Part XIII · Discipline · Chapter 13H

Managing total vega

The portfolio's exposure to volatility

1 · Why vega per position is not enough

The aggregate vega of a swing portfolio is rarely discussed in retail education. The Dealer Mechanics Lab calls it the single largest source of catastrophic loss in real premium-selling portfolios. Consider a swing trader running all of these at once:

• One SPX iron condor, short vega.
• Two short put credit spreads on SPY, short vega.
• One short strangle on QQQ, short vega.
• A program of covered calls on individual stocks, short vega.
• A LEAP collar on a core equity holding, with mixed vega.

Each position, viewed alone, looks reasonable, and the trader has given each one 1% to 2% of the portfolio's risk. But the aggregate vega, the total dollar P&L for each volatility point the VIX moves, is typically the equivalent of being short several hundred dollars of SPX vega per point. When the VIX moves from 15 to 25 over three days, a routine stress episode and not a tail event, the mark-to-market loss across the portfolio can exceed the sum of the individual stop-loss budgets. Each stop was set on the price of its underlying and sized for the loss a price move would cause. A jump in implied volatility adds a loss that no price stop sees, in every position at once, before any single stop triggers.

This is the vega correlation problem. Short-vega positions across SPX, SPY, QQQ and individual stocks are not diversified: they all lose together in a VIX-spike regime. The rule of risking 1% a trade, which works well for delta exposure, produces concentrated portfolio risk on the vega axis.

2 · Adding up the book's vega

For each position $i$, write $\nu_i$ for the option's vega per unit of the underlying per volatility point, the number a platform shows (Chapter 5C). With $n_i$ contracts, a contract multiplier $m$ of 100, and a sign $s_i$ of +1 for a long position and −1 for a short one, the position's vega in dollars per volatility point is

If your platform gives vega per unit of $\sigma$, the formula's own value, where 1.0 means 100 volatility points, multiply by 0.01 as well. The Dealer Mechanics Lab writes the rule with that factor for this reason.

Single-name vega does not move one for one with the VIX. So scale each position by the VIX beta of its underlying, the number of points its implied volatility typically moves for each point of the VIX. The book's aggregate vega is then the sum

and a VIX move of $\Delta$VIX points changes the book's value by about $V \times \Delta\text{VIX}$. The Lab gives typical VIX betas as ranges:

A single at-the-money SPX option carries a lot of vega. Chapter 5C puts a 30-day at-the-money SPX call at about 6.8 index points per volatility point with the index at 6,000 and implied volatility at 20%, about $680 a contract. Two short SPX puts like it hold about −$1,360 of vega per point on their own.

How much is too much

For a swing account of $250,000 to $1 million, the Lab puts a reasonable upper bound on aggregate short vega at $500 to $1,500 per volatility point. At that bound, a spike of 10 points in the VIX costs $5,000 to $15,000 on the marks. Beyond that, the portfolio becomes a directional bet on the VIX rather than a diversified swing book. The book in the example, at $668.50, sits inside the band.

The Lab also gives a stress threshold in proportion to the account. When aggregate short vega exceeds 1% of the net asset value (NAV) per volatility point, the portfolio is over-concentrated: trim or hedge before adding any new short-premium position. The two limits do not agree. On a $1 million account, 1% of NAV per point is $10,000 per point, almost seven times the top of the band, and a 10-point spike at that level costs 10% of NAV. On $250,000 it is $2,500 per point. Use the dollar band as the working limit, since it binds first, and treat the NAV threshold as the point past which no new short premium goes on at all.

3 · A monitoring cadence for vega

• Daily, at the end of the day: compute aggregate vega from the current marks and compare it with last week's number. A large change needs investigating.
• Weekly, at Friday's close: write a full report of the aggregate vega, each position's contribution, and the share of NAV that a spike of 10 volatility points would cost.
• Before each event, such as an FOMC meeting, a Consumer Price Index (CPI) release, Non-Farm Payrolls (NFP) or a major earnings report in a name the portfolio holds, recompute the aggregate vega. Consider trimming if it is elevated.
• At the stress threshold: above 1% of NAV per volatility point, trim or hedge before adding any new short-premium position.

4 · The vega portfolio monitor

The figure adds up a book by (13H.1) and (13H.2). Each row is a position: choose its instrument and side, and enter its contracts and its vega per contract in dollars per volatility point, from your platform at current marks. The VIX-beta adjustment uses a fixed value for each instrument: 1.00 for SPX and SPY, 0.90 for QQQ, 0.85 for IWM, 0.75 for NVDA, 0.70 for AAPL, 0.65 for MSFT and 0.50 for any other name. The figure opens with two positions, each short 5 SPY contracts at $25 of vega, and holds up to six. A strip marks the aggregate against the bands at $500 and $1,500 a point, and readouts give the book's P&L for VIX spikes of 5, 10 and 15 points.

Below the strip sits the macro vega overlay. Choose a target vega, flat, half or full, and set the VIX's spot level, 17 when the figure opens. The overlay counts the 90-day VIX calls, struck near 1.4 times the VIX's spot level and carrying about $80 of vega each, that take the book's vega to the target. Flat means a full hedge, down to zero vega. Half keeps half of today's vega. Full leaves the whole vega in place, so it buys nothing, and for a book that is long vega the overlay counts calls to sell instead. The monitor has no input for delta or for days to expiry: it needs only each position's vega at today's marks.

