Part XI · Swing trading · Chapter 11L

Single stocks and the index

When a stock trade is really an index trade

Where we are

Chapters 11H to 11K built setups that run on the index and on single stocks. This chapter chooses between the two for a given setup, since the vehicle sets the costs, the precision of sizing, the stock-specific risk and the tax. It ends with a trap: several stock positions can quietly add up to one index-sized bet.

What you need first

SPX, XSP, SPY and ES, their multipliers and notional (Chapter 2A), and Section 1256 against ordinary tax (Chapter 2E).
Beta, a stock's sensitivity to the market (Chapter 1G), and correlation and the variance of a portfolio (Chapter 1F).
The effective number of correlated signals, $k/[1 + (k-1)\rho]$ (Chapter 11C).
The four setups and their vehicles (Chapter 11H), pre-earnings momentum (Chapter 11I) and relative strength (Chapter 11B).
Sizing from a loss budget, $n \le \lfloor B_{\text{exp}}/L_{\max} \rfloor$ (Chapter 13C).

1 · Two kinds of risk in one stock

The four base setups and the pre-earnings overlay apply both to index instruments, SPX, SPY and QQQ, and to single names: the Mag-7, momentum stocks and sector leaders. The choice between them has consequences for execution friction, the precision of position sizing, idiosyncratic risk and tax treatment.

Idiosyncratic risk is the part of a stock's risk that belongs to the company alone. The one-factor model splits a stock's return into a market part and the company's own part:

R_i = \beta_i R_m + \varepsilon_i

(11L.1)

Here $R_i$ is the stock's return, $R_m$ the market's, $\beta_i$ the stock's beta and $\varepsilon_i$ the stock's own return, uncorrelated with the market's. The model leaves out a constant, which does not affect risk. Because the two parts are uncorrelated, their variances add:

\sigma_i^2 = \beta_i^2 \sigma_m^2 + \sigma_\varepsilon^2

(11L.2)

where $\sigma_m$ is the market's volatility and $\sigma_\varepsilon$ the volatility of the stock's own return. An index position carries only the first term, while a single-stock position adds the second.

Worked example: One stock, two risks

Step 1: the parts. A hypothetical stock has $\beta = 1.3$ , in a market with a volatility of 16%, and an own volatility of 25%. Its market variance is $1.3^2 \times 0.16^2 = 0.0433$ and its own variance $0.25^2 = 0.0625$ .

Step 2: the total. By (11L.2) its variance is 0.1058, a volatility of 32.5%. The market explains $0.0433/0.1058 = 41\%$ of the variance, leaving 59% as the company's own.

A single-stock trade is therefore worth its extra risk only when your view is about $\varepsilon_i$ : a catalyst, a sector tailwind or a company-specific dynamic. A view about $R_m$ is cheaper to hold through the index.

2 · When the index is the better vehicle

Use SPX, XSP, SPY or QQQ when:

The setup is regime-driven: the composite signal favors broad-market action (Chapter 11G).
You want to express a macro view without single-name idiosyncratic risk.
Tax efficiency matters, since XSP and SPX options are Section 1256 contracts while SPY gets what the source calls 1099 treatment. A swing trade's gain on SPY is then a short-term capital gain (Chapter 2E).
Execution liquidity is paramount: SPX and XSP have institutional-grade spreads at all strikes.
You are scaling size: the SPX contract multiplier permits efficient large position sizing.

The tax difference is easy to price: at a 37% ordinary rate and a 20% long-term rate, the 1256 blend is $0.6 \times 20\% + 0.4 \times 37\% = 26.8\%$ (Chapter 2E). A $10,000 swing gain then costs $2,680 in tax on XSP against $3,700 on SPY.

The multiplier explains the source's default for the trend-following setups, the Trend Pullback, the Compression Breakout and IBS + RSI: XSP for retail traders below a $1 million account, and SPX above that.

Worked example: Why XSP below $1 million

Step 1: one contract. With SPX at 6,000, one SPX option carries $600,000 of notional and one XSP option $60,000 (Chapter 2A). For a deep in-the-money call with a delta near 1, a stop 1% away costs about $6,000 a contract on SPX and $600 on XSP.

Step 2: a $500,000 account. Risking 1%, the budget is $5,000, so the loss-budget rule allows $\lfloor 5{,}000/6{,}000 \rfloor = 0$ SPX contracts but $\lfloor 5{,}000/600 \rfloor = 8$ XSP contracts.

Step 3: a $2 million account. The budget is $20,000, which buys 3 SPX contracts or 33 XSP contracts. Both fit, and SPX does the job in a tenth as many contracts.

The Dealer Mechanics Lab adds that the Trend Pullback in particular has decades of clean backtest support on SPX and SPY, while single-name applications have less robust evidence and more execution friction. The Edge Book is more doubtful that a retail account can capture short-horizon mean reversion at all (Chapters 1G and 11B).

