Help us double down on what's working, instead of guessing. Takes 5 seconds, totally optional.
optionsstrategies · 276 min read
Options strategies ranked: a reference guide with 50 light-backtest profiles
All 50 options strategy profiles ranked by light-backtest results on SPY, QQQ and IWM, 2020 to 2025, with constructions, what each sample showed, what to test next and a downloadable trade ledger for every profile.
You want to know which options strategy is worth your time. This guide ranks 50 of them by how they actually did on SPY, QQQ and IWM options from 2020 to 2025, priced from FlashAlpha's historical options archive, the same data behind the Historical API. Then it walks through each one: what it is, what happened, and what to try next.
Treat it as a research ideas list. The ranking shows where the interesting questions are, not what will work next year. Every strategy links to its own trade ledger, so you can check any number here against the trades behind it.
50
strategies
7,265
simulated trades
2020 to 2025
SPY, QQQ, IWM
Bid/ask + $0.65
every fill, every leg
All 50 strategies, ranked
Every % is a percentage of the value of 100 shares of the ETF at entry. On a $500 ETF that is $50,000, so +0.76% is about $380 a trade and -20% is a $10,000 loss. It is not a return on premium, margin or account equity. Prices are the ones you would get: buy at the ask, sell at the bid, $0.65 a contract each way. Column definitions follow the table.
42 monthly ETF strategies
Ordered by Sharpe ratio, the annualized return per unit of monthly volatility. Rank by mean shows where each strategy sits by average result, so you can see what moves when risk is counted.
Same rules for every row: pick the contracts on the first trading day of the month, get in at the next close, get out ten sessions later. Options about 35 days from expiry. SPY, QQQ and IWM, January 2020 to December 2025, up to 216 months per strategy. Every bid, ask and delta comes from FlashAlpha's options archive.Coverage by strategy
How to read the columns:
Mean after costs is the expectancy: the average result per trade. Median is the middle trade. When the two are far apart, a few big trades are moving the average.
Win rate is the share of trades that finished positive. Worst trade is the single biggest loss in the sample.
Sharpe is the annualized Sharpe ratio of the monthly results (each month's figure is the average across the ETFs traded that month; mean divided by standard deviation, times the square root of 12, no risk-free rate subtracted). Profit factor is total gains divided by total losses. Max drawdown is the largest peak-to-trough fall in the running total of monthly results.
Why the trade counts differ. Each strategy runs on every month where its contracts had clean quotes, so some have more trades than others. The earnings trades and the wheel run on their own schedules, so they get their own tables.
What it does not do. The tables do not equalize capital or margin. A credit spread and a long call can sit next to each other and tie up very different amounts of money.
Risk detail for the same 42 strategies
Same order, same 42 rows. This table adds the profit factor, the drawdown you would have sat through to collect the result, and the average win and loss.
Sharpe uses the monthly series, annualized, with no risk-free rate. Max drawdown is on the running total of monthly results, not compounded. For scale, 100 shares of the ETF with no options scored a Sharpe of 0.66 on the six-strategy matched windows.Download all risk metrics (CSV)
Running total of monthly results, 2020 to 2025, in % of the 100-share reference value (not compounded). Each line is one strategy on its own eligible months; the dashed line is 100 shares of the ETF at midpoint with no costs.
Covered callLong callVol-carry filtered put spreadCash-secured putIron condor100 shares of the ETF, no options, no costs
Three things stand out when you rank by Sharpe:
The filtered put spread moves to the top. The vol-carry filtered put spread has the best Sharpe in the table at 1.18, on 73 trades. Small sample, but the smoothest ride in the dataset.
The share overlays keep their expectancy and show their drawdowns. The covered call and protective put have max drawdowns near a quarter of the reference value. That is the stock, not the option: 100 shares with no options drew down -27.91% on the matched windows, and in the chart the covered call's line follows the ETF's dashed line almost month for month.
Every symmetric butterfly and single-right condor has a Sharpe below -2. Steady small losses, month after month, with almost no winners to offset them.
Four 2025 earnings each for AAPL, MSFT and AMZN. In one session before the report, out one session after, except the buildup trade. Six to twelve trades each, so read these as worked examples rather than proof of anything.
