Monthly Review — March 2026

Abstract

March 2026 was my first real geopolitical stress test. The U.S.–Israel strikes on Iran (Feb 28, a Saturday) compressed five weeks of extreme volatility into the period from March 2 (the first trading day after the attack) through April 9. The Strait of Hormuz was effectively closed for most of the period, WTI crude spiked from USD 67 to a peak of USD 112, and the S&P 500 drew down roughly 7.6% from its January ATH. My time-weighted return for the window was −3.48%, underperforming SPX (≈+1.85% to +2.5% depending on start date) by roughly 5–6 percentage points. The underperformance has two distinct sources: (1) a clean but expensive PGR capitulation on March 27 that realized −USD 1,053.83, closing a position that I had been continuously accumulating since June 2025 without ever properly re-validating the thesis, and (2) a structural cash build from 3.5% to 60% that protected me from further downside but caused me to miss the April 7 ceasefire rebound. The cash build was the mirror image of February’s stock-heavy violation: Core ETFs collapsed from 17.9% to 3.6% as I exited RSP and SCHG during the late-March selloff. The defining meta-lesson is the distinction between decision quality and outcome quality: my April 6 bearish posture was defensible given the information available at the time, but the deeper failure was that I had no pre-committed action plan tied to specific triggers — most of the period’s decisions were improvisational, dressed up as discipline. That, not the bearish view itself, is the real process failure.

Data window: March 2 to April 9, 2026. The window extends past month-end because the April 7 TACO ceasefire is inseparable from the March story and any clean cut would distort the narrative.

TL;DR

• TWR: −3.48% over the period; SPX ≈ +1.85% to +2.5%; relative underperformance ~5–6pp.
• PGR capitulation: −USD 1,053.83 realized on 3/27. This single trade accounts for 99% of the period’s realized loss of −USD 1,067.79. Excluding PGR, net realized P&L was approximately flat. PGR was a position I had been adding to since June 2025 — nine months of continuous accumulation without independent re-validation.
• Plan B rotation: sold SCHG (73 shares, −USD 285) and RSP (24 shares, −USD 178) on 3/30–4/2 to rotate out of broad ETFs during the selloff. Opened V at USD 295 on 3/27.
• Sleeve drift, mirror image of February: Core ETFs 17.9% → 3.6% (violation, far below floor). Cash 3.5% → 60% (violation, far above ceiling). Stocks 63% → 40%. Gold 15.3% → 7.9%.
• GLDM carried the month again: +USD 521 unrealized + USD 43 realized = +USD 564 total. Validated as the cleanest hedge of the period.
• Zone framework untested at design levels: VOO never reached Zone C (USD 565) or Zone D (USD 550); the lowest close was approximately USD 580. I am not loosening the zones in response — see §7.2.
• TACO ceasefire (4/7): I had assigned <10% probability to a two-week ceasefire. It happened anyway. SPX +2.51% on 4/8, WTI −16.4%. The portfolio participated in the equity side via its 40% stock sleeve but missed the larger rebound the 60% cash would have captured.
• Biggest unrealized pain: MSFT at −USD 576 (15 shares, avg cost USD 411.47). Q3 earnings on April 29 is the key catalyst.

1) What happened (context + market environment)

1.1 The Iran war arc

The war was launched on Saturday, February 28 — markets were closed. The first trading day under the new geopolitical reality was Monday, March 2. The rough timeline:

• Feb 28 (Sat): U.S.–Israeli strikes on Iran; Supreme Leader Khamenei killed. Iran’s command structure decapitated. Markets closed.
• March 2 (Mon, first session): Brent crude opens ~+6% to USD 77. SPX gaps lower. The IRGC moves to close the Strait of Hormuz over the following days.
• Early–mid March: WTI climbs from USD 67 into the USD 80s, then USD 90s. By early April, 187 tankers are stranded inside the Gulf with 172 million barrels of seaborne crude (Kpler).
• March 18: First U.S. strike on the South Pars gas field. European gas +7%, oil +5%.
• March 27: Selloff trough. SPX bottoms around 6,473 (−7.6% from the 1/28 ATH of 7,002). CNN Fear & Greed Index hits 10 — extreme fear. This is the day I capitulate on PGR.
• March 30: First sharp relief rally on premature ceasefire rumors. Dow +1,100 points. Rumor proves false within 24 hours.
• April 6 (Mon evening): Second strike on South Pars.
• April 7 (Tue): Iran’s “99th operation” hits U.S. bases and Jubail. Trump’s 8 p.m. deadline approaches. At roughly 8 p.m. ET, ceasefire announced — two weeks, conditional on Strait reopening. Brokered by Pakistan.
• April 8 (Wed): SPX +2.51% to 6,782. Dow +2.85% (+1,325pts, best day in a year). WTI −16.41% to USD 94.41 — largest single-day drop since April 2020. VIX −22%.
• April 9 (Thu): SPX +0.62% to 6,824. But WTI reverses back up to USD 97.87 (+3.67%) as the market realizes Iran still controls the Strait. Iran’s parliament speaker claims the U.S. has already violated the ceasefire. Only 4 ships transit the Strait on 4/8 — fewer than 4/7 (11) or 4/6 (9).

