[02] Stress Testing Your Life — What Happens at -30%, -50%, -60%?
This is Part 2 of a 6-part series: Building Investment Systems with Python
Banks Do This Every Quarter. You Never Have.
After the 2008 financial crisis, regulators required banks to run stress tests — hypothetical scenarios where markets crash 30%, 40%, 60% — and prove they could survive.
Your personal balance sheet faces the same risks. If you hold a securities-backed loan, a market crash doesn't just reduce your wealth — it can trigger forced liquidation at the worst possible time.
Today we build the stress test engine. One function. Every scenario. The exact cash needed to survive each one.
The Core Engine
This reads from the ALM database we built in [Episode 01] and runs every drop scenario from -5% to -70%.
# stress_test.py
import sqlite3
from alm_schema import DB_PATH
def run_stress_test(cash_reserves=4_500_000):
conn = sqlite3.connect(DB_PATH)
c = conn.cursor()
# Current portfolio value
c.execute("""
SELECT SUM(h.shares * p.close_price)
FROM holdings h
JOIN prices p ON h.ticker = p.ticker
WHERE p.price_date = (SELECT MAX(price_date) FROM prices)
""")
portfolio_value = c.fetchone()[0]
# Margin loan details
c.execute("""
SELECT balance, freeze_pct, margin_call_pct, forced_liq_pct
FROM loans WHERE collateral_type = 'portfolio'
""")
loan = c.fetchone()
if not loan:
print("No margin loan found.")
return
loan_balance, freeze_pct, call_pct, liq_pct = loan
# Unsecured credit line
c.execute("SELECT balance FROM loans WHERE collateral_type IS NULL")
credit_row = c.fetchone()
credit_available = credit_row[0] if credit_row else 0
conn.close()
total_defense = cash_reserves + credit_available
print(f"Portfolio Value: ¥{portfolio_value:>14,.0f}")
print(f"Loan Balance: ¥{loan_balance:>14,.0f}")
print(f"Current Ratio: {loan_balance/portfolio_value:>12.1%}")
print(f"Cash Reserves: ¥{cash_reserves:>14,.0f}")
print(f"Credit Line: ¥{credit_available:>14,.0f}")
print(f"Total Defense: ¥{total_defense:>14,.0f}")
print()
print("=" * 90)
print(f"{'Drop':>6} {'Collateral':>14} {'Ratio':>8} {'Status':>16} "
f"{'Repay to 65%':>14} {'Repay to 85%':>14} {'Survive?':>10}")
print("=" * 90)
for drop in range(5, 75, 5):
pct = drop / 100
collateral = portfolio_value * (1 - pct)
ratio = loan_balance / collateral if collateral > 0 else float('inf')
# Determine status
if ratio > liq_pct:
status = "🔴 FORCED LIQ"
elif ratio > call_pct:
status = "🟠 MARGIN CALL"
elif ratio > freeze_pct:
status = "🟡 FROZEN"
else:
status = "🟢 OK"
# Cash needed to bring ratio back to safe levels
repay_to_65 = max(0, loan_balance - collateral * 0.65)
repay_to_85 = max(0, loan_balance - collateral * 0.85)
# Can we survive with available cash?
can_survive = "✅ YES" if repay_to_85 <= total_defense else "❌ NO"
print(f"{drop:>5}% ¥{collateral:>13,.0f} {ratio:>7.1%} {status:>16} "
f"¥{repay_to_65:>13,.0f} ¥{repay_to_85:>13,.0f} {can_survive:>10}")
# Find maximum survivable drawdown
print()
print("─" * 90)
for drop in range(1, 100):
collateral = portfolio_value * (1 - drop/100)
if collateral <= 0:
break
ratio = loan_balance / collateral
repay_to_85 = max(0, loan_balance - collateral * 0.85)
if repay_to_85 > total_defense:
print(f"⚡ Maximum survivable drawdown: -{drop-1}%")
print(f" At -{drop}%, you need ¥{repay_to_85:,.0f} but only have ¥{total_defense:,.0f}")
break
if __name__ == "__main__":
run_stress_test()
Sample Output
Portfolio Value: ¥ 125,135,000
Loan Balance: ¥ 50,000,000
Current Ratio: 39.9%
Cash Reserves: ¥ 4,500,000
Credit Line: ¥ 8,000,000
Total Defense: ¥ 12,500,000
==========================================================================================
Drop Collateral Ratio Status Repay to 65% Repay to 85% Survive?
==========================================================================================
5% ¥ 118,878,250 42.1% 🟢 OK ¥ 0 ¥ 0 ✅ YES
10% ¥ 112,621,500 44.4% 🟢 OK ¥ 0 ¥ 0 ✅ YES
15% ¥ 106,364,750 47.0% 🟢 OK ¥ 0 ¥ 0 ✅ YES
20% ¥ 100,108,000 49.9% 🟢 OK ¥ 0 ¥ 0 ✅ YES
25% ¥ 93,851,250 53.3% 🟢 OK ¥ 0 ¥ 0 ✅ YES
30% ¥ 87,594,500 57.1% 🟢 OK ¥ 0 ¥ 0 ✅ YES
35% ¥ 81,337,750 61.5% 🟡 FROZEN ¥ 1,130,463 ¥ 0 ✅ YES
40% ¥ 75,081,000 66.6% 🟡 FROZEN ¥ 1,197,350 ¥ 0 ✅ YES
45% ¥ 68,824,250 72.6% 🟠 MARGIN CALL ¥ 5,264,238 ¥ 0 ✅ YES
50% ¥ 62,567,500 79.9% 🟠 MARGIN CALL ¥ 9,331,125 ¥ 0 ✅ YES
55% ¥ 56,310,750 88.8% 🔴 FORCED LIQ ¥ 13,397,963 ¥ 2,135,863 ✅ YES
60% ¥ 50,054,000 99.9% 🔴 FORCED LIQ ¥ 17,464,900 ¥ 7,454,100 ✅ YES
65% ¥ 43,797,250 114.2% 🔴 FORCED LIQ ¥ 21,531,788 ¥ 12,772,338 ❌ NO
70% ¥ 37,540,500 133.2% 🔴 FORCED LIQ ¥ 25,598,675 ¥ 18,090,575 ❌ NO
──────────────────────────────────────────────────────────────────────────────────────────
⚡ Maximum survivable drawdown: -62%
At -63%, you need ¥13,101,398 but only have ¥12,500,000
One table. Your entire risk profile. The number that lets you sleep at night.
