# Adopted Strategy After 5,000 One-Year Simulations

## Simulation Setup

- Date: 2026-06-06.
- Start capital: 103,000,000 KRW.
- Runs: 5,000 simulations.
- Horizon: 252 trading days.
- Objective: maximize KRW net profit and challenge completion while rejecting strategies whose downside pain is too large.
- Regime calibration: 10 years of historical adjusted-close market data from yfinance.
- Data type: historical-regime-calibrated Monte Carlo, not a deterministic historical backtest.

The simulator is available here:

- `/Users/jangbogeun/Documents/Codex/2026-06-06/2/outputs/strategy_monte_carlo_simulator.py`

Result CSV:

- `/Users/jangbogeun/Documents/Codex/2026-06-06/2/outputs/strategy_monte_carlo_results.csv`

Historical calibration files:

- `/Users/jangbogeun/Documents/Codex/2026-06-06/2/outputs/historical_regime_config.json`
- `/Users/jangbogeun/Documents/Codex/2026-06-06/2/outputs/historical_regime_stats.csv`
- `/Users/jangbogeun/Documents/Codex/2026-06-06/2/outputs/historical_regime_calibration_report.md`

## Strategies Tested

1. `passive_hold`: mostly keep current high-beta holdings.
2. `panic_sell_rebuy`: sell after drawdown, rebuy after signals improve.
3. `defensive_trend`: cash and quality-first trend defense.
4. `adaptive_barbell`: quality/core plus tactical leaders, cash, limited SOXL/SOXS.
5. `guarded_aggressive_barbell`: more aggressive SOXL/SOXS, but with cash, stops, and crash-risk override.
6. `aggressive_leverage`: strongest SOXL/SOXS usage when directional signals appear.

## Selection Rule

The strategy is selected by KRW net profit and challenge completion, not average return optics.

Disqualify a strategy if:

- P5 net profit is negative.
- Probability of ending below the start is above 5%.
- P5 max drawdown is worse than -30,000,000 KRW.

The -30,000,000 KRW rule exists because a strategy that earns a lot on average but demands a very large painful drawdown can still be wrong for the user's account and family stability.

## Results Summary

| Strategy | Median Net Profit | P5 Net Profit | Median Max DD | P5 Max DD | Avg Weekly Clears | Avg Monthly Clears | X2 Hit | Status |
| --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | --- |
| guarded_aggressive_barbell | 82,304,918 | 33,908,967 | -11,788,752 | -21,576,261 | 21.53 | 3.89 | 31.04% | SELECTED |
| adaptive_barbell | 63,679,758 | 26,424,197 | -9,252,026 | -16,997,409 | 18.29 | 2.50 | 10.60% | qualified |
| aggressive_leverage | 134,264,787 | 41,962,840 | -26,503,864 | -55,512,422 | 26.60 | 6.18 | 73.44% | disqualified: drawdown too large |
| defensive_trend | 41,864,063 | -11,676,729 | -9,441,990 | -25,947,094 | 13.19 | 1.05 | 1.26% | disqualified: negative P5 profit |
| panic_sell_rebuy | 22,837,804 | -9,578,609 | -13,380,435 | -19,105,389 | 10.72 | 0.71 | 0.16% | disqualified: negative P5 profit |
| passive_hold | 23,200,356 | -19,532,228 | -19,695,620 | -40,726,929 | 14.81 | 1.22 | 0.72% | disqualified: negative P5 profit |

## Adopted Strategy

Adopt:

`GUARDED AGGRESSIVE BARBELL`

This is the final default strategy for the current account-growth agent.

The raw-return winner was `aggressive_leverage`, but it was rejected because the P5 max drawdown was approximately -55,512,422 KRW in the 1-year, 10-year-calibrated run. That is too much pain for a strategy whose goal is repeated account-growth challenges, not bragging about average return.

The adopted strategy still uses leverage, but only inside a guarded structure.