Total return on $100,000 of starting cash per account, 2020 to 2025. One contract at a time, options held to expiry, assignment assumed on anything that finishes in the money, no interest on idle cash. The wheel is an account, not a single trade, so it gets its own scale.
Six strategies on exactly the same months
The big table compares each strategy on its own months. This one is stricter: six structures run on the same 133 months where all six could be traded. It is the fairest ranking in the article, ordered by Sharpe with the rank by average result alongside.
Midpoint prices every leg halfway between bid and ask with no fees. After costs uses the bid and ask you would actually get, plus $0.65 a contract each way. Win rates use the after-cost numbers. The last row is the ETF itself over the same windows, for scale.Download the ranking · Trade-level results
Three things stand out:
Winning often is not the same as making money. The bull put credit spread won 72.9% of the time and the long call won 55.6%. The credit spread still came third, at +0.09% a trade against +0.31% for the call. How big the losses are matters as much as how often they happen.
Costs reorder the table. The call calendar was -0.01% at midpoint and -0.17% once you pay the spread. The iron condor went from about flat to -0.10%. If a strategy only works at midpoint prices, it does not work.
First place is not an edge. The ETFs themselves rose an average of +0.94% per window with a Sharpe of 0.66, above every one of the six structures. Five of the six strategies have a 95% range for their average that includes zero. The long call's runs from about -0.19% to +0.79%. This is what happened on one sample, not a forecast. Claim audit
Where to start: six research ideas from the ranking
The top of the table still leans on the stock market. By Sharpe the leaders are the vol-carry filtered put spread, bullish risk reversal, covered call, protective put and collar. By average result the five best (covered call, protective put, collar, bullish risk reversal, long call) all own upside in ETFs that went up. Rerun each one against 100 plain shares on the same months. Whether the option added anything is the real question, and the ledgers answer it.
Premium selling wins often and loses big. The cash-secured put won 74.1% of the time and has the single biggest loss in the dataset, at -20.07%. The bull put credit spread won 70.4% of the time and still averaged slightly below zero. Sort any premium-selling ledger by loss and look at the dates. Then test what a size limit or an exit rule would have done on those dates.
Multi-leg trades pay the spread twice. Every mirror pair in the data (iron condor and reverse iron condor, iron butterfly and reverse, each ratio spread and its backspread) has exactly opposite midpoint results, and after costs both sides lose or barely break even. Symmetric butterflies and single-right condors won fewer than one trade in six. Test holding to expiry, and test only entering when the spread is tight. The historical chain returns every leg's bid and ask, so that filter is one line.
One volatility filter helped. Three did nothing. Selling the put credit spread only when implied volatility was at least 1.2 times realized volatility lifted the average from -0.01% to +0.17% a trade, skipped the worst month, and gave it the best Sharpe in the dataset. The skew and tail filters fired almost every month and changed nothing. Build the control sample and move the threshold around. Section 7 has the details, and /v1/strategies/vol-carry is the live version of the signal.
Selling earnings worked on twelve events. The short strangle won 10 of 11, the iron condor 9 of 11 and the short straddle 8 of 12. Buying the straddle lost 9 of 12. Twelve megacap earnings in one year show how the accounting works, not that it keeps working. A full year across a wide universe is the next test, and /v1/earnings/iv-crush gives you the implied move and the crush history for any name to start from.
The wheel was mostly cash. Three accounts finished six years up 22.4%, 21.9% and 4.8%, earning nothing on idle cash and never getting assigned early. Compare that with holding the ETF and with holding cash at a real rate before you draw a conclusion.
How to read the profiles
Each of the 50 profiles below has the same four parts:
Setup is the exact construction this study used, including delta targets and strike geometry.
Exposure is what the position gains and loses from, in plain terms.
In the sample reads the numbers: what the mean, median, win rate, Sharpe, drawdown and worst trade say about this profile on its own eligible windows.
Test next is the one change most likely to tell you something new.
Each profile ends with a generated Sample line and a link to its trade ledger. The numbers come straight from the frozen results files.