1.2 What the ceasefire actually is

A critical distinction I have been underweighting: this is a two-week pause, not a resolution. Key structural facts as of 4/9:

• Iran retains operational control of the Strait.
• Even in a ceasefire scenario, oil prices are floored well above the pre-war USD 67 baseline.
• The trust deficit between the two sides is described as “significant” by every major research house.
• Yardeni Research lowered U.S. recession odds from 35% to 20% post-ceasefire but explicitly warned “a two-week pause is not a resolution.”
• J.D. Vance publicly called it “a fragile truce” in Budapest on April 8.

My forward look continues to assume this ceasefire is more likely to fracture than to extend into a durable deal.

1.3 My mental state

The month broke into two phases. Through March 27 I was executing a coherent defensive playbook and felt in control. After the PGR capitulation I had a productive two weeks of detailed analysis — but in retrospect, I was over-indexing on a single scenario (sustained bearish) and treating every new data point as confirmation. When the ceasefire came, my first reaction was “we failed.” That framing was wrong, and recognizing why it was wrong is what produced this month’s most important meta-lesson.

2) Performance & attribution

2.1 Time-weighted performance

• Period TWR: −3.48% (IBKR calculation, 3/2–4/9)
• Benchmarks: SPX ≈ +1.85% to +2.5%; VOO ≈ USD 605 → USD 625.02 ≈ +3.3%
• Relative underperformance: approximately 5–6 percentage points vs. SPX

Two observations about the gap:

1. Through March 27, the portfolio was likely close to or ahead of SPX (SPX −5% from start, portfolio relatively protected by gold and cash). The V-reversal from 3/28 through 4/9 is where the gap opened.
2. The realized loss on PGR is permanent. The missed rebound is reversible. They are not equivalent sources of underperformance and I will not let one obscure the other in my self-assessment.

2.2 Net asset value bridge

Net deposits of USD 25,668 dominate the headline NAV increase. The “real” investment performance is the −USD 2,145 MTM plus the small positive items, netting to roughly −USD 1,806 for the period.

2.3 Realized P&L by ticker

Excluding PGR, the realized P&L is −USD 13.96 — essentially flat. The entire period’s realized loss is one position’s ghost.

2.4 Unrealized P&L (period end)

Combined realized + unrealized = −USD 1,195.93. For a five-week period that included a closed Strait of Hormuz, an oil spike to USD 112, and a ~7.6% SPX drawdown, this is not catastrophic — but it is worse than a pure hold-everything strategy would have produced over the same window.

3) Trades & actions

3.1 Phase 1 — Mechanical defensive setup (3/2–3/6)

Executed in the first week, before the situation could be properly assessed. The goal was to launch defensive posture per Playbook and use volatility to begin building conviction positions.

3.2 Phase 2 — Mid-war accumulation (3/8–3/25)

This is where the first judgment error surfaces. Buying 5 more ACN on 3/19 at USD 193 was an attempt to average down on a position that was already wrong at entry. I was conflating two questions: “is ACN cheap?” and “do I want more ACN?” The right answer to the first didn’t imply the right answer to the second.

Note on EUR: the 5K @ 1.153 buy on 3/8 was an emergency, separate-tranche transaction, not a fill on my 8K @ 1.14 limit. The original 8K @ 1.14 limit order is still pending and unaffected. This is operational housekeeping that I want to record clearly: the 5K was paid above target out of necessity, the 8K remains live waiting for the target.

3.3 Phase 3 — Capitulation day + Plan B (3/27–4/2)

The pivotal week. On 3/27 I cleared six positions in a single session, the largest being PGR. This was not a panic sale — it was a deliberate decision to stop paying the attention tax on a position whose sizing was wrong. The psychological relief of the next three weeks validated the decision even before any price recovery.

