Having reconstituted some backtests in Excel, it is clear that there are some backtests which have improved performance (similar CAGR, lower DD) if position-sizing is NOT based on 'margin allocation' but on the basis of '1 contract per x$' of capital. e.g.: sell a 50 delta put, with a 25-wide long below if this is entered at 3.30, then perhaps the margin requirement / max. loss is 2,000 if this is entered at 9:35 on a day when VIX is at 50, then perhaps the margin requirement is only 500 - leading to a (disproportionately) high number of contracts being traded but what if we could enter 1 contract per 20k$ of capital? As noted above, this may often smooth the profit curve. There are many other scenarios where this approach would be helpful
