Part X · Edges, one by one · Chapter 10G

Calendar and flow effects

Fed meetings, quarter-ends, expiry weeks and seasons

Where we are

Chapter 10F leaned the equity sleeve toward stocks with chosen traits. This chapter takes the third category of Chapter 10A, structural overlays: calendar and flow effects that scale exposure you already hold. Their economic size is typically 10 to 40 bps a year, so the whole question is whether you can capture them more cheaply than they pay.

What you need first

The four sources of edge (Chapter 1G) and the five categories, with the cost test for an overlay (Chapter 10A).
The Sharpe ratio (Chapter 1D) and covariance (Chapter 1F).
Trading costs (Chapter 2D), taxes (Chapter 2E), and ES futures and SPY (Chapter 2A).
Options expiration and pin risk (Chapter 5F), and same-day options (Chapter 9B).
The conviction score (Chapter 13B).

Overlays modulate exposure to other premia rather than serving as strategies of their own. Their magnitudes are small, so they work best as cheap changes in the size of positions you already hold. This chapter examines five: the FOMC cycle, end-of-quarter rebalancing, the OPEX week, the Halloween effect and the turn of the month.

In plain terms

An overlay earns nothing by itself. It changes how much you hold of something you already own, at moments chosen by the calendar or by a predictable flow. Well-chosen moments earn the same holdings a little more for each unit of risk, but with effects this small, the cost of the changes can easily exceed what they add.

1 · What an overlay is, and what it is not

An overlay is a rule that systematically scales exposure up or down according to the calendar date, a regime indicator or an institutional flow pattern. It does not generate return on its own; it modulates the return on other strategies. A common retail error is to treat an overlay as a strategy in its own right, and the Edge Book gives three examples:

Treated as a strategy	Treated as an overlay
"Trade SPX only in FOMC even weeks": too restrictive, and costs eat the small premium	Modulate equity sizing across the weeks of the FOMC cycle, using existing positions
"Trade the EOQ rebalance every quarter": a small premium per trade, only about 4 trades a year	Add small tactical positions at the end of a quarter when the divergence is extreme
"Sell premium specifically in OPEX week": a narrow window, and diluted since 0DTE	Be aware of OPEX dynamics without trading around them: context, not a strategy

Implemented cheaply, an overlay improves a portfolio's Sharpe ratio by 0.05 to 0.20. Implemented expensively, it lowers it.

The arithmetic of an overlay

Write $R_t$ for the return of a position in period $t$ and $k_t$ for the overlay's scaling of it: 1.0 for full size, 0.75 for a quarter off. The overlaid position returns $k_t R_t$ , and its expected value splits into two parts:

\begin{aligned} \mathbb{E}[k_t R_t] = {}&\mathbb{E}[k_t]\,\mathbb{E}[R_t] \\ &+ \operatorname{Cov}(k_t, R_t) \end{aligned}

(10G.1)

The first part is what a static position at the overlay's average size would earn anyway. The second, the covariance of Chapter 1F between the scaling and the return, is the overlay's whole contribution, and it must exceed the overlay's costs. An overlay adds expected return only if it holds more when returns are higher.

Worked example: Two kinds of week

Step 1: the static part. Suppose, hypothetically, half of all weeks return 0.6% and the other half 0.0%, and the overlay holds full size in the first kind and 75% in the second. A static position at the average size of 0.875 earns $0.875 \times 0.3\% = 0.2625\%$ a week.

Step 2: the overlay's part. The overlay earns $0.5 \times 0.6\% + 0.5 \times 0.75 \times 0\% = 0.30\%$ a week. The difference of 0.0375% a week is $\operatorname{Cov}(k_t, R_t)$ , about 1.9% a year before costs.