A quick glossary. Long means bought and short means sold. A debit is a net payment to open; a credit is cash received. ATM is at the money, near the current price; OTM is out of the money. A wing is an outer strike that limits or shapes the exposure. A vertical uses two strikes in one expiration, a calendar uses two expirations at one strike, and a diagonal changes both. Delta is how much an option's price moves for a small move in the underlying; the 50, 25 and 10 in this article are selection targets for absolute delta, not probabilities of profit. IV is implied volatility, the volatility level option prices imply. RV is realized volatility, measured from past returns. OIC glossary
A single bought option pays for exposure to a move. A debit vertical sells part of that payoff to lower the cost. A credit vertical collects premium and takes a defined spread risk. All three can express the same view and produce very different result distributions.
Setup Buy one call with delta near 0.50, about 35 days to expiration. Hold ten sessions.
Exposure You gain from a rise in the underlying and from a rise in implied volatility. You lose from time passing, from a fall in the underlying and from a drop in implied volatility. The most you can lose is the premium.
In the sample The top of the controlled comparison. The mean and median are close (+0.36% and +0.33%), so the average is not the product of one lucky month. The win rate is only a little above half. The underlying rose on average over the period, and that is the main reason a bought call did well: on the matched windows the ETF itself had the better Sharpe. The worst trade was -4.57% and the max drawdown -19.17%.
Test next Compare the call with a sized position in the underlying on the same windows. If the call's advantage disappears against 50 shares, what you are seeing is the market, not the option. The historical stock quote gives you the same-minute share price to do it.
Setup Buy one put with delta near 0.50, about 35 days out. Hold ten sessions.
Exposure You gain from a fall in the underlying or a rise in implied volatility. You lose from time, from a rise and from falling volatility. The premium is the most you can lose.
In the sample The market rose more often than it fell across 2020 to 2025, and a standing bearish bet paid for that. Fewer than three trades in ten were profitable. The median trade (-1.03%) was much worse than the mean (-0.31%), because a few large gains in selloffs offset many small losses. The drawdown of -34.76% is the cost of carrying that insurance for six years. That shape is what insurance looks like.
Test next Decide whether you are testing a forecast or a hedge. As a forecast, test any entry signal against unconditional entries on matched dates. As a hedge, attach the put to a stock position and measure loss reduction, not the put's own return.
Setup Buy a 50-delta call and sell a 25-delta call with a higher strike in the same expiration.
Exposure Bullish, with a lower cost than the long call and a capped payoff. The short call gives back the largest up-moves.
In the sample The win rate was a little higher than the long call's and the median trade was similar, but the mean was about a third of it. The cap removed the biggest winners, and two legs cost more to trade than one: the gap between midpoint and after-cost results was wider here than for the single call.
Test next Measure how much of the long call's total profit came from windows where the underlying moved past the short strike. That is what the cap cost.
Setup Buy a 50-delta put and sell a 25-delta put with a lower strike in the same expiration.
Exposure Bearish, with a lower cost than the long put and a capped payoff below the short strike.
In the sample The spread lost less than the outright put on average and won a little more often, because the short put cut the cost of being wrong in a rising market. It still lost on about two trades in three. The cap also removed the large gains that made the long put's selloff months worthwhile.
Test next Compare the spread with the long put on their shared windows, split into months where the underlying fell more than the spread width and months where it did not.
Setup Sell a 25-delta put and buy a 10-delta put with a lower strike. Collect a credit.
Exposure You keep the credit if the underlying stays above the short strike. You lose up to the spread width minus the credit if it falls through both strikes. Rising implied volatility hurts before expiration.
In the sample This profile shows why the two tables in this article differ. On the 133 matched windows it ranked third with a mean of +0.09%. On its own larger sample of 206 trades the mean was -0.01%. The extra windows include the worst trade in the directional group, at -7.05%. The win rate stayed above 70% in both samples. Seven wins in ten and a negative mean is what a short-tail structure looks like when the tail arrives.
Test next Sort the ledger by loss and look at the dates. Then ask whether any rule you would have followed at the time would have kept you out of those months.
Setup Sell a 25-delta call and buy a 10-delta call with a higher strike.
