The BTC iron condor ladder
Sell one small, defined-risk iron condor on BTC most days when implied volatility is rich, hold each to expiry, and spread the account's risk across many strikes and entry days.
- Target APY
- 15–40%
- Risk
- Elevated
- Complexity
- Advanced
- Min. capital
- $1k–$10k
- Where it runs
- DeFi
- Status
- Active
Published Nov 17, 2026. 10 min read, 18 min to listen.
What a year of selling condors looks like
Nothing cancels here the way a hedged pair does. One line is every credit collected, about four a week; the other is what goes back out at expiry on the weeks price left the short strikes. The net line is the difference, and it spends the year climbing in small steps and falling in a few large ones. Drag across the chart to see any day.
credit +130.6%settlement −75.4%costs −18.6%
| Day | Credit collected | Settlement paid | Net: account equity |
|---|---|---|---|
| 0 | +0.0% | +0.0% | +0.0% |
| 91 | +32.0% | −6.0% | +21.4% |
| 182 | +65.3% | −34.2% | +21.8% |
| 273 | +100.6% | −58.4% | +27.9% |
| 365 | +130.6% | −75.4% | +36.6% |
An iron condor sells one out-of-the-money put spread and one out-of-the-money call spread on the same expiry. You collect a credit up front, and your worst case is fixed the moment you open it: the width of the wider spread minus that credit. The trade makes money when the market stays inside the two short strikes until expiry.
The "ladder" is the part that matters more than the structure. Instead of putting one large condor on each week, you open one small condor on each of five days a week, all into the next Friday expiry. No single entry day and no single price level decides the year's result.
The version documented here runs on BTC options settled in USDC, uses Deribit's BTC DVOL index as its volatility filter, and was built as a written specification with a backtest behind it. It has been paper-traded, not run live. Everything below is that specification and its modelled results, not a track record.
How it works
| Instrument | European, cash-settled BTC options. Expiry and settlement at 08:00 UTC. |
| When | Once a day, just after settlement, Tuesday to Saturday. Sunday and Monday have no Friday expiry in the target window, so there is no trade. |
| Volatility filter | Only trade if the latest hourly Deribit BTC DVOL is 38 or above. In the backtest this skipped about a quarter of eligible days. |
| Expiry | The Friday weekly expiry that is 6 to 10 days away. Tuesday sells 10 days, Saturday sells 6, all into the same following Friday. |
| Short strikes | The put nearest −0.25 delta and the call nearest +0.25 delta. On average these sat about 4% below and 5% above spot. |
| Wings | Buy the put $4,000 below the short put and the call $4,000 above the short call, or the nearest listed strike further out. |
| Size | Contracts per leg = risk % × account equity ÷ (wing width − credit per 1 BTC). Risk % is 1.5% as a base case, 2% as the aggressive setting. |
| Risk cap | Total max loss across all open condors may not exceed 20% of equity. |
| Exit | Hold to expiry. No stop-loss, no profit target, no adjustments. Settlement is automatic. |
| Execution | Limit orders at or near mid. Never cross the full spread — costs are the largest controllable drag on the result. |
| Collateral | Stablecoin only. Posting BTC as collateral would bolt an unhedged long onto a short-volatility position. |
Five entry days into one weekly expiry means five to eight condors are usually open at once, each at different strikes, each sized off the equity at the moment it was opened. Positions grow as the account grows and shrink after losses, automatically.
Where the yield comes from
The same place as every other short-volatility trade: the variance risk premium. Option buyers pay for protection and convexity, and on average they pay more than the volatility that follows turns out to be worth.
Over the period studied (1 January 2025 to 31 October 2026), Deribit's 30-day BTC implied volatility index averaged 44.3, while volatility actually realized over the following two weeks averaged 41.6. Implied exceeded realized about 71% of the time. An iron condor is one way to sell that spread with a worst case you know in advance.
That gap is the entire edge. It is not a yield in the sense that staking is a yield — there is no cash flow and no counterparty paying you to hold something. You are being paid to carry the risk of a large move, and sometimes that move happens.
Worked example
Illustrative, from quotes taken on 2 November 2026 with BTC near $77,878 and DVOL at 38.5. Strikes, prices and spreads will be different whenever you read this.
| Leg | Strike | Delta | Mark (per 1 BTC) |
|---|---|---|---|
| Sell put | 75,000 | −0.27 | $934 |
| Buy put | 70,000 | −0.08 | $244 |
| Sell call | 82,000 | +0.22 | $596 |
| Buy call | 86,000 | +0.07 | $151 |
The $71,000 put was not listed, so the put wing came out $5,000 wide instead of $4,000. Credit at mid is ($934 − $244) + ($596 − $151) = $1,135 per 1 BTC. Crossing every spread instead would have left $995. Max loss is the wider wing minus the credit: $5,000 − $1,135 = $3,865 per 1 BTC.
