Week 11: Behavioral Biases and Rebalancing — Rules That Protect You From Yourself

---

1. Why This Is Important

Every portfolio in this course is two things at once: the strategy written down in the spreadsheet, and the behaviour of the human who has to hold it through a 35% drawdown, a screaming headline, and a brother-in-law who just tripled his money in a meme stock. Most of investing's documented underperformance does not come from picking the wrong fund. It comes from the gap between what the spreadsheet recommends and what the human actually does.

You need to understand behavioural biases for four reasons.

The Dalbar gap is real, and it is measured in your dollars.

The DALBAR QAIB studies have followed actual cash flows in and out of US equity mutual funds for thirty years. The headline finding, replicated every year, is that the average equity-fund investor earns roughly 2 to 4 percentage points per year less than the fund itself. The fund didn't underperform. The investor did, by buying after rallies and selling after drawdowns. Compounded over thirty years, that gap eats more than half the terminal wealth. The strategy was fine. The hand on the keyboard was not.

Markets stay irrational longer than retail can stay liquid.

The classic Keynes warning: even when you are correctly identifying a mispricing, the price can move further against you for longer than you can finance the position. Behavioural biases are the mechanism that creates those long irrational stretches — it is the herd's loss aversion and FOMO that drives a bubble for an extra two years past where the spreadsheet said it should have ended. If you don't understand herd behaviour, you don't understand why the trade you got right on Day 1 still blew you up by Day 500.

**Fat tails punish leverage, and leverage is what overconfidence

builds.** The volatility tail wags the portfolio dog. Overconfident investors take more leverage than their conviction warrants, because the bell-curve model in their head puts the next −30% year at "once in a century." It isn't. Pre-2008 leverage was a textbook overconfidence-meets-fat-tail pairing, and individual investors are running the same trade today, just with smaller decimal points.

Knowing about a bias does not eliminate it. This is the most

important paragraph in the lesson. Reading about loss aversion does not make you immune to loss aversion. Even Kahneman, who won the Nobel Prize for naming it, said in interviews that he still feels the asymmetry as strongly as he did at twenty. The bias is wired in. The only durable fix is system design — automate the contributions, automate the rebalance, *remove the discretion*, and accept that on the days you most want to override the system, the system is most right and you are most wrong.

This lesson covers loss aversion, recency bias, anchoring, herding and FOMO, the disposition effect, narrative fallacy, and overconfidence — the seven behaviours that, between them, account for nearly all of the Dalbar gap. Then it covers the single most powerful counter-system to those biases: rebalancing. Rebalancing is usually taught as portfolio mechanics. It is more honestly understood as a behavioural rule — a pre-committed instruction that forces you to do the right trade at the moment your wired-in biases would have you do the opposite. We close on the broader set of system designs (automated contributions, written policy statements, position-size limits) that work because they take the decision out of your hands.

---

2. What You Need to Know

2.1 The Dalbar Gap — How Much Behaviour Costs

The DALBAR Quantitative Analysis of Investor Behavior (QAIB) study compares two return series: the time-weighted return of the average US equity mutual fund (what the fund actually delivered) and the dollar-weighted return of the average investor in those funds (what the investor actually earned, given the timing of their contributions and redemptions). The gap between them is pure behaviour.

The gap is not random fees or fund-selection. Two-thirds of it is sequence-of-flows: investors put more money in after the market has run, and they pull money out after it has fallen. The remaining third is fund-shifting (jumping from the year's loser to the year's winner) and panic exits. Compounded over a thirty-year career at 3.5% per year, the behavioural drag turns a 10x wealth multiple into roughly a 4x one. **The single most expensive financial mistake the typical retail investor makes is not which fund they pick — it is when they buy and sell.**

2.2 Loss Aversion — The 2.5x Asymmetry

The foundational bias, named by Daniel Kahneman and Amos Tversky in their 1979 prospect-theory paper. The pain of losing $100 is roughly 2 to 2.5x as intense as the pleasure of gaining $100. Their value function looks like this:

Three consequences for portfolio behaviour:

• Holding losers, selling winners. Closing a losing position

realises the loss — converts a paper pain into a confirmed one — and the asymmetry makes that conversion feel intolerable. So you hold. Closing a winning position realises the gain, and that conversion feels good, so you do it too quickly. The combined effect is a portfolio that systematically over-weights the names that have already proved they were the wrong call. This is the disposition effect, and it is the single most documented behavioural pattern in retail brokerage records.