The figure runs the example through a made-up year of 52 weeks. Raise the cost per dollar traded, and the net falls to zero at 15 bps. Then press shuffled: the covariance vanishes, and the costs stay.

return gap 0.60%

cut 25%

cost per $ traded 0 bps

cuts fall on

The overlay earns 0.3000% a week. A static position at its average size, 0.875, earns 0.2625%. Cov(k, R) = +0.0375% a week, or +1.95% a year before costs.

costs = 0.00% a year, 52 switches

break-even cost = 15.0 bps

net = +1.95% a year

An overlay earns the covariance, less its costs. The bars are a made-up year of expected weekly returns R, alternating between good and poor weeks a return gap apart, as the FOMC cycle's even and odd weeks do, and averaging 0.3% a week. The line above is the size the overlay holds, k, and the filled part of each bar is what it keeps, k × R. Costs charge every change of size for each dollar traded. Press shuffled, and the cuts land on good and poor weeks alike: the static part stays, the covariance falls toward zero, and only the costs remain.

2 · The FOMC cycle

In plain terms

The Federal Reserve's rate-setting committee meets about every six weeks. Cieslak, Morse and Vissing-Jorgensen (2019) showed that since 1994, the entire equity premium has been earned in the even weeks of the cycle between meetings. The odd weeks returned near zero or less. The proposed mechanism is the "Fed put": the Fed's reaction function reduces the risk of a large fall around the windows in which it is expected to speak.

The Federal Open Market Committee (FOMC) is that committee. Count the weeks from each meeting: the meeting week is week 0, the week after is week 1, and so on to week 5, since the cycle is six weeks long. Even weeks are 0, 2 and 4; odd weeks are 1, 3 and 5. The Fed put is the market's belief that the Fed responds to a sharp fall by easing policy.

Worked example: Where the premium sits

Step 1: the average day by week. Cieslak and colleagues found an average return of about 14 bps a day in week 0, about 11 bps a day in weeks 2 and 4, and about 0 bps in weeks 1, 3 and 5.

Step 2: a whole cycle. Five trading days a week give $5 \times (14 + 11 + 11) = 180$ bps from the even weeks and about none from the odd ones. Holding equity only in the even weeks, 3 of every 6, would have captured most of the long-run equity premium with half the time exposed.

Formally, the claim is that expected returns differ sharply by where a day falls in the cycle:

\begin{aligned} &\mathbb{E}[R_t \mid t \in \{w_0, w_2, w_4\}] \\ &\quad \gg \mathbb{E}[R_t \mid t \in \{w_1, w_3, w_5\}] \end{aligned}

(10G.2)

Here $R_t$ is the day's return and $w_0$ to $w_5$ are the weeks of the cycle. If the odd weeks truly return nothing, cutting them changes the risk but not the return, so the overlay's gain shows up in the Sharpe ratio.

Worked example: What cutting the odd weeks does to the Sharpe ratio

Step 1: the assumption. Suppose, hypothetically, that every day has the same volatility $\sigma$ . Over the 30 days of a cycle, a full position then earns 180 bps with a variance of $30\sigma^2$ .

Step 2: cut the odd weeks by 25%. The return stays 180 bps. The variance falls to $15\sigma^2 + 15 \times 0.75^2\sigma^2 = 23.4\sigma^2$ , so the Sharpe ratio rises by $\sqrt{30/23.4} = 1.13$ , or 13%. Leaving the odd weeks entirely would raise it by $\sqrt{30/15} = 1.41$ , or 41%, before any costs or taxes.

The evidence

Cieslak, Morse and Vissing-Jorgensen (2019, Journal of Finance, JF) is the original paper, with a sample running from 1994 to about 2014.
An important caveat, which a reviewer of the Edge Book corrected: out-of-sample (OOS) replication after 2019, after COVID and through the hiking cycle of 2022 is limited. Some practitioner replications suggest the effect held through 2019. Data through the aggressive hiking cycle of 2022 to 2024 are thin, and the effect was possibly disrupted.