Exposure You keep the credit if the underlying stays below the short strike. You lose up to the width minus the credit on a rally through both strikes.
In the sample Selling upside into a rising market lost money more often than not. Windows that span an ex-dividend date are excluded for any short call, so this profile has fewer eligible trades (180) than its put-side twin, and the two samples are not directly comparable.
Test next Separate the bearish forecast from the standing short-upside position. Run the spread only on dates where a stated signal says the market is extended, and compare with unconditional entries.
For these profiles, option premium alone is a misleading score. The shares move, dividends accrue, cash sits idle and assignment changes the inventory. The monthly overlays hold 100 shares plus one option for ten sessions. The wheel is a separate, stateful account ledger.
Because the underlying rose on average over 2020 to 2025, every overlay that holds shares shows a positive mean. That is the stock, not the option. The useful comparison is against the same 100 shares with no option.
Setup Hold 100 shares and sell one 25-delta call against them, about 35 days out. Windows that span an ex-dividend date are excluded.
Exposure You keep the shares' gains up to the strike, plus the premium. Above the strike you give up further gains. Below, you carry almost all of the shares' loss, less the premium.
In the sample The highest mean in the monthly table, at +0.76%, with a median much higher at +1.71%. Most months the shares rose. A few bad months pulled the average down, and the worst trade, -11.84%, is a stock loss the call did nothing to stop. The max drawdown of -24.13% is the shares' drawdown, less the premium collected along the way.
Test next Compare total wealth, including dividends, with the same shares and no call, on the same windows. Then split by whether the underlying finished above the strike.
Setup Sell one 25-delta put and hold enough cash to buy 100 shares at the strike. The profile marks the option over ten sessions; it does not pay interest on the cash.
Exposure You keep the premium if the underlying stays above the strike. Below it, you carry the shares' loss from the strike down, less the premium.
In the sample The highest win rate of any monthly profile (74.1%) and the largest single loss in the whole dataset (-20.07%, about a fifth of the reference value). The median was comfortably positive. The mean was small and the Sharpe close to zero. This is the clearest example in the data of a strategy whose average is set by its rare months. Readers often ask why it sits so far below the covered call when the two are synthetically the same trade. Here they are not: the covered call sells a 25-delta call above the market and holds the shares, the cash-secured put sells a 25-delta put below the market and holds cash, and the two profiles are eligible in different months. Same payoff shape, different strikes, different sample.
Test next Rebuild the result as an account return with cash earning a stated rate, then compare with the covered call on shared windows. The two are often called equivalent; check whether they were here.
Exposure You keep the shares' upside, less the premium. Below the strike, the put caps further loss for as long as it is held.
In the sample The mean was close to the covered call's, and the worst trade was about 2.5 percentage points smaller. That gap is what the put bought in the worst month. In ordinary months the premium was a drag, which shows in a median below the covered call's.
Test next Score the hedge on loss reduction in the worst 10% of windows, not on its average. Then price the drag in the other 90% and decide whether that trade suits the portfolio.
Setup Hold 100 shares, buy a 25-delta put and sell a 25-delta call.
Exposure Loss is limited below the put strike and gain is limited above the call strike. The net premium can be a debit or a credit depending on skew.
In the sample The lowest mean of the three share overlays and the smallest worst trade. That is the design working as intended: less upside, less downside. A collar is not free. Check the net premium in the ledger before assuming zero cost.
Test next Report three numbers together on each window: downside protected, upside given up, net option cost. A collar study that reports only one of them is not finished.
Setup Start with $100,000 in cash for each of SPY, QQQ and IWM. Each month, if no option is open, sell a 35-day 25-delta put while holding cash, or a 35-day 25-delta call while holding 100 shares. Hold to expiry. Assume assignment on any option that finishes strictly in the money. Accrue dividends while holding shares. Sell any remaining shares at the end of 2025. No early assignment, no interest on cash.
Exposure A cycle of short put, share ownership, short call and back to cash. The account is mostly idle cash with one contract's exposure. Returns depend on how long the shares are held and what they do while held.