• Aversion to drawdowns at the strategy level. A 60/40 portfolio

that loses 22% in one year (2022) feels worse than twice as bad as one that loses 11% in two consecutive years (the actual cumulative is similar). The temporal compression of the loss amplifies the pain, even when the dollars are identical. This is why one bad year changes long-term allocations far more than a slow grinding decade does.

• The wrong volatility check. Loss aversion makes investors

check their portfolio more often during drawdowns than during rallies. The more you check during a drawdown, the more the asymmetric pain accumulates, the more likely you are to capitulate at the bottom. The brokerage app's red number is a behavioural weapon pointed at your own wealth.

2.3 Recency Bias — The Last Year Predicts Forever

Recency bias is the tendency to extrapolate the most recent experience as the new permanent state. When stocks have just returned 25% for two years running, your brain quietly resets the expected-return prior to "stocks return 25%." When they've been flat for three years, the prior resets to "the era of stock returns is over." The actual long-run distribution does neither.

This is why the worst time to raise your equity allocation is after a great equity decade — the regime that produced those returns is already priced into current valuations, and your forward expected return is lower, not higher. And the worst time to cut equity is after a 30% drawdown — the forward expected return after the drawdown is higher, because prices fell faster than fundamentals. Recency bias has the wrong sign in both cases. Every time.

The 1999 retail allocation to internet stocks was peak recency bias. The 2009 retail flight to cash was peak recency bias. The 2021 retail allocation to crypto and meme stocks was peak recency bias. The pattern repeats because the wiring repeats.

2.4 Anchoring — Why Your Cost Basis Should Not Exist on Your Screen

Anchoring is the brain's tendency to over-weight the first number it encountered as a reference point. In investing, the anchor is almost always your cost basis — the price you paid. Cost basis is relevant for tax purposes. It is irrelevant for *should I hold this* purposes.

The honest framework is the inheritance test: if you inherited this position today at the current market price, with no cost basis attached, would you choose to hold it? If yes, the anchor is noise. If no, you are only holding because of the anchor — and the anchor is a bookkeeping artefact, not a piece of investment thesis.

Other anchors that infect retail portfolios: the 52-week high ("it was 30% higher last year, so it must come back"); the IPO price ("it must be worth at least the IPO price"); a round number ($100, $1,000); a friend's cost basis. None of these have any information content about the current fair value. All of them distort your hold/sell decision.

2.5 Herding and FOMO — The Beauty Contest

Keynes' beauty-contest analogy is the cleanest description of herding ever written: the investing game is not picking the prettiest face, it is picking the face *the average voter* will pick — and the average voter is in turn picking what they expect the average voter to pick, all the way down. The correctness of the underlying valuation is many layers removed from the proximate driver of price.

Herding has two flavours and they reinforce each other:

• FOMO (fear of missing out). Watching neighbours, coworkers,

and Twitter accounts get rich on a trade you didn't take. The behavioural pressure builds asymmetrically: a 50% rally in a position you don't own hurts more than a 50% rally in a position you do. The rational response is to ignore other people's outcomes; the behavioural response is to chase, often near the top, for fear of being the only one left out.

• Capitulation. The mirror image: when everyone else is

selling, the social proof of mass exit overrides any analytical view that the sell-off is overdone. The 2008 March bottom and the 2020 March bottom both saw record retail redemptions; both were near-perfect bottoms. The herd was wrong, the herd was loud, and following the herd cost you the recovery.

The reason herding behaviour creates the long irrational stretches that bankrupt good traders is exactly the beauty-contest mechanism: even correct contrarians can be early by years, and being right on Day 1 doesn't help if you can't finance the position to Day 500.

2.6 Narrative Fallacy and Overconfidence

Two related biases that compound each other.

Narrative fallacy — the brain's preference for a clean causal story over a noisy statistical one. Every 1% market move on the news is paired with a confidently-stated reason. *"Stocks rose on strong earnings." "Stocks fell on rate fears."* These narratives are post-hoc fits to noise. The actual driver of any given day's move is rarely a single clean reason; usually it is the chaotic sum of millions of independent decisions, some on data, some on flows, some on liquidity, most on nothing in particular. Treating the day's narrative as causally informative builds a false picture of how markets work.