3 · When a single stock is the better vehicle

Use single names, the Mag-7 or momentum stocks, when:

The setup is specifically idiosyncratic: the stock has a unique catalyst, a sector tailwind or a company-specific dynamic.
Pre-earnings momentum applies, which is viable only on single names with earnings events (Chapter 11I).
Relative-strength (RS) dynamics matter, as with a strong stock inside a sideways index (relative strength, Chapter 11B).
Single-name dispersion is high: the index is range-bound, but its members are moving.
You are explicitly diversifying away from index correlation.

Dispersion measures how far the members' returns over a period spread around their average:

D = \sqrt{\frac{1}{n}\sum_{i=1}^{n} \left(R_i - \bar R\right)^2}

(11L.3)

Here $n$ is the number of members, $R_i$ is member $i$ 's return and $\bar R$ their average. Four equally weighted members returning +3%, −2%, +5% and −6% leave their index flat, at $\bar R = 0$ . Their dispersion is $\sqrt{(9 + 4 + 25 + 36)/4} = 4.3\%$ , a flat index hiding members that are all moving.

The Mag-7 are NVDA, AAPL, GOOGL, MSFT, META, AMZN and TSLA: Nvidia, Apple, Alphabet, Microsoft, Meta, Amazon and Tesla. Their options have institutional-grade liquidity comparable to SPY's, with clean bid-ask spreads, depth and term-structure availability. Outside the Mag-7, single-name options execution deteriorates noticeably. Only the top 30 to 50 names by options volume have execution clean enough for swing-options strategies.

The spread is the cost to watch: an option quoted $4.90 bid and $5.10 ask has a spread of $0.20, 4% of its $5.00 mid. Buying at the ask and selling at the bid costs that 4% of the premium even if the mid never moves. At the thin end, Chapter 1F cites pre-earnings single-stock option spreads averaging 31.7% of the mid-quote (Govindaraj, Li and Zhao, 2020).

4 · Several stocks, one bet

Single-name positions are often less diversifying than they appear. Mag-7 stocks have pairwise correlations of 0.5 to 0.8 during normal conditions and 0.8 to 0.95 during stress. A portfolio of NVDA, AAPL and GOOGL long swing positions is fundamentally one large bet on Mag-7 momentum, not three independent bets.

The one-factor model explains why stress raises correlations. Two stocks share only the market term of (11L.1), so their covariance is $\beta_i\beta_j\sigma_m^2$ . Taking each $\sigma$ from (11L.2), their correlation is

\rho_{ij} = \frac{\beta_i\,\beta_j\,\sigma_m^2}{\sigma_i\,\sigma_j}

(11L.4)

When the market's volatility rises and the stocks' own volatilities do not, the shared term comes to dominate both numerator and denominator, and $\rho_{ij}$ climbs toward 1.

Worked example: Correlation rising with the market's volatility

Step 1: calm. Two hypothetical stocks each have $\beta = 1.2$ and an own volatility of 20%. With the market's volatility at 15%, the shared variance is $1.44 \times 0.0225 = 0.0324$ and each stock's variance $0.0324 + 0.04 = 0.0724$ , so (11L.4) gives $\rho = 0.0324/0.0724 = 0.45$ .

Step 2: stress. The market's volatility rises to 40%, while the stocks' own volatilities stay at 20%. The shared variance is $1.44 \times 0.16 = 0.2304$ and each variance 0.2704, so $\rho = 0.85$ .

How many independent bets do $N$ correlated positions make? Chapter 11C answered the same question for indicators: $N$ positions of equal risk, with a common correlation $\rho$ , diversify as much as $N_{\text{eff}}$ independent positions, where

N_{\text{eff}} = \frac{N}{1 + (N-1)\,\rho}

(11L.5)

At $\rho = 0.55$ , three positions are worth $3/2.1 = 1.43$ independent bets. At $\rho = 0.9$ they are worth $3/2.8 = 1.07$ : one bet.

Familiar ground

A response trial counts patients, not lesions. Three metastases in one patient share that patient's biology, so their responses move together. Counting them as three responses would overstate the evidence. Three Mag-7 longs are three lesions in one patient: the factor they share, Mag-7 momentum, is the unit that responds and so the unit of risk.

Treat correlated positions as one

The source's operational rule for portfolio-level sizing is to treat correlated single-name positions as a single position for risk-budget purposes. If your portfolio loss budget is 2% per day for active swing positions, and you have three Mag-7 longs running, each should be sized at 0.67% individual risk, not 2% each. In stress, with the correlation near 1, the three stops tend to be hit together, so the loss is the sum of the three risks.

