Pick a methodology. The besties' book moves with it.
5 presets run against the same six years of All-In Podcast calls (the 4 OG besties — Chamath, Sacks, Friedberg, Calacanis). We change the rules — stop losses, harvest, hold window, conviction floor, portfolio mode — and recompute what would have happened. Default is Balanced, the data-optimal cohort methodology (+143pp shared bucket). Try Let It Ride to see the besties' own "let your winners ride" mantra in action. The two Concentrated/Equal-Weighted presets are high-variance counterfactuals — useful for transparency, not recommendations.
Data-driven default. 18mo hold, -10% stop, gain-harvest at +50/+100, time-decay glide last 9mo. Wins for cohort books.
Methodology shape
Data-optimal default: moderate stop, medium hold, stepped gain harvest.
The cohort numbers above are the result of running the engine with this preset's exact parameters (stop loss, hold window, gain harvest schedule, conviction floor, portfolio mode). The recipe itself isn't published. Subscribers can see the full parameter readout and replicate the methodology end-to-end.