On a $5,000 account:
| 1.5% risk | 2% risk | |
|---|---|---|
| Risk budget per condor | $75 | $100 |
| Size per leg | 0.019 BTC | 0.025 BTC |
| Credit received | $21.57 | $28.38 |
| Max loss | $73.43 | $96.63 |
Those are small numbers, and that is the point. The account's return comes from putting twenty-odd of these on per month, not from any one of them.
What the backtest showed
$5,000 start, compounding, every trading day from 1 January 2025 to 31 October 2026, with fees and a half-spread charged on entry — limit orders filled between mid and the touch.
| 1.5% risk | 2% risk | |
|---|---|---|
| Ending equity | $8,859 | $10,538 |
| Total return / annualised | +77% / 36.6% | +111% / 50.2% |
| Max drawdown | −17.3% | −22.6% |
| Sharpe / Sortino | 1.65 / 2.09 | 1.65 / 2.09 |
| Trades / win rate | 351 / 74.9% | 351 / 74.9% |
| Average win / average loss | +$36.9 / −$66.6 | +$53.1 / −$95.9 |
| Profit factor | 1.66 | 1.66 |
| Months positive | 17 of 22 | 17 of 22 |
| Worst / best month | −13.3% / +13.8% | −17.4% / +18.6% |
A 75% win rate with losses roughly twice the size of wins is the normal shape of a short-volatility book. You are right most of the time and wrong expensively.
Costs decide a lot of it. The same rules, priced three ways:
| Cost assumption | Annualised (1.5%) | Max drawdown | Sharpe |
|---|---|---|---|
| Cross the full half-spread | 24.6% | −19.0% | 1.19 |
| Limit orders near mid (base case) | 36.6% | −17.3% | 1.65 |
| Zero costs (upper bound) | 55.2% | −15.7% | 2.29 |
Twelve points of annual return sit between patient limit orders and market orders. Execution is not a detail here.
The losses cluster. Every large losing trade in the test was a 9–15% move inside a single week, and when it happened, several open condors on that expiry lost together. Two of the five worst trades shared the same expiry.
One path is not a distribution. Resampling the weekly results into 5,000 random 52-week sequences gives a range for a normal year:
| Outcome over one year | 1.5% risk | 2% risk |
|---|---|---|
| Bad year (1 in 20) | +1% | 0% |
| Typical year (median) | +47% | +65% |
| Good year (1 in 20) | +105% | +159% |
| Chance of a losing year | 4.5% | 5.1% |
| Typical worst drawdown within the year | −13% | −17% |
| Bad-case drawdown (1 in 20) | −25% | −33% |
Resampling only reshuffles weeks that already happened. It cannot produce a regime the test period did not contain — a sustained 40% crash with collapsing liquidity, for example. Treat the table as the range under normal conditions and size for the bad case, not the median.
Plan on less than any of this. The option prices in the backtest were modelled from DVOL with an assumed skew rather than taken from historical venue quotes, and 22 months is a short sample that happened to be friendly to selling BTC volatility. A prudent planning range is roughly 15–30% a year at 1.5% risk and 20–40% at 2%, with at least one 20–30% drawdown along the way. The yield range on this page is that planning range, not the backtest.
Why these settings
Each rule was tested against the obvious alternatives over the same period.
| Decision | What the alternatives did |
|---|---|
| Hold to expiry | Selling 14-day condors and closing after 7 days lost money in every quarter: the first week holds only about 10% of the credit while carrying the full risk, and closing four legs weekly doubles costs. |
| No stop-loss | Stops at 1×, 1.5×, 2× and 3× credit, strike-touch stops and delta stops all reduced profit after costs. Short-dated BTC pushes through a strike and comes back before expiry often enough that stops fired on 35–70% of trades. |
| DVOL ≥ 38 | No filter: Sharpe 1.58 vs 2.04, and a much worse bad case. Filters based on implied being above trailing realized vol roughly halved returns — implied tops realized after quiet spells, which often end in breakouts. |
| 25 delta | 15 delta: credit too thin to cover costs. 35 delta: deepest drawdowns. 25 and 30 were close; 25 had the higher win rate and smaller worst trade. |
| Daily ladder | With one fixed entry day a week, results depended on which day: Monday and Thursday entries differed by about 2× under identical rules. |
| Skip Sunday and Monday | Selling the 11–12 day expiry on those days raised peak open risk from 13% to 20% of equity and cut Sharpe from 1.65 to 1.36 by concentrating risk in one expiry. |
| $4,000 wings | $2,000 wings lost in every variant — costs eat the small credit. Wider wings earn more but the worst trade grows in proportion. |
Running it
This is a bot strategy. The daily sequence is short but every step has to fail closed.
- Reconcile open positions against the exchange and your own journal, mark anything past expiry as settled, and read current equity. If the two disagree, stop and alert.