4) Structural analysis

4.1 Sleeve allocation (Feb → Mar)

This is the mirror image of February’s violation. In February I was stock-heavy (63% vs. 50% ceiling) and cash-light (3.5% vs. 5% floor). In March I ended cash-heavy (60% vs. 30% ceiling) and Core-light (3.6% vs. 35% floor). The Playbook bands have now been violated in both directions within two months. That itself is a process signal worth flagging: rules that get violated in opposite directions in adjacent months are not really constraining behavior — they are documenting it after the fact.

4.2 Why the Core ETF collapse was partially correct but partially not

The Plan B reasoning on 3/30–4/2 was: during a war with open-ended duration, broad SPX exposure (SCHG, RSP) carries beta risk without compensating me for the quality I actually want to own. Better to sell broad beta, hold cash, and redeploy into quality individual names at Zone C/D levels.

What was correct: the logic is internally consistent. If Zone C/D had triggered, the redeployment would have looked good.

What was wrong: I never built a “Zone C/D never triggers” branch into the plan. By selling SCHG and RSP at March lows, I made an implicit directional bet that the market would go lower. It didn’t. The broad ETFs I sold at USD 28.30 and USD 190 would be worth more at 4/9 prices (USD 30.45 and USD 198.08) than I received. Net drag: roughly USD 120 on SCHG, USD 70 on the RSP lots.

The structural lesson: never liquidate an existing position on the implicit assumption that a future entry opportunity will materialize. If the framework says “wait for better prices,” the right action is to do nothing, not to raise cash against a specific price that may never come.

4.3 End-of-period concentration

Software/data: MSFT + SPGI = 14.3% (down from 21.1% after ACN exit). The implicit sector bet has been reduced but still exists.

5) What went wrong

5.1 PGR — the original sin compounded over nine months

PGR was not a February mistake. I started buying PGR in June 2025 and continued to add to it across the second half of the year and into early 2026 — almost nine months of continuous accumulation. At the start of the war I held 29.95 shares with a cost basis around USD 236, having made adds on at least four separate occasions. Not once during those nine months did I sit down and write an independent valuation of PGR. I treated each add as “buying lower,” and each addition felt rational in isolation, but the cumulative position size and the cumulative cost basis were never tested against the question “would I buy this position at this size today, from cash, if I had no prior history with it?”

The −USD 1,054 realized on 3/27 is not the cost of a bad exit. It is the cost of nine months of un-validated accumulation, paid in a single line item. The exit on 3/27 was correct — once the sleep test had been failing for weeks, getting out cleanly was the only useful action left. But the lesson is not about exits. It is about the silent compounding of an error that is too small to trigger any single review.

Rule: Any position I have added to more than twice without writing a fresh independent valuation note is, by definition, oversized. Continuous “buy lower” without re-validation is not averaging down — it is anchoring to a sunk cost.

5.2 Zone framework was never stress-tested at design levels

Zone C was set at VOO = USD 565 and Zone D at USD 550. The actual March low on VOO was approximately USD 580 — about 2.7% above the Zone C trigger. The framework was never tested. A naive read of this would be “the zones are too aggressive, recalibrate them looser.” I am explicitly not doing that.

The reason is structural. The fact that VOO bottomed at only −5% from the start of the war, during a period when the Strait of Hormuz was closed, oil hit USD 112, and Iran was striking U.S. bases, is itself a piece of information that deserves more attention than the surface read suggests. A normal market in a normal regime would have produced a deeper drawdown. This one didn’t — and the rebound from the lows was characterized by:

• Thin volume on up days
• Narrow breadth (mega-caps doing the work)
• A TACO-driven sentiment reversal rather than fundamental improvement
• Reluctance to test the lows even when the catalysts (April 6 South Pars strike, April 7 Jubail attack) genuinely justified it

The most charitable interpretation is that passive flows and earnings beats are mechanically supporting the index. A less charitable interpretation is that the rally has the texture of distribution into retail strength — a setup, not a recovery. I am not confident enough in the second interpretation to act on it aggressively, but I am confident enough not to loosen the zones in response to a non-test. The next real downside test may produce the trigger. If I recalibrate the zones now, I will have surrendered the framework to a benign single observation.

5.3 No pre-committed action plan tied to specific triggers

This is the meta-failure that ties most of the period’s errors together. I had a Playbook. I had Zones. I had stop-loss rules. But almost every individual decision during the war — the ACN average-down, the timing of the PGR exit, the SCHG/RSP liquidations, the V opening — was made on the fly, in response to whatever the day’s narrative looked like. None of those decisions executed a specific pre-committed rule. They felt disciplined because I could justify each one in the moment, but justification is not pre-commitment. A real rule would have produced the same action regardless of the day’s news.

