One normalized position per night, naked, real Tardis ledger JanโMay 2026 (limit fills at mid;
exec column = full cross fills for honesty). Companion to the overnight report.
Verdict: yes โ ~3% OTM beats ATM on risk-adjusted basis. Win rate rises monotonically with distance
(put: 81โ89โ96โ100%), MaxDD shrinks faster than premium does โ r/DD peaks at 3% OTM. With a fixed risk budget you
simply sell MORE units at 3% for the same tail. ATM earns more raw bps per contract but carries 2โ3ร the drawdown.
5% OTM is past the cliff โ no premium left. Touch-statistics agree: a 3% strike is touched in only 3โ4% of nights
and settles ITM in 1โ2% (stable 2024โ2026).
โ ๏ธ Caveats: 4 months real data, one high-IV bear (OTM premiums were fat โ in calm regimes the 3%-strike premium
thins toward the 5% cliff, so the sweet spot may sit closer to 2โ2.5%). Limit fills assumed; exec column shows the
cross-fill reality โ the further OTM, the more the spread matters. ETH-call rows benefit from the bear regime
(see the asymmetric call rules on weekly_research).
Source: real ledger 0dte_sell_trades.json (41,860 trades) ยท build_strike_compare_page.py.