## Operating Policy

### Mode 1: Attack

Use when:

- Market label is `HEALTHY RISK-ON`, `SELECTIVE MOMENTUM`, `TACTICAL REBOUND ACTIVE`, or `TREND REPAIR`.
- QQQ and SMH/SOXX confirm together.
- VIX is stable or falling.
- Leaders such as NVDA, AVGO, AMD, MU, TSM, ASML stabilize or lead.
- USD/KRW and current price are fresh.
- Main account orderable cash is available.
- User fatigue is not high.

Allowed:

- SOXL/TQQQ-style long leverage in main account only, normally 10-18% before upgrades.
- Tactical leader entries.
- 2-7 trading day swing only if daily revalidation is scheduled.

Default exposure guide:

- Quality/core: 25-35%.
- Leaders/high relative strength: 20-30%.
- SOXL/TQQQ tactical: 10-18% by default; 20%+ only for exceptional A-grade confirmation.
- Cash: 5-15%.
- High-beta legacy holdings: only if they still beat alternatives.

### Mode 2: Guarded Attack

Use when:

- Trend is improving but not fully confirmed.
- Rebound is possible but fragile.
- Crash risk is below 70 but not low.
- Cash is available but setup is not A-grade.

Allowed:

- Smaller SOXL/TQQQ entries.
- Staged entry, not all-in.
- Trim weak high-beta holdings into rebound.
- Keep 20-40% cash optionality.

Default exposure guide:

- Quality/core: 30-40%.
- Leaders: 10-25%.
- Leverage: 0-15%.
- Cash: 20-40%.

### Mode 3: Defense / SOXS Watch

Use when:

- Semiconductor breakdown is confirmed.
- QQQ/SMH/SOXX lose key levels with volume.
- Strong jobs/yields/Fed pressure is still active.
- Bellwether earnings reset the sector.
- Liquidity absorption or forced selling is active.
- Market flow is `DISTRIBUTION`, `BULL TRAP`, `LIQUIDITY AIR POCKET`, or `TREND BREAK`.

Allowed:

- SOXS/SQQQ only in main account.
- Cash-first defense.
- Partial tax-loss reset.
- Sell into rebound for weak high-beta names.

Default exposure guide:

- Quality/core: 20-35%.
- High-beta legacy holdings: 0-15%.
- SOXS/SQQQ: 5-20% only with stop, target, time stop.
- Cash: 40-70%.

### Mode 4: Rest / Cash

Use when:

- No edge.
- User fatigue is high.
- Data is stale.
- Broker cash is not orderable.
- Setup moved too far while waiting for cash.

Allowed:

- No new trade.
- Cash.
- Watchlist only.
- Prepare 10,000,000 KRW cash option if a first SOXL/SOXS signal may appear in 2-3 trading days.

## Hard Rules

- Never use sub account for leveraged or inverse ETFs.
- Never recommend SOXL/SOXS without target, stop, time stop, max loss KRW, and account eligibility.
- Never force sell an underwater holding to fund SOXL/SOXS unless net rotation edge is positive after cash delay and missed rebound risk.
- If the user is stressed or sleep-deprived, downgrade one mode automatically.
- If crash risk is 70+ and cash is not ready, do not chase late SOXS.
- If an A-grade opportunity becomes a missed window, mark `MISSED WINDOW` and move on.

## Why This Was Adopted

Pure passive holding performed worst because high-beta concentration creates large drawdowns without tactical control.

Panic sell/rebuy also performed poorly because it tends to lock losses and rebuy after the best rebound.

The best raw-return strategy was aggressive leverage, but it required a P5 drawdown of roughly -55.51M KRW from peak in the 1-year historical-regime simulation. That violates the challenge-first objective.

Therefore the adopted policy is guarded aggression: pursue large KRW net profit, but reject strategies whose downside path can damage execution, sleep, and family stability.

## Required Report Output Going Forward

Every major report should include:

- Current mode: `ATTACK`, `GUARDED ATTACK`, `DEFENSE`, or `REST`.
- Reason for the mode.
- Whether aggressive leverage engine is active or overridden.
- Whether the trade is still within the guarded aggressive barbell limits.
- SOXL/SOXS status.
- Main account orderable cash.
- Sub account restriction.
- Top 3 actions.
- Top 3 forbidden actions.
- Mode downgrade trigger.
- Mode upgrade trigger.