The honest summary: my decisions during the war were improvisational, dressed up as discipline. The Playbook acted as a vague background reference, not as an executable program. When the ceasefire came, I had no pre-committed response — and the absence of a response is itself evidence that the framework wasn’t operating as a framework.

Rule: If a decision cannot be traced to a specific pre-committed rule with a specific trigger, it is improvisation. The acceptable frequency of improvisation is low — say, no more than 1 in 5 trades. Anything above that means the Playbook needs more rules or the existing rules need more triggers.

5.4 Sub-optimal EUR execution

The 8K EUR @ 1.14 limit was placed correctly and remains live. The 5K @ 1.153 emergency buy on 3/8 was forced by the German residency deadline. This is not strictly a market mistake — it is an operational lesson: FX orders tied to hard external deadlines should be placed early enough that the limit can be set conservatively without risking the deadline. I delayed setting up the EUR conversion until the war was already in motion, which was structurally late.

6) What went right

6.1 Discipline preserved the account

Despite a five-week war, a 7.6% SPX drawdown, a USD 1,054 realized loss, and unprecedented noise on financial media, I made zero panic-driven trades. Every action was documented and at least loosely referenced a thesis. The system held — even though, per §5.3, it held more loosely than I want.

6.2 PGR sleep-test action was correct

Clearing PGR when it was causing genuine attention drag was the right call, regardless of loss magnitude. The “sleep test” framework — if a position is measurably degrading my attention, the hidden cost already exceeds the expected reversion value — proved itself in real time. I felt the difference within 48 hours.

6.3 GLDM as the period’s anchor

GLDM delivered +USD 564 total contribution (≈+USD 521 unrealized + USD 43 realized). This is the second consecutive month it has been the top positive attribution. The 10-share trim on 3/27 at USD 90 was well-timed; the remaining 60 shares at USD 94.38 continue to carry.

6.4 Cash floor was respected

The USD 4,000 hard cash floor was never at risk. In fact, cash was over-preserved. The discipline of never letting cash go below the floor, even under active-deployment pressure, is now second nature.

6.5 Ceasefire rebound participation

The 40% stock sleeve did participate in the 4/8 ceasefire rebound. GLDM, MSFT, AMZN, SPGI, V all moved favorably. The portfolio’s relative position at period end is meaningfully better than at the 3/27 low, even if worse than a buy-and-hold counterfactual.

7) Lessons learned

7.1 Continuous accumulation without re-validation is not averaging down

PGR was added to repeatedly across nine months without ever being re-validated as if it were a fresh position. Each individual add was small enough to feel like “DCA”; cumulatively they built a 13% concentration that was never independently justified. Anchoring to a falling cost basis is not the same as updating a thesis. The new rule (§5.1): more than two adds without a fresh valuation note → position is by definition oversized.

7.2 Untested zones do not get loosened — they wait for the next real test

The Zone framework wasn’t tested in March because the drawdown was shallow, not because the zones were too aggressive. The shallowness of the drawdown is itself a structural data point, possibly signaling that the rally has support that is mechanical (passive flows, buybacks) or worse, that it has the texture of a setup into retail. Either way, recalibrating the zones based on a non-test would be a soft surrender. The zones stand. The next real test will reveal whether they need adjustment.

7.3 Decision quality ≠ outcome quality

The single most important meta-lesson of the period. On April 6, with South Pars just bombed, Iran’s 99th operation hitting U.S. bases, Trump’s 8 p.m. deadline approaching, and the Strait still closed, a bearish posture was defensible given the information available. On April 7, a low-probability ceasefire materialized. If I judge the decision by the outcome, I will conclude “I should have gone all in” and will take reckless positions next time. The correct framing: the directional view was reasonable, but the absence of pre-committed counter-scenarios was the real failure.

7.4 Improvisation dressed as discipline is the most expensive failure mode

I had a Playbook, Zones, and rules. I also made nearly every decision of the period on the fly. The Playbook was a vibe, not a program. The fix is not more rules — it is making the existing rules executable: every rule must have a specific trigger and a specific action, written down before the situation arises. If a decision cannot be traced to such a rule, it is improvisation, and improvisation has a hard cap.

