Risk & Portfolio

Seven positions, the largest at 30% and 20%. Capping each at 15% and spreading the excess over the others pro rata gives 15, 15, 15, 15, 15, 13.3 and 11.7%. If the top name then halves, the portfolio loses 7.5% instead of 15%. The cap does not make the picks better; it bounds what one mistake can cost.

• Caps on single names miss correlated exposure: seven positions in one sector are one position.
• Limits on market value miss leverage and derivatives; Archegos built huge exposure through swaps that did not show up as holdings.
• Rules like UCITS 5/10/40 and the U.S. 75-5-10 test set minimums for funds, not good practice for every portfolio.

w = pd.Series({'A': 0.30, 'B': 0.20, 'C': 0.15, 'D': 0.10, 'E': 0.10, 'F': 0.08, 'G': 0.07})
cap = 0.15
while (w > cap + 1e-12).any():                             # trim to the cap, share the excess pro rata
    excess = (w - cap).clip(lower=0).sum()
    w = w.clip(upper=cap)
    room = w < cap
    w[room] += excess * w[room] / w[room].sum()
hit_before = 0.30 * 0.5                                    # the top name halves
hit_after = w.max() * 0.5

Assumes import numpy as np and import pandas as pd. Each snippet carries its own example numbers; the comments say which are made up or simulated.