- Calendar gate. Find the Friday expiry 6–10 days out. If there is none, or a condor was already opened today, stop.
- Volatility gate. Fetch DVOL. Below 38, or the data is stale or the request failed, stop — no trade that day.
- Select strikes by delta from the live chain, then the wings at least $4,000 out. If a wing strike does not exist, skip the day rather than widen.
- Size from risk % of equity and the remaining room under the 20% open-risk cap, rounded down to the venue's increment.
- Execute wings first, then the shorts, so the account is never short an uncovered option. Start at mid and step toward the touch every 60 seconds, stopping once the expected credit would fall below the natural price. If the shorts are not both filled inside 15 minutes, cancel and sell back any filled wings — never hold one side.
- Record strikes, size, fills, credit, max loss, DVOL and spot. Then do nothing until settlement.
Additional guards worth writing into the bot: pause new entries if equity falls 30% below its peak and review before resuming; keep at least 2× total open max loss in free collateral; log every skipped day with its reason.
Risks
Trend weeks. A 9–15% move in a few days puts every open condor on one expiry at a loss simultaneously. Expect a −10% to −17% month a couple of times a year.
The model is not the market. The backtest priced options from a volatility index plus an assumed smile, not from historical quotes on the venue. A spot check against live mid prices matched within about 4%, but far wings are exactly where that kind of model is least reliable.
Liquidity. Weekly BTC options on a smaller venue can show thin open interest, with only a fraction of strikes quoted two-sided. Fills may be slower and wider than modelled, and that difference showed up as 12 points of annual return in the cost sensitivity table.
Minimum order size. Venues impose a minimum notional per order. At a few hundred dollars of premium per leg on a $5,000 account, the long wings are small enough that this can bind — confirm how the minimum is measured before assuming the size table works.
Venue and settlement risk. Settlement is against the venue's own oracle. Downtime at entry costs you a day, which is acceptable; the real exposure is custody and solvency, as with any exchange.
Short volatility is not a yield. There is no coupon here. The strategy is being paid to absorb the risk of a large move, and a single bad week can take back several good months.
When it stops working
- Implied stops exceeding realized. This is the whole edge. Track average P&L per trade live; if it is negative over 100-plus trades, stop and review rather than assuming variance.
- Costs rise. If spreads widen or your fills drift from mid toward the touch, the market-order row of the sensitivity table is where your returns go.
- Liquidity thins further. If the 25-delta strikes or the wings stop being quoted reliably, the strategy cannot be executed as specified, and widening the wings to get filled changes the risk you signed up for.
- A regime the test never saw. A sustained crash with collapsing liquidity is outside the resampled range above. The 30%-below-peak pause exists for exactly that, and it was never tested — it is a circuit breaker, not a backtested rule.
- You start adjusting. Every discretionary intervention tested here made results worse. If you find yourself managing positions mid-week, you are running a different strategy than the one that was measured.
Key takeaways
- The edge is the variance risk premium: implied volatility on BTC options has on average exceeded the volatility that followed. The condor is just a defined-risk way to sell it.
- Laddering daily entries into one weekly expiry removes entry-day luck, but it does not remove correlation. Everything open loses together in a trend week.
- Holding to expiry and not using stops both tested better than the alternatives, because short-dated BTC pushes through strikes and comes back often enough to make stops expensive.
- Execution is worth more than most of the parameter choices: limit orders near mid versus crossing the spread was a 12-point difference in annual return.
- Plan on 15–30% a year at 1.5% risk with a 20–30% drawdown, not on the backtest's 36.6%. The modelled numbers are an upper bound on a short and friendly sample, and the specification has been paper-traded, not run live.
Found this useful? Share it.
Related strategies
- 8–25%target APY
Covered calls on BTC and ETH
Hold BTC or ETH and sell out-of-the-money calls against it. Buyers pay a premium for the upside above the strike. You keep most of the downside.
Options incomeCentralized exchange, DeFi8 min read
- Risk
- Elevated
- Complexity
- Intermediate
- Min. capital
- $1k–$10k
- 6–20%target APY
Perpetual funding-rate arbitrage
Buy spot and short the same amount of the perpetual future. Price exposure cancels out, and the short leg collects the funding that leveraged longs pay.
Funding-rate arbitrageCentralized exchange, DeFi8 min read
- Risk
- Low
- Complexity
- Intermediate
- Min. capital
- $1k–$10k
- 10–25%target APY
Cross-venue prediction-market arbitrage
Buy YES on one venue and NO on another for the same event. When the two prices add up to less than a dollar, the difference is yours at resolution. A month of live trading between Polymarket and outcome.xyz, fees included.
Prediction-market arbitrageDeFi10 min readAudio
- Risk
- Moderate
- Complexity
- Advanced
- Min. capital
- $10k–$100k
Comments
No comments yet. Questions about sizing, venues or risks are welcome.