Overconfidence — the tendency to over-estimate one's own prediction precision. The classic finding: when investors say they are 90% sure of a forecast, they are right 70% of the time. Combined with narrative fallacy, this builds a portfolio of high conviction in stories that are mostly noise. Two specific manifestations:

Excessive trading. Barber and Odean's work on retail

brokerage records found that the highest-turnover quintile underperformed the lowest by roughly 6 percentage points per year. Every trade is an act of confidence; over-confidence manifests as over-trading.

Under-diversification. "I know what I own" becomes a license

to hold ten names instead of five hundred. The ten names carry the same expected return as the index but vastly more idiosyncratic risk — the textbook free-lunch anti-portfolio.

2.7 Why Knowing Doesn't Cure — System Design Beats Willpower

This is the part of the lesson that survives the rest. The biases are not solved by being smarter. They are solved by *taking the decision out of your hands at the moments your hands are worst*.

Concrete system designs:

• Automatic monthly contributions that bypass the brain. The

decision to invest was made once, when you set up the transfer. The behavioural decision not to invest after a 30% drawdown never gets made because there is no decision to make.

• Rules-based rebalancing on a fixed calendar (annually or

semi-annually). Drift back to target. No view-based timing. The rebalance trade mechanically buys the loser and sells the winner — the opposite of the recency-bias trade you'd make by hand.

• Long horizons announced in advance. "I will not look at this

account for thirty years" is a behavioural commitment device, not an investment thesis. The strategy is the same; the behavioural drag falls because you have removed the surveillance that activates loss aversion.

• Position-sizing rules instead of conviction-sizing. A

maximum-2%-per-name rule, applied mechanically, removes the overconfidence channel by which "I really like this one" turns into a 30% portfolio bet that blows up.

• A written investment policy statement that pre-commits how

you will respond to common scenarios (a 20% drawdown, a 50% rally, a job loss, an inheritance). Pre-committed responses beat in-the-moment responses because the in-the-moment brain is the System-1 brain Kahneman warned about.

The interactive panel at the bottom of this lesson lets you toggle on and off four classic bias-driven actions ("sell after a 20% drawdown", "buy after a 20% rally", "switch to bonds after two down years", "chase last year's winner") and watch what they do to a 1928–2024 backtest against simple buy-and-hold. The pattern in every combination is the same: the bias-driven version underperforms, by roughly the Dalbar-gap-size or worse, in nearly every parameter combination. Behaviour is the most expensive line item in the portfolio.

2.8 Rebalancing as the Anti-Bias Rule — Drift, and Why a Portfolio Will Not Stay Where You Put It

Rebalancing is usually taught as portfolio mechanics — set a target allocation, trim what has drifted up, top up what has drifted down. It is genuinely all of those things. But the reason it survives as a discipline, after seventy years of academic and practitioner debate, is behavioural. Rebalancing is the cleanest implementation of "buy low, sell high" that exists, and it is the only one that works without requiring the human holding the keyboard to override loss aversion, recency bias, and FOMO at the exact moments those biases are loudest.

Start with the mechanical case. Two assets with different returns and different volatilities cannot hold their relative weights for long. Higher-return assets compound their share of the pie; lower-return assets shrink in proportion. Even in a single calm decade the drift is large.

The image below runs the 2010–2019 decade — stocks compounding at roughly 13.6% per year and 10-year Treasuries at roughly 3.7% — on a $100,000 starting 60/40 portfolio under two policies. The top panel holds no rebalances at all over the ten years. The bottom panel rebalances back to 60/40 every January.

The end-state of the no-rebalance portfolio is not 60/40 and never was after the first year. By 2019 it sat at about 78/22 — closer to the textbook "aggressive" allocation than to the "balanced" one the investor originally chose. When the COVID crash arrived in March 2020 that drifted portfolio dropped roughly two percentage points more than the rebalanced version, on a much larger asset base. The investor who never rebalanced did not consciously become more aggressive in 2019. They drifted there.

