AMD bull put spread: credit, risk, strikes

You are still selling downside — just not all of it. The long $550 put cuts the tail off below that level, which is why this needs $2,883 of buying power instead of the $59,000 a cash-secured put would tie up.

Above $590 at October 30, 2026, both puts expire worthless and you keep the full $1,117. Below $550, you lose the maximum $2,883. Breakeven is $578.83.

Return on risk is 39% for 33 days — 429% annualized. That headline is the reason people prefer spreads to cash-secured puts, and the reason spreads blow up accounts: the same capital supports several times the notional risk.

The risk is leverage, not the structure. $2,883 per spread is small; the temptation to sell ten of them because the buying power allows it is how a 68%-win-rate trade produces a losing year.

Between the strikes the loss scales linearly, so most of the damage happens fast when AMD breaks $590. There is no assignment-and-hold escape hatch: the long put you own expires the same day.

Liquidity is a risk, not a convenience. The moment you most want out of a short-premium position is the moment the spread is widest, and the exit price you modelled at mid will not be available.

The correlation is the risk nobody prices. Short premium on AMD and short premium on NVDA is one position with two tickets, and it gets tested on the same afternoon — AMD moves on NVDA's guidance as reliably as on its own. If both are in the book, size them as a single semi-cycle exposure and halve each.

AMD's Oct 30 strikes are $10 apart near the money (1.59% of spot). That is a coarse ladder: one rung is a large fraction of the implied move, so precision on the short strike is an illusion. 12k contracts of open interest on Oct 30 is thin, and a structure that needs four separate fills will pay for it. 41 strikes on that expiry — 100% of the board — carry prints that agree with their own implied volatility and hold up across the ladder, and those are the strikes priced here. Good depth around the money, thinner in the wings than NVDA's; keep condor widths inside where the open interest actually is.

Skew is inverted: the 25-delta CALL implies 1.6% more vol than the put. That is the market pricing upside risk above downside risk — a squeeze, a takeover rumour, or a crowded short. Selling calls into an inverted skew pays better than usual and is riskier than usual for exactly the same reason. The term structure is in contango: the back month implies 4.5% more vol than the front. Calm now, uncertainty later — which rewards selling the front month and makes the back month an expensive thing to own outright.

At 55% ATM implied vol, the Oct 30 options are pricing a one-standard-deviation move of $104.08 over 33 days — roughly −16.5% to +16.5%, or $526.55 to $734.71. The structure above sells the part of that distribution the market thinks it will not reach. Whether that is a good trade is entirely a question of whether 16.5% is too much or too little for AMD over 33 days — the delta table cannot answer that, and neither can we.

The specific way people lose money on AMD: Selling AMD premium as a diversifier from an NVDA position. It is the same trade at a higher beta.