Replicating Evans & Archer (1968)
How many stocks does it actually take to diversify? A simulation study.
A project led under Dr. Daniel Rodriguez through the FinTech Club, replicating a classic finance result from first principles.
The original claim
Evans and Archer (1968) asked a deceptively simple question: as you add randomly-chosen stocks to a portfolio, how fast does risk fall? The answer shapes how every retail investor should think about their holdings.
What we did
Built a Monte Carlo simulation in Python (Pandas, NumPy, SciPy) over historical price data: repeatedly draw random portfolios of size $n$, compute the standard deviation of returns, and trace how that standard deviation behaves as $n$ grows.
The result
Risk falls steeply at first and then flattens hard. Most unsystematic risk is gone by 10–20 assets, and past that, adding names buys you very little. The remaining variance is systematic: market risk you cannot diversify away, only choose to bear.
Why replicate something already known
Because replicating a result you expect is the cheapest possible way to learn whether your methodology is sound. If the simulation hadn't reproduced the 1968 curve, the bug would have been mine, and I'd rather discover that on a known answer than on an unknown one.
It's the same instinct that shows up in my execution research: build the control that could prove you wrong, and run it first.