That drift is the behavioural problem before any human ever pushes a button. Even if the investor did nothing — which would seem to be the most behaviourally inert outcome possible — the portfolio itself silently re-priced its risk by roughly fifty percent. The "do nothing" investor ends up holding a portfolio they never agreed to hold, sized for a risk tolerance they never committed to. The next bear market then arrives in a stranger's portfolio.

2.9 Calendar vs Threshold, and Why the Choice Barely Matters

There are two principled ways to decide when to rebalance.

Calendar rebalancing picks a fixed date — typically annual, occasionally semi-annual or quarterly — and trims back to target on that date regardless of how far the portfolio has drifted. The merit is operational: one calendar entry per year, automatable by most brokerages, no surprises in tax planning.

Threshold (or band) rebalancing ignores the calendar and only triggers a rebalance when one asset class has drifted beyond a fixed percentage band — say, ±5 absolute percentage points around the target, so a 60% stock target rebalances at 55% or 65%. The merit is efficiency: in a calm year you do nothing; in a violent year (1987, 2008, 2020) you rebalance multiple times, capturing the mean reversion that calendar rules miss.

The image below compares four policies on the full 1928–2024 Damodaran dataset of US stocks and 10-year Treasuries: never rebalance, rebalance annually, rebalance semi-annually, and a 5% band. The bars show geometric annualised return, realised volatility, and worst-year drawdown for each policy.

Three readings come out of this chart.

First, **rebalancing reduces volatility and drawdowns more than it changes the return.** All three rebalanced policies finish within about ten basis points of each other in geometric return. The story is in the risk columns, not the return column.

Second, **the difference between annual, semi-annual, and 5% band is genuinely small.** Pick one, run it for thirty years, and you will get within a few basis points of any other reasonable rule. This is a place where over-engineering pays nothing.

Third, **never-rebalance is the worst of the four on a risk-adjusted basis.** Higher absolute return, but realised volatility runs about 16% versus roughly 12% for the rebalanced policies — a 30% jump in risk for a return that is itself an artefact of the portfolio having drifted to near-90/10 by the end. The investor who refuses to rebalance is not running a 60/40 portfolio; they are running a slowly-creeping all-equity portfolio.

There is a small additional reward for rebalancing called the rebalancing premium (or volatility harvesting, after Claude Shannon's coin-flip thought experiment). Periodic rebalancing of two volatile, imperfectly-correlated assets extracts a small extra return on top of the weighted-average return of the components, purely from the path of prices. For a 60/40 of US stocks ($\sigma=16\%$) and Treasuries ($\sigma=6\%$) at correlation $\rho=-0.3$, the premium works out to about 0.3% per year. Real, but small — and it disappears when correlations turn positive, as they did in 2022. Frame it correctly: the premium is not the reason to rebalance. The reason to rebalance is risk control and behavioural discipline. The premium is a lagniappe.

2.10 Why Rebalancing Is the Anti-Recency, Anti-FOMO, Anti-Capitulation Trade

Now connect the mechanics back to the seven biases above.

Look at what the rebalance trade does, mechanically, at the moments it fires:

• In March 2009, with the S&P down 56% from its 2007 high,

the textbook 60/40 rebalancer was a mechanical buyer of stocks at the cycle low. They did not have a view. They did not call a bottom. They were forced by the rule to top up the bond-heavy portfolio with cheap equity. Five years later that purchase had doubled. **That is the exact opposite of the capitulation trade** that retail investors made in record numbers that month — the trade that produced the largest individual contribution to the Dalbar gap of the entire twenty-year window.

• In December 2021, with the S&P up 28% on the year and

bonds soft, the same rebalancer was a mechanical seller of stocks at what turned out to be a cycle peak. Again no view, no call. That is the exact opposite of the FOMO trade — adding equity exposure after a great year — that recency bias was screaming at every retail investor to make.

• In any year a single name drifts to 30% of the portfolio,

the rebalance rule says trim it back to its target weight. That is the exact opposite of the disposition effect — riding winners and refusing to ever ring the cash register — and it is the exact opposite of the anchoring mistake of refusing to sell because "the cost basis is so low."