The opening book's two positions of $-5 \times 25 = -\$125$ make −$250 a point, so a spike of 10 points costs $2,500. At the half target the overlay buys $\lceil 125/80 \rceil = 2$ calls struck near $1.4 \times 17 \approx 24$, where $\lceil\ \rceil$ rounds up. Then enter the worked example's six positions and switch between the targets.

positions (2/6)

+ add position

Position	Ctr	Vega / ctr	β · vega_adj
SPXSPYQQQIWMNVDAAAPLMSFTOther shortlong ×			β=1.00 · −⁠$125
SPXSPYQQQIWMNVDAAAPLMSFTOther shortlong ×			β=1.00 · −⁠$125

aggregate vega = −⁠$250 / vol point

VIX +5 P&L = −⁠$1,250

VIX +10 P&L = −⁠$2,500

VIX +15 P&L = −⁠$3,750

target vega flat half full

VIX spot

Macro vega overlay. To reach the half-vega target (−⁠$125/vp), buy approximately 2 90‑DTE VIX calls at strike ~24. Informational sizing only; actual vega per VIX call varies by IV and DTE.

For the worked book the aggregate reads −$669 a point, the figure's rounding of −$668.50, inside the band, and a spike of 15 points costs $10,028. A half target buys $\lceil 334.25/80 \rceil = 5$ calls, a flat target 9, and a full target none. Switching a covered call from NVDA to "other" shrinks its contribution, since the figure's beta for other names is 0.50. The counts are informational only, because a VIX call's vega varies with implied volatility and time to expiry.

5 · Hedging short vega: the macro vega overlay

Three practical hedges fit a book with too much short vega:

1. Long-dated VIX calls, 90 to 180 days to expiry. The cleanest macro hedge of vega. Strike them at a VIX of 25 to 30, well above the mean of about 17 for the curve from the VIX9D to the VIX3M. Size them to cover half of the aggregate short vega, since a full hedge costs too much carry and defeats the purpose of harvesting the VRP. Roll them quarterly. The figure's strike of 1.4 times spot lands in that range when the VIX is between about 18 and 21.
2. Long SPX puts, 90 to 180 days to expiry, struck at 90% to 92% of the index. These carry direction along with vega. They are cheaper than VIX calls in absolute dollars but more correlated with the delta book, so they are less pure as a hedge of vega. They are useful when the swing portfolio is already net long delta.
3. Purchases of long-dated SPX call spreads or put spreads, 60 to 90 days to expiry. They have defined risk and a small positive vega. They are less powerful than the first two but cheaper, and useful as a partial offset when the short vega is modest, below 0.5% of NAV per volatility point.

Do not hedge short vega with more option sales. Selling deeper out-of-the-money (OTM) puts to "fund" the hedge increases the aggregate short vega rather than reducing it. The hedge must consume premium, not generate it. Chapter 12F lists the same first two hedges for a short-volatility book, and Chapter 10I sets these VIX calls apart from the VIX exchange-traded products it rejects.

6 · Common failure modes

7 · The trade journal

Every trade in the framework is logged with the same fields. Implement the template in any tool, such as Notion, Obsidian, Sheets or plain markdown, and keep the field set the same across trades:

Trade ID: YYYY-MM-DD-N

=== PRE-TRADE ===
Date and time of decision:
Symbol and instrument:
Setup type: [Trend Pullback | Compression Breakout |
             Undercut and Reclaim | IBS+RSI |
             Pre-earnings momentum | Other]
Macro regime (composite signal):
Stage classification:
Required conditions (check all):
  [ ] Composite signal conditions met
  [ ] Setup-specific entry trigger fired
  [ ] No event-day overlay restrictions
  [ ] Position size within portfolio limits
  [ ] Conviction score:
        E (evidence):
        M (mechanism):
        R (regime):
        X (execution):
        S (sizing):
        C (product):

=== EXECUTION ===
Mid price at decision:
Limit order submitted:
Fill price:
Fill time:
Bid-ask spread at fill:
Slippage against mid (bps):
Time of day:

=== POSITION ===
Risk budget (dollars):
Risk budget (% of portfolio):
Entry price:
Stop level:
Position size:
Notional deployed:
Notional as % of portfolio:
Time stop:
Profit target:

=== MANAGEMENT ===
[Adjustments during the trade: additions, partial exits, stop moves]
[Allowed: a stop trailed up behind a winner, or a full exit.
 Not allowed: stop migration, moving the stop away from the price.]

=== OUTCOME ===
Exit reason: [profit target | stop | time stop |
              discretionary | event-driven]
Exit price:
Exit time:
Days held:
P&L (dollars):
P&L (% of portfolio):
R-multiple (P&L / risk budget):

=== POST-TRADE REVIEW ===
What went as planned:
What surprised me:
Errors made:
Framework lessons:
Should I have skipped this trade, in hindsight? Why?
Did the regime change during the hold?

The management line allows only two stop moves: a trailing stop raised behind a winning long position, or a full exit. It forbids stop migration, moving the stop further from the price as the price approaches it (Chapter 11N), which turns a planned loss into a larger one. A trade's R-multiple is its P&L divided by the risk budget, R (Chapter 11H).

Three principles make the journal work:

1. Write everything before the trade. The pre-trade fields are filled in before you click buy, which catches lazy setups.
2. Do not editorialize the outcome. Record what happened, not what should have happened.
3. Review monthly. Scan the trades of the prior 4 to 6 weeks and look for patterns: which setups produced positive R-multiples, and which regimes worked.

The journal is the framework's external memory. Without it, the framework cannot improve over time.

8 · Check yourself