owed is 3,600 times 0.15 which equals $540

Put that cost into context. $540 represents 5 percent of the $10,800 trade and about 0.4 percent of the total $132,000 portfolio. That's the real price of restoring the target mix here: a few tenths of a percent of the portfolio paid once to undo an 8-point drift in risk. Instead run the same trade inside an IRA and that $540 itemIt will disappear completely since there is no tax on a sale within a protected account. That one difference taxable versus protected is often the biggest leverage an investor has on the cost of staying disciplined

Case Study: Norway's Sovereign Wealth Fund

For a real-world example of threshold rebalancing on a massive scale look at Norway's Global Government Pension Fund the sovereign fund created from the country's oil and gas revenues and managed by Norges Bank Investment Management on behalf of Norway's Ministry of Finance. It is by most measures the largest single fund of its kind in the world at more than $1 trillion

Norway's Ministry of Finance sets a strategic benchmark a mix of targets heavily weighted toward global equities and the rest divided between fixed income and a smaller portfolio of real estate and infrastructure. The manager does not choose that target. His job is to manage the portfolio around him and rebalance it and the fund publishes the mechanics of that rebalancing as an explicit written policy rather than leaving it to the discretion of a manager. When the actual equity stake deviates from the strategic benchmark by more than a set number of pointspercentages the fund rebalances toward the target. I wouldn't swear what the exact trigger is as policy has been adjusted over the years but the mechanism is exactly the threshold approach this article describes simply executed on a portfolio the size of the entire economy of a medium-sized country

What makes the case study interesting is what that rule forces during a crash. When global stocks fell sharply during the 2008 financial crisis a fund heavily weighted in stocks relative to its benchmark heavily affected by falling prices would have found itself underweight stocks relative to its target simply because of the drop.So I bought stocks financed by cutting the fixed income side. Norway's fund did almost exactly that during the crisis contributing to the stock market's decline because the rule said so not because anyone in the fund felt confident that stocks had bottomed

It's about rebalancing discipline at the least comfortable time it's asked to work and it's also the time when the argument for having a rule rather than a sentiment matters most. Ordinary confidence doesn't buy aggressively in a market that's still falling. A fund that follows a written rebalancing policy does exactly that because the rule was written before the fear hit not during it

Where This Breaks: Trends Instead of Swings

I want to honestly argue for the other side because the previous section already told you where the vulnerability lies: it all depends on prices reversing rather than following a trend and trending is a real thing that markets do

The academic term for this is momentum and it is one of the most replicated findings in market research. Stocks and other assets that have outperformed over the past few months to a year tend on average to continue outperforming over the following months before any potential reversal. This is close to the opposite of the mean reversion that rebalancing quietly assumes over similar horizons. If rebalancing at a quarterly threshold the asset being shorted could be in the midst of a momentum streak.instead of a mean reversal swing. Sell it and the winner who was statistically likely to continue winning for a while longer will be eliminated and the laggard who was statistically likely to continue lagging will be bought

The most acute version of this problem manifests itself in concentrated positions in single stocks rather than diversified asset classes. Someone who had a founder-sized stake in a company on a genuine multi-decade secular path and trimmed it to a fixed target percentage each year was giving up a significant amount of terminal wealth compared to simply holding on because the position was not hovering around a level but rather moving away from one. This is the classic example of why some very successful long-term investors refuse torebalancing a concentrated core position on a mechanical schedule and I think that instinct is defensible for a single high-conviction position although I wouldn't extend it to a diversified portfolio of asset classes

Taxes make the same mode of failure worse. A strict rebalancing threshold on a taxable account applied to volatile holdings can accumulate a tax burden that exceeds any risk control or return benefit the rebalancing was supposed to generate especially for an investor in a high tax bracket who sells short-term gains rather than long-term gains. In that case the discipline technically works and still leaves the investor worse off after taxes than if the drift had been left alone for a while longer

None of this means that rebalancing is a bad idea. It means that these are long-term trends in diversified portfolios not a law that applies to all assets every year regardless of cost

The Behavioral Problem, Not the Analytical One

This is what I think is really difficult about rebalancing: none of it is difficult to calculate. All of this is difficult to do

Selling something that has made you money and buying something that has made you lose money is not a neutral action. It combats the same instinct that causes people to hold losing stocks too long in the hope of matching them again and sell winners too soon because the gain seems fragile. Rebalancing requires the opposite of both impulses at once on purpose on a schedule set before anyone knew which asset would go up and which would go down

I think that's why a lot of the actual advice on rebalancing is actually advice on how to move away from the decision. Automatic contributions that target the underweight asset. A calendar reminder rather than a gut check. A written threshold rather than a vibe about whether you now feel good. A default that requires a deliberate override rather than deliberate action to rebalance at all. None of them are analytically sophisticated. They exist because the analytically simple version of rebalancingIt's emotionally one of the hardest things to execute especially in the middle of a sell-off when buying what's falling feels in the moment like catching a falling object rather than following a rule

How I Actually Use This

My own approach is closer to the threshold than the calendar mainly because I don't trust myself to sit in a five-point drift for eleven months just because the date on the calendar hasn't arrived yet

The way I actually use this: I check several times a year not constantly and I only act when something has strayed about five percentage points or more from target. Below that I don't do anything on purpose because I'd rather outsource a small drift than pile on tax events for a risk shift that barely exists. When I rebalance I try to do it with new money first directing what I'm adding that month toward whatever is underweight before considering a sale thatwould generate a tax bill. Selling a winner to finance a purchase is the last tool I use not the first

I admit that I find it really difficult to accept the question of trend versus mean reversion more difficult than I expected when I first read about rebalancing bonuses as if they were an established fact. I don't know in real time whether a given rally is a swing that will partially reverse or the beginning of something that will continue for years. No one does. What I have decided is to treat rebalancing as a policy for the diversified core of a portfolio where I feel comfortable betting onlong-term reversal across entire asset classes and being much more reluctant to apply the same mechanical adjustment to a single concentrated position held for a specific high-conviction reason. That's a distinction I didn't make the first time I seriously thought about this and I think my previous thinking was the worse for omitting it

The honest summary of my own use of this is not glamorous. Set a wider threshold than seems satisfactory prefer new money over sales keep it inside tax-sheltered accounts whenever the account type allows and treat the discipline as more valuable for what it prevents me from doing - chasing what you just raised - than for any bonus returns I'm sure it will deliver

The Bottom Line

Rebalancing returns a portfolio to its target mix by selling what has surpassed the target and buying what has fallen below it and its first task is to control risk not increase returns. Calendar and threshold rebalance both work and the real difference between them shows up in the cost: taxes on realized gains and transaction friction which is why widening a threshold and funneling new contributions to the underweight asset before selling something is usually better than a strict trigger ruleeasy.The so-called rebalancing bonus is real over long periods but it is a bet that prices will reverse rather than follow a trend over the rebalancing horizon not a guaranteed law and momentum concentrated winners and tax burden are all real ways the bet can lose. What makes discipline difficult is not arithmetic. You are doing as planned the one thing that always seems wrong to you: selling what went up to buy what went down