Notice the pattern. Every bias we covered above pushes the investor toward a specific trade. Every rebalance fires the opposite trade. That is not a coincidence. That is what makes rebalancing the single most effective behavioural-defence rule in the toolkit: it is the systematic engine that places the correct contrarian trade at the moment the human is least capable of placing it on their own.

This is why "I'll rebalance when I feel like it" or "I'll skip this year's rebalance because the news is bad" defeats the entire point. The rule's value is exactly in the moments the human wants to override it. Discretionary rebalancing is no rebalancing at all. It is a re-entry of the bias under a different name.

2.11 Tax-Aware Implementation — How to Rebalance Without Bleeding

The behavioural case for rebalancing is overwhelming. The operational case has one wrinkle: every rebalance trade in a taxable account is a potential realisation of capital gains. A 0.3% rebalancing premium is wiped out by a single realisation event with a meaningful unrealised gain. So while you must rebalance, you should rebalance cleverly.

The hierarchy of tax-aware rebalancing, in increasing order of desirability:

Sell-to-rebalance in a taxable account. Worst case.

Realises gains, locks in tax. Use only when no other lever works.

Sell-to-rebalance with tax-loss harvesting. Pair the sale

of appreciated overweight assets with the sale of any genuinely underwater positions to offset the gains.

Direct new contributions. Funnel monthly cash inflows into

whichever sleeve is below target. No realisation, no tax. This is the right answer for any investor still in the accumulation phase, and it is enough to keep most portfolios within a few percent of target indefinitely.

Rebalance in tax-advantaged accounts only. If your full

portfolio is split across IRA, 401(k), and taxable brokerage, the rebalance trades happen inside the IRA where all gains are tax-deferred, while the taxable account holds whichever slice of the allocation needs the least turnover (typically the long-term equity sleeve).

Use dividends and interest. In retirement, the cash thrown

off by the bond sleeve and dividend stocks naturally lands in cash. Direct that cash to whichever sleeve is below target before reinvesting on autopilot.

For most readers of this course, the working answer is a combination of (3) during the working years and (4) once contributions slow. (1) is reserved for the quinquennial cleanup when bands have stretched too far for contributions alone to fix.

A note on the 2022 case. In 2022 the S&P 500 returned −18.1%, the 10-year Treasury returned −17.8%, and the 60/40 portfolio finished near −18.0%. The naive rebalancer asks: do I sell stocks to buy bonds, or sell bonds to buy stocks? Both are down nearly equally. There is no winner to trim and no loser to top up. The mechanical answer is unsatisfying but correct: **rebalance back to target weights anyway.** If at year-end you sit at 59/41, trim 1% off bonds and buy 1% of stocks. The trade is small, the rebalancing premium that year is essentially zero, but you are placing the correct trade for the next year. A regime change in correlation does not change the rule; it just changes the size of the premium the rule earns. Skipping the rebalance because "it doesn't matter this year" is a discretionary act — exactly the kind of discretion the rule exists to prevent.

---

3. Common Misconceptions

Misconception 1: "I'm a rational investor; biases are for amateurs."

The biases are wired. They affect Nobel laureates, professional traders, and institutional allocators just as they affect retail. The professional difference isn't immunity — it's *system design that constrains* the bias. If you don't have such systems, you have the biases at full strength.

Misconception 2: "Loss aversion is just risk aversion."

It is not. Risk aversion is a smooth preference for less variance at a given expected return. Loss aversion is a kink at zero — a discontinuity where losses are weighted ~2.5x as heavily as equivalent gains. The kink causes path-dependent behaviour (the disposition effect) that pure risk aversion does not predict.

**Misconception 3: "If I check my portfolio more often I'll catch problems sooner."**

The opposite. Checking more often amplifies loss aversion, because you experience more individual loss events (every red day) without gaining much new information. The optimum frequency for a long-term portfolio is roughly once a year for a rebalance. More often is behavioural exposure, not improved decision-making.

**Misconception 4: "I held through 2008/2020/2022, so I won't panic-sell next time."**

Survivorship of past drawdowns is a weak predictor. The next drawdown will have a different cause, a different narrative, a different speed, and you will be at a different stage of life with a different account size. Past discipline is comforting but the correct prior is "I will face the same temptation again."

Misconception 5: "The Dalbar gap is just fees."

