Some takeaways...
Every source of uncertainty should reduce position size
Large positions require disproportionately better information. Kelly is nonlinear. Going from 5% to 15% isn't simply "three times more conviction." It’s like ~6 times more conviction.
Your biggest risk probably isn't misjudging an individual company, it's overestimating how independent your best ideas really are.
Kelly is an upper bound. Kelly's growth curve is surprisingly flat below the optimum but falls off rapidly above it. A modest reduction in size sacrifices little long-term compounding, while overbetting can dramatically impair it.
Improvements in position sizing can contribute more to long-term wealth creation than incremental improvements in security selection.