Worked example: Three names in one budget

Step 1: the split. On a $300,000 account a 2% daily budget is $6,000, so three Mag-7 setups firing together get $6{,}000/3 = \$2{,}000$ each, 0.67% of the account.

Step 2: one position. For NVDA, with a stop $5 below entry, the loss-budget rule gives $\lfloor 2{,}000/5 \rfloor = 400$ shares.

Step 3: the stress day. If all three stops are hit together, the loss is $6,000, the budget. Sized at 2% each, the same day would cost $18,000, or 6%.

This is the principle worked through in Example 10 of Chapter 13D, where a SPY Trend Pullback and an AAPL Undercut and Reclaim share a 2% budget of $8,000, split $5,000 and $3,000.

The figure shows the seven names' pairwise correlations as a grid, each cell shaded more strongly as the correlation rises. One slider, the regime stress dial, blends an illustrative calm matrix into an illustrative crisis one. A label names the regime, from calm through mixed and stress to crisis, beside a readout of the average off-diagonal correlation. Both matrices are illustrative, set at the rough size of realized correlations in the mid-2020s rather than measured. The figure's calm matrix also runs below the prose: its cells span 0.35 to 0.70, against this chapter's 0.5 to 0.8. Its crisis cells, 0.82 to 0.92, sit inside the 0.8 to 0.95 of stress. Start at calm and read the TSLA row, then drag the dial to crisis.

regime stress dial Calm

avg corr = 0.52

Diversification leaves when it is needed. Pairwise correlations of the seven Mag-7 names, from an illustrative calm matrix to an illustrative crisis one, each set at the rough size of mid-2020s realized correlations rather than measured. The dial blends one into the other: the average off-diagonal correlation climbs from 0.52 at calm to 0.88 at crisis, where the seven trade as a single asset. Measure concentration risk against the stress correlation, not the average.

At calm the average reads 0.52, and TSLA's row, at 0.35 to 0.45, is the closest thing to a diversifier. At crisis the average is 0.88, and TSLA's row has caught up with the rest, at 0.82 to 0.88: the seven trade as a single asset. The source's trio, NVDA, AAPL and GOOGL, sits at 0.55 in every calm pair and at 0.88 to 0.92 in crisis, so by (11L.5) it falls from 1.43 independent bets to about 1.07. As the caption says, measure concentration risk against the stress correlation, not the average.

5 · The decision matrix

The source maps each setup to a default vehicle and an alternative:

Setup type	Default vehicle	Alternative if available
Trend Pullback (broad regime)	XSP for retail, SPX for size	SPY for share-level execution
Compression Breakout (broad)	XSP or SPX	SPY
Compression Breakout (single name, strong RS)	Mag-7 equity directly	Long call spread for defined risk
Undercut and Reclaim (broad)	XSP or SPX	SPY
Undercut and Reclaim (single name)	Mag-7 equity directly	Long call spread
IBS + RSI Mean Reversion	QQQ shares	SPY shares, or XSP options
Pre-earnings momentum	Mag-7 equity directly	Long call spread or diagonal
Catalyst macro event	XSP or SPX option structure	None: a single-name event would be a different framework

The matrix puts IBS + RSI Mean Reversion on QQQ shares, while the source's default above puts all three trend-following setups, IBS + RSI among them, on XSP below $1 million. The QQQ default matches the setup's own vehicle in Chapter 11H and the note on QQQ below.

6 · A note on QQQ

QQQ deserves mention as a hybrid instrument. It is an index exchange-traded fund (ETF), with cleaner execution and diversification than single names. It is also heavily concentrated in the Mag-7, with 60% of its weight in its top 10 holdings. For swing-trade purposes QQQ behaves as "Mag-7 lite": broad enough to avoid pure single-name idiosyncratic risk, and tech-concentrated enough to capture momentum-driven dispersion when growth leads. Its options are equity options, taxed like SPY's (Chapter 2E).

How much company risk a fund keeps depends on its weights. If the members' own returns are uncorrelated, a fund with weights $w_i$ keeps an own variance of $\sum_i w_i^2 \sigma_{\varepsilon,i}^2$ , which concentration keeps large.

Worked example: How much company risk a concentrated fund keeps

Step 1: the top ten. Take a hypothetical fund of 100 stocks, each with an own volatility of 25%. Its top 10 hold 6% each, 60% in all, and contribute $10 \times 0.06^2 \times 0.0625 = 0.00225$ of own variance.

Step 2: the other 90. They share the remaining 40%, about 0.44% each, and add only 0.0001.

Step 3: the comparison. The fund's own volatility is $\sqrt{0.00236} = 4.9\%$ , against 25% for one stock and $\sqrt{100 \times 0.01^2 \times 0.0625} = 2.5\%$ for the same 100 stocks equally weighted. The stocks of a tech-heavy fund also share a growth factor, which this sum leaves out, so the fund keeps more than 4.9%.

