S-B SPY put spread

Paper trading, not investment advice. Every account and book on these pages trades paper money: the Alpaca paper accounts DAILY ($1,000,000 from Sep 30, 2026), INTRA1 ($100,000 from Sep 30, 2026) and INTRA2 ($100,000 from Sep 30, 2026); the simulated books SIM-TT ($100,000 from Sep 29, 2026), SIM-V2B ($100,000 from Sep 28, 2026) and SIM-V2B-TOM ($100,000 from Sep 28, 2026), filled at the official close; and market-analysis's ledgers, one contract per position against a $100,000 reporting equity. No real money is at stake. Positions, trades and P&L are published the same day, as the engines record them, with each number's source and cost model labelled. Nothing here is a recommendation to buy, sell or hold any security, and nothing is tailored to you. Each strategy is a pre-registered experiment, read once on a date or at a trade count fixed in advance; until then its record says little either way. I may hold or have held positions in securities shown here. Markets can and do lose money.

Sells a SPY put credit spread about three weeks out, once a week, when the volatility gate is on: VIX's term structure sloping up and VIX above its recent median. This is the champion book, the rule as frozen before any result; its exploration twin ignores the gate. The put spread is S-B's primary structure.

Gate: VIX3M above VIX (G1) and VIX above its median of the 20 sessions before (G2), both on the prior close; a missing value means no entry.
Entry: the last session of the week, each leg priced at its average traded price from 15:00 to 16:00 ET; of the expiries that traded that day, the Friday one (the Thursday before a Friday holiday) nearest 21 calendar days out, between 7 and 30.
Strikes: short put at S(1 − m), long put at S(1 − 2m), with S SPY's price at the day's last option print and m = VIX/100 × √(DTE/365), DTE the calendar days to expiry and VIX at that day's close; each the strike nearest its target, within 0.25m × S of it, that printed at least 5 contracts from 15:00. If a leg has no such strike in that expiry, there is no entry that week. S and m are taken at the end of the day, up to an hour after the prices the legs are entered at; S-B's specification accepts this gap.
Spread: the short put's quoted spread at its prints from 15:00, averaged by size, at most 10% of the quote's midpoint; otherwise, or with no quote, no entry that week. The long put's is charged, not filtered.
Size: max loss per contract is the width less the credit. As many contracts as keep the max loss within 3% of $100,000, but never fewer than one, so one contract can risk more; at most 3 open.
Exit: held to expiry and settled at intrinsic value against the official close (or SPY's price at that day's last option print if the close is missing). Early assignment and dividends are not modelled.

Name	Value	Unit	Where it lives	Pinned by
target_dte_cal	21	calendar days	engine/config.py:125	tests/test_sb_frozen_params.py:31
dte_cal_min	7	calendar days	engine/config.py:126	tests/test_sb_frozen_params.py:31
dte_cal_max	30	calendar days	engine/config.py:127	tests/test_sb_frozen_params.py:31
short_sigma	1	m, the short leg's distance from S	engine/config.py:128	tests/test_sb_frozen_params.py:31
wing_sigma	2	m, the long leg's distance from S	engine/config.py:129	tests/test_sb_frozen_params.py:31
strike_band_sigma	0.25	m × S, the most a strike may sit from its target	engine/config.py:130	tests/test_sb_frozen_params.py:31
leg_size_min	5	contracts printed from 15:00, each leg	engine/config.py:131	tests/test_sb_frozen_params.py:31
short_spread_max	0.1	of the quote's midpoint, short leg	engine/config.py:132	tests/test_sb_frozen_params.py:31
median_window	20	sessions	engine/config.py:133	tests/test_sb_frozen_params.py:31
equity	100000	USD	engine/config.py:140	tests/test_sb_frozen_params.py:31
max_loss_frac	0.03	of equity	engine/config.py:141	tests/test_sb_frozen_params.py:31
max_open_per_sleeve	3	open positions	engine/config.py:142	tests/test_sb_frozen_params.py:31
LATE_START	15:00:00	ET, the late window's start; it runs to the close	engine/config.py:42	tests/test_frozen_params.py:39
COMMISSION_PER_CONTRACT	0.65	USD a contract, each leg	engine/config.py:55	tests/test_frozen_params.py:39
SB_SCALE_MIN_POSITIONS	40	graded paper positions before more than one contract	engine/config.py:176	tests/test_sb_frozen_params.py:47

What is read: the go or no-go on S-B's bar for this book, fixed before any result: a three-year backtest priced by model and one priced from real option prints since March 2026, on entries through 2026-11-06. A go puts one contract live from the first gate-on week after the read. The paper record shown here is not part of this go or no-go; it sets when a live book may trade more than one contract, at 40 graded positions with a positive mean.
Read on: Dec 1, 2026
Trades needed: —
Trades logged: —
Hard stop: —
Shipped on: —
Deployed commit: 6e25a3f

Closed trades: 1
Won: 1 of 1
P&L: +$159.73
Net P&L: +$155.40
Closed without a P&L: —
Closed without a net P&L: —
Open: 2

Record GitHub ledger

Cost model

half of each leg's quoted spread plus $0.65 a contract at entry; $0.65 a contract at expiry

Opened	Closed	Account	Symbol	Side	Size	Entry	Exit	P&L	Net P&L	Return	Held	Exit by	Status	Record
Oct 2, 2026	—	MA	SPY	short	1	1.208684	—	—	—	—	—	—	open	GitHub ledger (1)
Sep 25, 2026	—	MA	SPY	short	1	1.325587	—	—	—	—	—	—	open	GitHub ledger (1)
Sep 11, 2026	Oct 2, 2026	MA	SPY	short	1	1.597298	0.00	+$159.73	+$155.40	—	21 days	expiry	closed	GitHub ledger (1)

(1)
Record GitHub ledger
Cost model
half of each leg's quoted spread plus $0.65 a contract at entry; $0.65 a contract at expiry

market-analysis
S-B's specification (findings market-analysis/DESIGN/80)

What it is

The rule

Its read

Record so far

Trades

Where the research is