The standard framework for position sizing — risk a fixed percentage of capital per trade — is built on an assumption that is almost never stated explicitly and almost never true in practice. The assumption is that each trade’s outcome is independent of every other trade’s outcome. If that assumption holds, the math works cleanly: five positions at one percent each produce one percent losses when any individual trade stops out, five percent when all five stop out simultaneously, and the probability of the latter is the product of the individual stop probabilities.
When the assumption breaks down — when positions move together because they are driven by the same underlying factor — the math changes entirely. Five correlated positions at one percent each can produce five percent losses not just when all five are wrong, but whenever the shared underlying factor moves against them. The individual position limits mean nothing if the positions themselves are not independent. The risk model l…