Visualizing the Danger Zones
# stress_chart.py
import matplotlib.pyplot as plt
import matplotlib.patches as mpatches
import sqlite3
from alm_schema import DB_PATH
def plot_stress_test():
conn = sqlite3.connect(DB_PATH)
c = conn.cursor()
c.execute("""
SELECT SUM(h.shares * p.close_price)
FROM holdings h
JOIN prices p ON h.ticker = p.ticker
WHERE p.price_date = (SELECT MAX(price_date) FROM prices)
""")
portfolio = c.fetchone()[0]
c.execute("SELECT balance, freeze_pct, margin_call_pct, forced_liq_pct FROM loans WHERE collateral_type = 'portfolio'")
balance, freeze, call, liq = c.fetchone()
conn.close()
drops = list(range(0, 75))
ratios = []
for d in drops:
collateral = portfolio * (1 - d/100)
ratios.append(balance / collateral * 100 if collateral > 0 else 100)
fig, ax = plt.subplots(figsize=(12, 6))
fig.patch.set_facecolor('#0a0a0a')
ax.set_facecolor('#0a0a0a')
# Danger zones
ax.axhspan(0, freeze*100, alpha=0.15, color='#00ff88', label=f'Safe (<{freeze:.0%})')
ax.axhspan(freeze*100, call*100, alpha=0.15, color='#ffaa00', label=f'Frozen ({freeze:.0%}-{call:.0%})')
ax.axhspan(call*100, liq*100, alpha=0.15, color='#ff6600', label=f'Margin Call ({call:.0%}-{liq:.0%})')
ax.axhspan(liq*100, 120, alpha=0.15, color='#ff0000', label=f'Forced Liquidation (>{liq:.0%})')
# Ratio line
ax.plot(drops, ratios, color='#00ff88', linewidth=2.5, zorder=5)
# Threshold lines
for pct, color, style in [(freeze*100, '#ffaa00', '--'), (call*100, '#ff6600', '--'), (liq*100, '#ff0000', '-')]:
ax.axhline(y=pct, color=color, linestyle=style, alpha=0.7, linewidth=1)
ax.set_xlabel('Portfolio Drawdown (%)', color='#888', fontsize=12)
ax.set_ylabel('Margin Ratio (%)', color='#888', fontsize=12)
ax.set_title('Stress Test: Margin Ratio vs. Drawdown', color='white', fontsize=14, pad=20)
ax.tick_params(colors='#888')
ax.spines['bottom'].set_color('#333')
ax.spines['left'].set_color('#333')
ax.spines['top'].set_visible(False)
ax.spines['right'].set_visible(False)
ax.legend(loc='upper left', facecolor='#111', edgecolor='#333', labelcolor='#ccc')
ax.set_xlim(0, 74)
ax.set_ylim(30, 120)
plt.tight_layout()
plt.savefig('stress_test.png', dpi=150, facecolor='#0a0a0a')
print("Chart saved: stress_test.png")
if __name__ == "__main__":
plot_stress_test()
This produces a single chart that maps your margin ratio against every possible drawdown level, with color-coded danger zones.
Historical Context
How realistic are these scenarios? Here's what the data says for the Nikkei 225:
| Event | Period | Peak-to-Trough |
|---|---|---|
| Bubble Collapse | 1989 → 2003 | -81% (over 13 years) |
| IT Bubble + Banking Crisis | 2000 → 2003 | -64% (over 3 years) |
| Lehman Shock | 2007 → 2008 | -62% (over 17 months) |
| COVID Crash | Jan → Mar 2020 | -31% (over 2 months) |
| Aug 2024 Crash | Jul → Aug 2024 | -27% (over 3 weeks) |
The -60% scenarios happened twice in modern history — and both took years, not days. That's critical: you have time to react. The stress test tells you the thresholds; the timeline tells you there's usually a window to repay.
The Key Insight
The stress test reveals something counterintuitive:
A lower loan balance doesn't always make you safer — more available cash does.
Compare:
- Loan ¥40M, Cash ¥1M → forced liquidation at -55%, no cash to repay
- Loan ¥50M, Cash ¥12.5M → forced liquidation at -53%, but cash covers repayment through -62%
The second scenario has more debt but higher survivability because the defense fund is orthogonal to the collateral.
This is why the "just pay down debt" advice can be wrong. It depends on the structure.
What We Built
- A stress test engine that models every drawdown from -5% to -70%
- Automatic calculation of repayment amounts for each danger zone
- Maximum survivable drawdown computation
- A matplotlib visualization of margin ratio vs. drawdown with danger zones
Next week: [03] Designing a Personal Commitment Line — "Two loans, one defense system."
Series: Building Investment Systems with Python — Engineering financial independence with code.