Why it exists

The FOMC cycle is an institutional and structural effect, Source 3 of Chapter 1G. The Fed's policy reaction function reduces the tail risk of equities during the windows in which an announcement is expected. The Fed put creates an asymmetry between upside and downside in the even weeks.

The chapters drawn from the Dealer Mechanics Lab look at the meeting itself through dealer hedging. Implied volatility is bid up into the announcement and collapses after it. Dealers who are short puts then buy back hedges, producing a ramp after the event (Chapter 5D). Chapter 7B puts the most that flow typically adds to the SPX move at 0.3% to 0.5%, and calls it highly reliable across years. Chapter 11K turns it into a brief long trade in the 30 to 90 minutes after the announcement, at no more than 0.5% of the portfolio at risk. Chapter 13F halves new swing entries from two days before the meeting to one day after it. The Edge Book does not test that trade. It treats the FOMC calendar only as an overlay on the equity sleeve, and it rejects the standalone FOMC trade it does test, the drift before announcements (Chapter 10I).

In practice

The Edge Book's implementation is ETF rotation: full SPY exposure in even weeks and reduced exposure in odd weeks. Use no derivatives unless the capital in the overlay exceeds $200k.

Sizing: hold equity exposure at 1.0 times the baseline in weeks 0, 2 and 4, and reduce it by 20% to 30% in weeks 1, 3 and 5.
Cost discipline: if costs exceed 100 bps a year, simplify, for example by reducing only in week 1 rather than in all the odd weeks.

The Edge Book's own worked example, in Chapter 13E, reaches a stricter verdict for a 60% allocation held in SPY. Every sale realizes gains, and in a taxable account that tax can cost more than the cycle pays. It concludes that you should run no separate rotation, even in a tax-deferred account. Instead, time new contributions and rebalances to land in even weeks, or change the size of a futures position taxed under Section 1256 (Chapter 2E).

When it fails

The effect could decay with publication and with reforms to the Fed's transparency, as the drift before FOMC announcements did after 2015 (Chapter 10I). The aggressive hiking cycle of 2022 to 2024 may have disrupted the pattern. And the out-of-sample evidence is thinner than the original paper suggests.

Decision rule

Implement the cycle modestly, and treat it as a T2 overlay, not a structural certainty. Reduce equity by 20% to 30% in odd weeks through ETF rotation, and hold full exposure in even weeks. In a taxable account, Chapter 13E's stricter reading, which moves only contributions and rebalances onto the cycle, is the safer default. Track your performance against a static allocation, and suspend the overlay if the rolling two-year return of the odd weeks reaches or exceeds that of the even weeks.

Common mistake: "the FOMC cycle is robust because it has a clear Fed-policy mechanism"

A plausible mechanism does not make an effect persist. The drift before FOMC announcements had a clear mechanism and decayed after 2015, when the Fed's transparency reforms took hold. The even-week cycle could decay the same way. Use it, but do not bet the portfolio on it.

3 · End-of-quarter rebalancing

In plain terms

Pension funds with a 60/40 mix, target-date funds and similar mandates rebalance toward their target weights at the end of each quarter. If equities outperformed bonds during the quarter, such a fund is now overweight stocks, so it sells equities and buys bonds. If equities underperformed, it does the reverse. Front-running this flow earns a small premium.

Rebalancing is trading a portfolio back to its target weights, and front-running here means trading ahead of a flow you expect. Chapter 1G showed the forced trade for one fund; this section adds the direction rule and the threshold.

Worked example: A strong quarter for stocks

Step 1: the drift. The Edge Book's example is the first quarter (Q1) of 2024, with SPX up 10% and the 10-year Treasury down 1%. A fund that starts the quarter at 60/40 holds $66 of stocks and $39.60 of bonds for every $100 it began with. That is 62.5% stocks, which the Edge Book rounds to 63/37.

