Open Claude Open ChatGPT

Chat with the numbers

Financial simulations inside ChatGPT and Claude. Ask about retirement, sabbaticals, or big purchases — get a range of outcomes, not a single guess.

Quant-grade simulations Free and open source We never see your data

Chat flow

Ask a question in ChatGPT. The response includes a compact, interactive widget.

Can I retire at 50? I'm 35, make $150k, spend $50k, saved $800k.
Chubby Chubby
Plan Coverage
This plan stays funded through Age 80
(end of simulation)
Typical assets at that age: ~$4.8M
Starting Point
Age35
Assets$800K
Income$150K
Spending$50K
Scheduled Events
🏖️Age 50Early Retirement$0
🏛️Age 62Social Security+$28K/yr
🏥Age 65Medicare−$4K/yr
What Happens to Your Money
35
$800K
40
$1.2M
50
$2.8M
60
$3.6M
70
$4.1M
80
$4.8M
P10 → P50 Median
through Age 80 100 paths

Simulation summary

I ran Chubby simulations and summarized the trajectory range — P10, P50, and P75 — across 100 paths.

Outcome highlights

  • P50 retirement age: 51 (P10: 54 · P75: 48).
  • Primary constraint: sequence‑of‑returns risk in the first decade.
  • Typical assets at age 80: ~$4.8M (median path).

Considerations / tradeoffs

  • Retiring at 50 looks viable in median outcomes — but early drawdowns matter.
  • Lower spending by ~$5–10k improves the P10 case materially.
  • Adding 12–18 months of cash buffer reduces downside paths.

Answer

Based on your assumptions: potentially yes — if you can keep spending stable and tolerate early‑year volatility. If you want, we can stress‑test lower returns, higher spending, or a first‑year drawdown.

What if I spend $60k instead of $50k?
ChatGPT can make mistakes. Check important info.

Try these

Copy a prompt and paste it into ChatGPT or Claude.

How it works

Each simulation runs 100 paths through a discrete-event engine with realistic market dynamics.

Volatility

GARCH(1,1) with Student-t innovations. Crashes cluster; tails stay fat.

σ²(t) = ω + α·ε²(t-1) + β·σ²(t-1)
Correlation

8-asset Cholesky-decomposed correlation matrix. Assets move together in crashes.

Σ = L·Lᵀ → r = L·z
Events

Priority-sorted queue: income, expenses, contributions, withdrawals, taxes, markets.

Income(25) → Expense(60) → Market(110)
Aggregation

100 paths aggregated into percentiles. P10 = harder markets, P50 = median, P75 = favorable.

quantiles(paths) → P10 / P50 / P75

Privacy

Widget links use URL fragments — the part after # that browsers never send to servers. Your simulation data stays in your browser.

widget.chubby.fyi/#d=eJxLTEo... ← never sent to servers

No accounts, no sessions, no persistence. Same seed + same inputs = identical output. Results are deterministic and reproducible.