QQQ is the preferred vehicle for:

IBS + RSI Mean Reversion setups, where tech-heavy dispersion creates cleaner mean-reversion signals than the broader SPX.
Compression Breakouts during tech-led regimes, when growth stocks lead the index.
Setups where you want broad exposure, but to growth and tech specifically rather than to value and cyclicals.

In practice

Group positions by factor before you enter. List the open positions by the risk they share: the index, Mag-7 momentum, growth. Add the planned risks within each group, and hold each group's sum to one budget.
Price the spread before the setup. Quote an option's spread as a share of its mid, and set the round trip against the move the setup targets.
Put index exposure in 1256 contracts in a taxable account: XSP below $1 million, SPX above, and SPY only for share-level execution (Chapter 2E).
A stock brings its own calendar. Every single-name position carries an earnings date, and sometimes a binary event (Chapter 11K), where the index carries only the macro calendar.

When it fails

Diversification across stocks fails when it is needed. Correlations rise with the market's volatility, as (11L.4) shows, so a book of names sized as independent is most concentrated in a sell-off.
A correct company view can be stopped out by the market. In the first worked example the market explains 41% of the stock's variance, so a broad fall can take out the stop of a sound thesis about the company.
Liquidity is a property of calm markets. In stress, option spreads widen (Chapter 1F), which raises the cost of leaving any position.

Decision rule

A regime-driven setup or a macro view: trade the index, with XSP below a $1 million account, SPX above it and SPY for share-level execution.
An idiosyncratic setup: trade the stock, only among the Mag-7 or the top 30 to 50 names by options volume, and only if you can name the company-specific driver.
Several positions on one factor: give the group one risk budget and divide it, so that three Mag-7 longs in a 2% budget get 0.67% each. Chapter 11N suggests capping simultaneous long Mag-7 swing positions at two.
Growth-tilted exposure, or IBS + RSI: QQQ shares, the source's preferred vehicle for both.

Common mistake: counting three tickers as three bets

NVDA, AAPL and GOOGL at 2% risk each look like three separate bets. In stress they move as one, so the book risks 6% on a single factor. Size the group, not the ticket.

What to carry forward

A stock's variance is market variance plus its own, $\beta^2\sigma_m^2 + \sigma_\varepsilon^2$ . Trade a single name only when your view is about its own part.
The index wins on regime-driven setups, macro views, tax, liquidity and scale: XSP below a $1 million account, SPX above.
Mag-7 options trade with SPY-like liquidity, while execution deteriorates outside the top 30 to 50 names by options volume.
Correlations rise with the market's volatility, so correlated names are one bet in stress and should share one budget.
QQQ is "Mag-7 lite": an index fund that keeps some of its top holdings' company risk and a tilt to growth.

7 · Check yourself

Question 1

A stock has a beta of 1.5, the market's volatility is 20% and the stock's own volatility is 30%. What share of its variance is market risk?

Answer

By (11L.2), the market part is $2.25 \times 0.04 = 0.09$ and the own part is also 0.09, so market risk is half the variance, with the stock's volatility at $\sqrt{0.18} = 42.4\%$ .

Question 2

Two stocks each have a beta of 1.0 and an own volatility of 25%, with the market's volatility at 25%. What is their correlation, and what does it become if the market's volatility doubles?

Answer

The shared variance is 0.0625 and each variance $0.0625 + 0.0625 = 0.125$ , so (11L.4) gives $\rho = 0.5$ . At 50% market volatility the shared variance is 0.25 and each variance 0.3125, so $\rho = 0.8$ .

Question 3

You hold four single-stock longs with a common correlation of 0.6. How many independent bets are they worth, and how many at 0.9?

Answer

By (11L.5), four positions at a correlation of 0.6 are worth $4/(1 + 3 \times 0.6) = 4/2.8 = 1.43$ independent bets, and at 0.9 only $4/(1 + 3 \times 0.9) = 4/3.7 = 1.08$ .

Question 4

Your ordinary rate is 37% and your long-term rate 20%. What tax does a $10,000 short-term swing gain cost on SPY and on XSP?

Answer

On SPY it is taxed at 37%, or $3,700, against $2,680 on XSP at the 1256 blend of 26.8%, which is $1,020 less.

Question 5

Which vehicle does the matrix give to a broad-regime Trend Pullback in a $400,000 account, to pre-earnings momentum in NVDA, and to an IBS + RSI setup?

Answer

XSP for the Trend Pullback, since the account is below $1 million. NVDA shares for pre-earnings momentum, or a long call spread or diagonal for defined risk. QQQ shares for IBS + RSI, with SPY shares or XSP options as alternatives.