Step 2: the trade. Getting back to 60/40 means selling about 2.5% to 3% of the fund in stocks and buying the same in bonds. Across the roughly $5 trillion held in 60/40 mandates, 3% is about $150 billion of expected stock selling and bond buying in the last week of March.

The imbalance is the drift of the stock weight $w_t$ from its target $w^*$ :

\text{Imbalance}_t = w_t - w^*

(10G.3)

When stocks have outperformed, $w_t > w^*$ and rebalancers sell stocks. The front-runner does the same first: short ES, the E-mini S&P 500 future of Chapter 2A, and long ZN, the 10-year Treasury note future. The size of the imbalance follows from the quarter's two returns.

How far a 60/40 fund drifts

Start with weight $w^*$ in stocks, which return $R_s$ , and $1 - w^*$ in bonds, which return $R_b$ . At the end of the quarter the stock weight is

w_t = \frac{w^*(1 + R_s)}{w^*(1 + R_s) + (1 - w^*)(1 + R_b)}

(10G.4)

For small returns, a first-order expansion gives $w_t - w^* \approx w^*(1 - w^*)(R_s - R_b)$ . With $w^* = 0.6$ , $R_s = 10\%$ and $R_b = -1\%$ , the exact weight is $0.66/1.056 = 0.625$ , and the approximation gives $0.24 \times 0.11 = 0.026$ .

The approximation shows why the signal is the SPX-bond divergence, $R_s - R_b$ , the gap between the quarter's stock and bond returns. At a divergence of 1.5%, a 60/40 fund drifts only $0.24 \times 1.5\% = 0.36\%$ of its value, too small a flow to trade ahead of.

The evidence

Etula, Rinne, Suominen and Vaittinen (2020), "Dash for Cash", in the Review of Financial Studies (RFS).
NBER Working Paper 33554 (2024), "The Unintended Consequences of Rebalancing": predictable next-day effects on equity returns of about 17 bps when stocks are overweight, and larger effects when the divergence is extreme.

Why it exists

End-of-quarter rebalancing is an institutional friction, Source 3 of Chapter 1G. Mandate-driven flow is mechanical and driven by no fundamental view. The premium compensates the front-runner for providing liquidity.

In practice

Signal: a divergence between SPX and bonds of more than 1.5% either way over the quarter.
Direction: if stocks outperformed bonds, go short ES and long ZN, because rebalancers are about to sell stocks and buy bonds. If stocks underperformed, go long ES and short ZN.
Timing: enter 5 trading days before the quarter ends, and exit 1 to 3 trading days into the new quarter.
Sizing: very small until you have your own data, at 0.5% of the portfolio at risk per trade for the first 8 quarters.

Chapter 13E works a quarter through for a $1 million portfolio, with one ES contract short and one ZN contract long.

Get the sign right

An earlier version of the Edge Book had this sign backward, and its reviewer caught the error. The book calls the correction the most important mechanical one in its text. It restates the rule plainly: if equities outperformed bonds in the quarter, you go short equities and long bonds. Rebalancers sell their equity overweight to buy their bond underweight, and front-running their flow means doing what they are about to do.

When it fails

The effect needs a meaningful divergence, above 1.5%, so skip flat quarters. Other market participants sometimes pre-empt the flow. And the paper is recent: retail replication is untested.

Decision rule

Treat end-of-quarter rebalancing as T2: "promising, replication required". Trade tiny sizes for the first 8 quarters of your own data, and do not size up until you have statistical confidence in your own execution. Two kill criteria apply:

After 4 quarters, suspend the trade if its hit rate is below 50% or its net P&L is negative.
If the quarter's divergence is below 1.5%, skip the trade.

Common mistake: "I can size this big because the academic Sharpe is good"

The NBER paper of 2024 is recent, and live retail replication is untested. A significant gap separates "the academic effect exists" from "I can capture it after slippage and execution". Size this like a research project, not a foundation strategy.