It is not. The fund-level returns DALBAR cites are net of fund fees. The gap between fund and investor is purely the timing of flows. Eliminating fees does not close it.

Misconception 6: "If I read more news, I'll be better-informed."

Almost the opposite. More news exposes you to more narrative fallacy, more recency bias (the headlines are by definition "what just happened"), and more herding signals. The best portfolios in the long-run sample are run by people who don't follow market news daily.

Misconception 7: "FOMO can be controlled with willpower."

FOMO is a social-comparison loop, and the loop is now in a phone in your pocket. Willpower lasts minutes; the loop runs years. The structural fix is to mute, unfollow, and set portfolio rules that don't require you to react to what your network is doing.

Misconception 8: "If I diversify enough, behaviour doesn't matter."

A perfectly diversified portfolio still has full equity-market volatility (~16-20% σ), which means a 30% drawdown is roughly once-per-decade. Diversification limits idiosyncratic risk; it does not remove the behavioural temptation to sell during a systematic drawdown.

**Misconception 9: "Rebalancing is a market-timing trade in disguise."**

It is the opposite of market timing. Market timing forms a view on near-term direction. Rebalancing has no view; it mechanically restores a fixed target. The two strategies do opposite trades at exactly the moments they disagree — the market timer chases the recent winner; the rebalancer trims it.

Misconception 10: "Rebalancing always increases return."

It does not. The rebalancing premium is small (typically 0.1%–0.4% per year for a 60/40) and disappears entirely when the correlation between sleeves runs positive. The reason to rebalance is risk control and behavioural discipline, not return enhancement.

Misconception 11: "I shouldn't rebalance during a bear market."

This is the most expensive rebalancing mistake on the list. The bear market is exactly when the rebalance trade is most valuable, because that is when you are buying the lower-priced asset. Investors who suspended rebalancing in 2008–09 and resumed in 2010 locked in a permanent underperformance against rule-followers. Skipping the rebalance is, behaviourally, a capitulation trade dressed up in cautious language.

Misconception 12: "More frequent rebalancing is better."

There is essentially no difference in long-run outcomes between quarterly, semi-annual, and annual rebalancing on a stock-bond portfolio. Daily or weekly rebalancing is actively worse — trading costs and bid-ask spreads dominate the negligible incremental premium. Annual is the institutional default for a reason.

---

4. Q&A

Q1: Is overconfidence really worse for men than women?

A: Yes — Barber and Odean's "Boys Will Be Boys" (2001) study found men trade roughly 45% more than women, and that the extra turnover translates into about 1.4 percentage points per year of lower net return. The mechanism is overconfidence: men are more likely to believe they have an edge, so they trade more on it. The behavioural cost is the trade itself, not the gender.

Q2: What's the single most useful system to install today?

A: Automatic monthly contributions to a broad-market index ETF in a tax-advantaged account, with a one-time annual calendar reminder to rebalance. That single system bypasses recency bias (you don't time entries), bypasses loss aversion at the contribution level (no decision to make in drawdowns), and dramatically reduces FOMO exposure (you're already invested, so you can't miss out).

Q3: How do I design a written investment policy statement?

A: Three sections, one page each. (1) Allocation: target weights and rebalance rule. (2) Triggers: pre-committed responses to specific events (drawdown of X%, allocation drift of Y%, life event Z). (3) Forbidden actions: things you commit to not doing (e.g., "I will not move >5% of the portfolio based on a single news headline"). Sign it. Date it. Re-read it during drawdowns before trading.

Q4: How does the disposition effect interact with taxes?

A: It compounds the damage. Holding losers and selling winners is exactly the opposite of optimal tax behaviour. Tax-loss harvesting says realise losers (use the loss against gains) and defer winners (let them compound untaxed). The disposition effect makes you do the inverse, costing both behavioural alpha and tax alpha simultaneously.

Q5: If I know a bubble is forming, should I short?

A: No. Markets stay irrational longer than you can stay solvent. Even if you correctly identify a bubble in year 1, the bubble can run another 2–3 years. A short position financed with margin will be stopped out long before the fundamentals matter. The behavioural fix is to avoid bubbles (don't add new money in), not to bet against them.