7.5 TACO pattern is more durable than I priced

The June 2025 twelve-day war ceasefire taught me the Trump-Always-Chickens-Out pattern once. The April 2026 ceasefire taught me again, in a larger and longer conflict. I should stop treating TACO as “one specific Trump negotiating tactic” and start treating it as “the default base case for geopolitical conflicts in this administration.” The burden of proof is now on the non-ceasefire scenario, not the ceasefire one. This does not contradict §7.2 — TACO is about the de-escalation of a specific conflict; the structural-rally concern is about the broader equity market regime. Both can be true simultaneously.

8) Rule updates (process changes after this month)

8.1 Re-validation requirement for accumulated positions (new)

After two adds to any position, no further adds are permitted until I have written a fresh valuation note that ignores the existing position. The note must answer: “If I were starting from cash today and this position did not exist, would I buy it at this price and at this proposed size?” If the answer is no, no add. This closes the PGR loophole.

8.2 Pre-committed rule discipline (new)

Every trade that is not a simple stop-loss execution must reference a specific pre-committed rule by name (e.g., “Zone C trigger,” “post-earnings add per MSFT plan,” “sleep-test exit”). If I cannot name the rule, the trade is improvisation. Improvisation cap: no more than 1 in 5 trades in a calendar month. If the count crosses that threshold, the entry rule for the next month is “no new positions, only stop-loss execution,” until the rule set is updated.

8.3 Counter-scenario plan requirement (new)

Any directional thesis driving trade decisions must be accompanied by a written Plan B for the opposite outcome. The Plan B must specify: (1) what the counter-scenario looks like, (2) what I would do if it materializes, (3) what pre-committed cash or position I will deploy. If I can’t write the Plan B, the main thesis is too strong for the evidence.

8.4 Never liquidate on an implicit price assumption

Selling an existing position to fund a hypothetical future entry is prohibited unless the replacement entry is also pre-committed as a specific market-order plan (not a limit that may never fill). This closes the SCHG/RSP loophole.

8.5 Core ETF rebuilding priority

Because Core ETFs collapsed to 3.6%, first-priority rebalancing in April and May is all new cash into RSP or VOO until Core returns to at least 25%. No new individual-stock positions until this condition is met, with one exception: pre-committed MSFT adds at USD 345/USD 335 (legacy Zone D limits) remain valid.

8.6 Zones explicitly held

Despite not being triggered in March, Zone C (USD 565) and Zone D (USD 550) remain unchanged. They will be re-evaluated only after the next real downside test, not after a non-test.

9) Forward look — April and May

9.1 The ceasefire scenario tree

The two-week ceasefire clock runs to approximately April 21. Between now (4/9) and then, four branches matter:

In all four scenarios the first-priority action converges: rebuild Core ETFs with new cash, not by liquidating existing positions.

9.2 Key dates

9.3 Order book adjustments

• Cancel: MSFT limit @ USD 350 (redundant; superseded by Zone D USD 345/USD 335).
• Keep: MSFT USD 345 and USD 335 limits (legacy Zone D).
• Keep: AMZN stop-loss USD 175, AXP stop-loss USD 260, GLDM sell @ USD 100, VOO sell @ USD 640.
• Adjust: TLN sell limit from USD 450 to USD 480 (oil “new normal” raises fair value upward).
• Keep: EUR 8K @ 1.14 limit unchanged (separate tranche from the 5K @ 1.153 emergency buy).

9.4 What I will not do

• I will not chase the 4/8–4/9 rebound with market orders. Forced entries after a surprise rally is exactly the wrong response to being wrong-footed.
• I will not reduce MSFT before 4/29 earnings. The thesis (forward PE ~20x on a wide-moat business) is intact; the −USD 576 is a catalyst-dependent drawdown, not a thesis break.
• I will not expand TLN. The position is the right size for the AI-power thesis.
• I will not re-enter PGR. The exit was correct; the lesson was bought with the tuition; do not re-engage.
• I will not treat the ceasefire as permanent peace. Plan C (fracture) remains the second-most-likely scenario until the April 21 expiry says otherwise.
• I will not loosen the Zone framework based on March’s non-test.

10) End-of-period portfolio snapshot (4/9)

Total NAV: USD 71,450.53 Market value (equities): USD 28,544.33 Cash: USD 42,884.46 (USD USD 41,151 + HKD USD 1,733) Unrealized P&L: −USD 128.14 Realized P&L (period): −USD 1,067.79 (securities) + −USD 58.50 (EUR FX)

Holdings

Signed off 2026-04-10, during a fragile two-week ceasefire whose durability I do not trust.