4 · The OPEX week

In plain terms

The third Friday of each month is the monthly options expiration (OPEX). Stivers and Sun (2013) documented an excess return of about 30 bps a week on large companies with active options during OPEX week. The mechanism is unclear, possibly related to dealer hedging into expiry.

The toy strategy holds SPY only during the week that ends on the third Friday. Historically, that captured about 30 bps a week of excess return, concentrated in option-heavy large companies: about 3.6% a year on paper over twelve such weeks, if the effect persisted and cost nothing.

The evidence

Stivers and Sun (2013), in the Journal of Banking and Finance (JBF), is the original SPY OPEX-week strategy.
A decay caveat: since 2022, weekly and 0DTE options have exploded, and the importance of the monthly third-Friday expiry has diluted significantly. The effect has not been retested rigorously in the regime since 0DTE.

Why it exists

The OPEX week is a microstructure and institutional effect, Sources 4 and 3 of Chapter 1G. Dealer hedging into expiry creates flow patterns, and option-related rebalancing concentrates around OPEX.

The Dealer Mechanics Lab chapters describe those flows in detail and agree that they are small. Charm, the drift of an option's delta with time, draws dealer buying into OPEX Fridays (Chapter 5D). Gamma concentrates at the strike in the final hours, the source of pin risk (Chapter 5F). Chapter 9C treats pinning as real but small, a tiebreaker rather than a trade. Chapter 2A adds the risk of the settlement print on expiry morning.

In practice

Do not allocate to the OPEX week explicitly: use it as awareness only. If you implement it anyway, remember that the regime since 0DTE may already have eroded the effect.

When it fails

Since 0DTE options arrived, the monthly expiry's effect has been diluted. It may already be priced in. And its small per-trade premium is dominated by a retail account's transaction costs.

Decision rule

Do not trade specifically around OPEX. Use your awareness of it to avoid being caught in moves driven by dealer hedging around expiry, but do not size positions to capture the effect.

5 · Halloween: sell in May

In plain terms

"Sell in May and go away": equities have historically returned more from November to April than from May to October. The Halloween effect persists across more than 36 countries, despite having no clean mechanism.

Worked example: Two halves of the year

Step 1: the averages. For SPX from 1950 to 2024, the November-to-April half returned about 7% on average and the May-to-October half about 2%.

Step 2: the gap. The difference of 5% is large, persistent and global. Yet no satisfying causal mechanism for it exists.

The evidence

Bouman and Jacobsen (2002, American Economic Review, AER) is the original study: 36 of 37 countries show the effect.
Jacobsen and Zhang (2021), in the International Review of Financial Analysis (IRFA), cover 109 markets and 323 years in aggregate: 82 of the 109 show the effect.

Counting markets is not the same as counting independent trials, because stock markets move together (Chapter 1F). A toy shows how much that matters.

Worked example: Twenty markets that behave like four

Step 1: independent markets. If 20 independent markets each had an even chance of showing a winter gap by luck, 18 or more of them would show it with probability $(190 + 20 + 1)/2^{20} = 211/1{,}048{,}576$ , about 1 in 5,000.

Step 2: correlated markets. Suppose instead that the 20 move so closely together that they carry the information of 4 independent markets. All 4 agreeing then happens by luck 1 time in $2^4 = 16$ . The same headline count is far weaker evidence.

Why it exists

There is no clean mechanism. The possibilities include:

Tax timing: selling at the year's end to harvest losses, then reinvesting in January.
Reduced participation during summer vacations.
Seasonal mood, as in seasonal affective disorder (SAD).

The empirical persistence is striking despite an unclear mechanism, and that is itself a flag. Either the mechanism is being missed, or the pattern is a bias specific to the sample.

Familiar ground

Read against the Bradford Hill criteria, the Halloween effect is strong on consistency, across dozens of countries and centuries of data, and has no plausibility and no experiment. An association with that profile earns a watch, not a change of practice. Even its consistency is weaker than it looks, as sites in a multicenter trial that share one protocol are not independent confirmations.

