I Asked Claude to Invest $10,000. Then I Changed the Starting Date

Description and use

Demonstrates how entry date choice can dramatically alter investment outcomes using a rolling‑window backtest to assess return distribution.

Problem

Investing a fixed amount at a single start date can give misleading performance results because the outcome depends heavily on the chosen entry point.

Outcome

The rolling‑window backtest showed that the same $10,000 allocation could have yielded $19,135 in three years if started in January 2019, but only $12,382 if started in January 2022—a $6,753 difference driven solely by entry timing. The distribution of returns revealed the strategy’s true risk, benchmark performance, and the importance of considering multiple start dates.

Workflow

• Define four portfolios (Claude, Magnificent 7, S&P 500, 60/40).
• Collect adjusted close prices for all tickers via EODHD API.
• Build a function to download price data for a ticker.
• Create a DataFrame of prices for all tickers, dropping missing values.
• Simulate portfolio performance for a given start date and 3‑year horizon, rebalancing annually.
• Loop over all 105 start dates, recording final portfolio values.
• Compare distributions against benchmarks and visualize results.

Evidence

Same portfolio, same $10,000, but instead of one start date I tested 105 of them: every month from January 2015 to September 2023, each one held for three years.

Source

If you want to reproduce this, EODHD's free tier is enough to start.

Source

$6,753 of difference, and none of it came from the portfolio.

Source