**Q6: What do I do when my brother-in-law triples his money on a meme stock?**

A: Congratulate him sincerely. Do not change your strategy. The sample of one is not data; the strategy you can articulate the risk of has positive expectancy and the strategy he ran (single name, no edge, lottery payoff distribution) does not. Survivorship bias makes the lottery winners visible; the losers are silent.

Q7: Is dollar-cost averaging just a behavioural trick?

A: Largely yes. Mathematically, lump-sum investing beats DCA in roughly 70% of historical entry windows because markets trend up. But behaviourally, DCA reduces the regret of bad timing, which makes investors more likely to actually invest the money rather than hold it in cash forever. A behavioural trick that gets the money into the market is worth more than a mathematical optimum that leaves the money in cash.

Q8: How does fat-tail awareness intersect with overconfidence?

A: Directly. Vol tail wags dog. Overconfident sizing assumes the realized volatility distribution looks like the recent quiet sample. Fat tails mean the next extreme event is larger than your sized-for sample suggested. Overconfidence-built leverage gets crushed when the fat tail materialises. The fix is to size for the tail you have not yet seen, not the centre you have already lived through.

Q9: How does this lesson connect to the barbell?

A: The barbell (covered Week 14) is in part a behavioural design: by holding extreme safety on one end (cash, T-bills) and small concentrated speculation on the other (long calls, asymmetric bets), the middle — where loss aversion is loudest, where the Dalbar gap is largest — gets stripped out. You cannot panic-sell cash; you cannot panic-sell a position that is already at its maximum loss. The barbell removes the fuel that behaviour burns.

Q10: Is automation actually a substitute for understanding?

A: It is a substitute for willpower, not for understanding. You still need to understand the strategy well enough to set it up correctly and to recognise the genuinely rare moments when the system needs to be adjusted (a regime shift, a major life event, a tax-law change). Automation is "trust the spreadsheet most of the time"; understanding is "know which 1% of the time the spreadsheet is wrong."

Q11: What is the optimal rebalancing frequency?

A: For tax-deferred accounts, annual is the conventional answer and dominates by simplicity. For taxable accounts, event-driven rebalancing — only when a 5%–10% band has been breached — minimises realisations. Either is fine. The least good answer is "monthly," which adds friction without adding return, and "whenever I feel like it," which silently re-introduces the biases the rule exists to suppress.

Q12: What is the right band width for threshold rebalancing?

A: 5 absolute percentage points around the target is a standard default for a 60/40. 10pp is reasonable for an investor who values inactivity. 1pp is too tight — the band fires constantly and the turnover eats the premium. The choice depends on tax cost and brokerage friction more than on theory.

**Q13: Should I rebalance my 401(k), IRA, and brokerage account separately or jointly?**

A: Jointly. Treat the household portfolio as one allocation. Place the high-turnover trades inside the IRA or 401(k), where they are tax-free, and keep the taxable account as the long-term buy-and-hold sleeve. This single decision saves more in lifetime taxes than nearly any other portfolio choice.

**Q14: Can I just rebalance with new monthly contributions and never sell anything?**

A: Yes, as long as the contributions are large relative to the drift. A working professional contributing 15% of salary into a balanced portfolio rarely needs to do a sell-rebalance trade in the accumulation phase. Once contributions slow (retirement, sale of a business), supplementary calendar or band rebalancing kicks in.

**Q15: What about rebalancing in a year like 2022 when both assets fell?**

A: Trade the rule. The premium that year is near zero, but the trade still positions you correctly for the next year. Skipping the rebalance is a discretionary act and exactly the kind of discretion the rule exists to prevent.

Q16: What does Horace personally do?

A: A January calendar rebalance for the IRA sleeve (auto-rebalanced by the broker, cost: zero, time: zero), contributions-based rebalancing for the taxable equity sleeve, and a 10pp band check for everything in between, fired only if the January look-through shows the portfolio out of the band. The total annual time commitment is under an hour. Anything more complex is a hobby, not a portfolio.

The interactive panel below lets you toggle on and off four classic bias-driven actions and watch the resulting wealth path versus simple buy-and-hold of the S&P 500 from 1928 to 2024. The pattern repeats: every bias rule loses to "do nothing." The strategy your spreadsheet recommends beats almost any version of you that overrides it.