In practice

At most, use a soft tilt. Reducing equity exposure by 10% to 20% from May to October has a small positive expected effect. That is not enough to justify trading costs unless you implement it at the level of your regular rebalancing.

Decision rule

Use a soft tilt only. If you rebalance once a year, time the rebalance to add equity in November and to reduce it in May. Do not run a separate strategy for the effect.

6 · The turn of the month

The first 4 to 5 trading days of a month show modestly positive average returns. The turn-of-the-month effect is attributed to month-end pension contributions and 401(k) retirement deposits being invested in the first days of the next month. The evidence is a modest empirical pattern, whose persistence is narrowing as the patterns of these cash flows change.

Decision rule

Treat the turn of the month as marginal, and do not allocate against it. It is useful only as a soft consideration when you time a larger discretionary purchase. If you have one to make, prefer the first week of the month to the last.

7 · How this book files the calendar effects

The Edge Book files two of these effects differently in different chapters, as Chapter 10A explains. This book uses one classification throughout, the one in the table, and Chapter 1G's filing agrees with it:

Effect	Source (Chapter 1G)	Mechanism	Use
FOMC cycle	Institutional (Source 3)	The Fed's reaction function	Built into equity rebalancing rules
End-of-quarter rebalancing	Institutional (Source 3)	Mandated flows	A small tactical trade below 0.5% of the portfolio
OPEX week	Microstructure, with an institutional part (Sources 4 and 3)	Dealer hedging into expiry	Awareness only
Halloween	Behavioral seasonality (Source 2), provisionally	None found	The timing of an annual rebalance
Turn of the month	Institutional (Source 3), by its proposed mechanism	Contributions being invested	Awareness only

The Halloween filing is provisional, since no mechanism has been found. The turn of the month is filed by its proposed mechanism, which the Edge Book states without classifying.

8 · Adding the overlays together

The Edge Book estimates what each overlay adds to a portfolio's Sharpe ratio:

Overlay	Addition to the Sharpe ratio
FOMC even-week tilt	0.10 to 0.30, with caveats
End-of-quarter rebalancing	0.05 to 0.15, untested live
OPEX week	About 0
Halloween, through rebalance timing	0.05 to 0.10
Turn of the month	About 0

The net realistic addition is 0.10 to 0.30 if all of them are implemented well, and near zero or negative if they are implemented expensively. Adding the rows gives 0.20 to 0.55, well above that net, because overlays do not stack cleanly. They overlap, since the FOMC cycle and Halloween both time the same equity exposure. They share sensitivities to the regime, and implementation costs eat their small premiums. Chapter 10A's worked examples show how correlation caps a combined Sharpe ratio and how quickly costs consume a gain of this size.

The operational principle is that overlays are cheap tilts. Once you pay meaningful transaction costs, taxes or attention to capture them, they are not worth it:

FOMC cycle: built into your equity rebalancing rules, with no separate execution.
End-of-quarter rebalancing: a small tactical sleeve at under 0.5% of net asset value (NAV) per trade.
Halloween: the timing of your annual rebalance.
OPEX and the turn of the month: awareness only.

If an overlay requires extensive operational infrastructure, it is miscategorized.

When it fails

Overlays fail when their costs, taxes or attention exceed their small gains, as they usually do when run as separate strategies. They fail when their mechanism changes, as the drift before FOMC announcements did. They also fail when the effect was a feature of its sample, which is the open question for Halloween.

Decision rule

Express every overlay through positions you already hold, as a change of size, never as a new sleeve.
Check the cost first, and simplify or drop any overlay whose costs and taxes approach its expected gain.
Keep the end-of-quarter trade below 0.5% of the portfolio at risk until 8 quarters of your own data support it.
Apply the kill rules of each section, and suspend an overlay the moment one triggers.

Common mistake: "several overlays can stack to add 0.5 or more of Sharpe ratio"

Overlays are cheap to add but also cheap to overstate. Their aggregate addition is 0.10 to 0.30, and even that is optimistic, because they overlap, share sensitivities to the regime and lose their small premiums to costs. Do not build a portfolio around overlays.

Overlays contribute least to a portfolio's Sharpe ratio, yet they are the most operationally seductive, with specific dates and clear stories. Treat them as small tilts. Chapter 10H turns to regime filters, which also resize other positions but are read from the market rather than the calendar.

What to carry forward

An overlay scales positions you already hold and earns only the covariance between its scaling and the returns, less its costs.
The FOMC cycle is the strongest overlay, a T2 effect with thin evidence since 2019: cut equity by 20% to 30% in odd weeks, as cheaply as your account allows.
End-of-quarter rebalancing is T2 and untested live: if stocks beat bonds by more than 1.5%, go short ES and long ZN, at tiny size.
The OPEX week and the turn of the month are awareness only, and Halloween is a soft tilt in an annual rebalance.

9 · Check yourself

Question 1

State the finding of Cieslak, Morse and Vissing-Jorgensen, the rule for implementing it, and why the thin out-of-sample evidence matters.

Answer

Since 1994, the entire equity premium was earned in weeks 0, 2 and 4 of the FOMC cycle, while the odd weeks returned near zero. Hold full exposure in even weeks, cut it by 20% to 30% in odd weeks, and suspend the tilt if odd weeks match even weeks over two years. The paper's sample ends around 2014, and pre-FOMC drift shows that an effect with a clear mechanism can vanish when the Fed changes how it communicates.

Question 2

One quarter, stocks return +2% and bonds +1.5%. Another, stocks return −6% and bonds +2%. What does the end-of-quarter rule do in each?

Answer

In the first quarter the divergence is $2\% - 1.5\% = 0.5\%$ , below the 1.5% threshold, so you skip the trade. In the second it is $-6\% - 2\% = -8\%$ . Stocks underperformed, so 60/40 funds are underweight stocks and must buy them: you go long ES and short ZN, 5 trading days before the quarter ends, at no more than 0.5% of the portfolio at risk.

Question 3

Why has the OPEX-week effect probably diluted since 2022, and what do you do about it?

Answer

Weekly and 0DTE options have exploded, so the monthly third-Friday expiry carries a smaller share of the options market and of the hedging around it. The effect has not been retested rigorously since. You do not trade around OPEX. You use your awareness of expiry to avoid being caught in hedging-driven moves.

Question 4

State the Halloween effect. Why is the lack of a mechanism a concern, and how would you use the effect?

Answer

Equities have returned more from November to April than from May to October: about 7% against 2% for SPX from 1950 to 2024, with the effect in 36 of 37 countries in one study and 82 of 109 markets in another. With no mechanism, the pattern could be a feature of the sample. Use it only to time an annual rebalance, adding equity in November and reducing it in May.

Question 5

With scores like those of Chapter 13E's worked example, score rotating SPY on the FOMC cycle as a standalone strategy in a taxable account: $E = 0.50$ for T2 evidence, $M = 0.65$ , $R = 0.65$ , $X = 0.40$ and $S = 0.85$ . What binds?

Answer

$C = 0.50 \times 0.65 \times 0.65 \times 0.40 \times 0.85 = 0.072$ , far below the 0.3 threshold of Chapter 13B. Execution, $X$ , binds: in a taxable account every rotation of SPY realizes gains, and the tax can exceed the cycle's premium.

Question 6

Why are overlays modulators rather than strategies, and which is expected to add the most?

Answer

Their premiums are small, typically 10 to 40 bps a year, so a separate strategy's costs and taxes consume them, and only a change in the size of existing positions is cheap enough. The FOMC even-week tilt is expected to add the most, 0.10 to 0.30 of Sharpe ratio with caveats.