In the sample The three accounts finished 2020 to 2025 with total returns of 22.4%, 21.9% and 4.8% on initial cash, across 106 completed cycles. A wheel that sits in cash most of the time is not the same bet as holding the shares, and that is by design. Those figures also come from an account that never earned interest on its cash and never faced early assignment.
Test next Rerun with interest on idle cash and with a stated early-assignment rule around ex-dividend dates. Then compare with holding the ETF and with holding cash.
Sample 106 completed cycles across three accounts. Account and inventory ledger, not comparable with the per-trade figures above.
Setup Buy an 80-delta call in the expiration nearest one year out (270 to 540 days) and sell a 25-delta call about 35 days out, with the long strike below the short. Close both legs after ten sessions.
Exposure A diagonal that stands in for a covered call. The long call replaces the shares at a fraction of the capital, but it decays, it pays no dividend and it is not shares.
In the sample At midpoint the profile was solidly positive (+0.53%). After costs it was the worst of the overlays at -0.61%. That is the largest cost drag in the dataset. Long-dated deep-in-the-money calls have wide spreads, and this profile crosses that spread twice in ten sessions. A real PMCC holds the long call for months and rolls only the short. This sample does not test that program.
Test next Hold the long call and roll the short call monthly for a year, charging the spread only when a leg actually trades. Compare with a covered call on the same shares.
Cboe's buy-write, put-write and collar indices are a useful outside reference for these overlays. Their rules differ from this study, so compare the rules before you compare the numbers. Cboe strategy benchmark indices
3. Volatility and range trades
These structures are usually called volatility trades. Direction still matters once the underlying moves away from the starting strikes. Buying both sides does not profit from any move, only from a move larger than the premium paid. Selling both sides is not neutral under stress.
Four of the profiles in this section come in mirror pairs. The iron condor and reverse iron condor, and the iron butterfly and reverse iron butterfly, are the same legs with opposite signs. Their midpoint results are exact opposites. Their after-cost results are all negative. That is the cleanest evidence in the article that the bid/ask spread, not the direction of the trade, decided these outcomes.
Setup Buy a call and a put at the ATM strike, about 35 days out.
Exposure You gain from a large move in either direction or from a rise in implied volatility. You pay two premiums and lose them to time if nothing happens.
In the sample At midpoint the straddle was slightly positive. After costs it was slightly negative. The median was well below the mean: most months the move was too small, and a few months of large selloffs and rallies paid for the rest. The worst trade lost a large part of the premium paid.
Test next Split windows by whether the realized ten-session move exceeded the straddle's entry price. That separates "the market moved" from "the market moved enough".
Setup Sell a call and a put at the ATM strike. Ex-dividend windows are excluded because of the short call.
Exposure You keep the premium if the underlying stays near the strike and volatility falls. Loss grows without limit on a rally, and on a selloff it grows until the underlying reaches zero. Margin is not modeled.
In the sample About half the trades won, but the mean was clearly negative and the worst trade was among the largest losses in the dataset. The midpoint result was negative too, so costs are not the whole story here: the underlying moved further than the premium covered often enough to make the strategy a net loser over these six years.
Test next Add a margin model and a forced-exit rule. Without them, a short straddle backtest measures a position nobody could hold. Minute-level replay lets you mark the position through the day rather than only at the close. OIC short straddle
Exposure Like the straddle, but cheaper to open and needing a larger move to pay. Loss is limited to the premium.
In the sample Cheaper did not mean better. The strangle won about one trade in three and lost more on average than the straddle, while its worst trade was smaller because less premium was at risk.
Test next Plot each trade's result against the size of the realized move. The strangle needs a bigger move than the straddle; find out how often the sample provided one.
Exposure You keep the premium if the underlying stays between the strikes. Loss is unlimited above the call strike and large below the put strike. Rising volatility hurts before expiration.
In the sample A high win rate, a positive median and a slightly negative mean. The strangle lost less on average than the short straddle because the strikes gave the underlying room, and its worst trade was roughly half the straddle's. The pattern is the same: many small wins, a few large losses, a mean below zero.
Test next Test the volatility-carry filter from section 7 on this structure. The filtered condor result there suggests a broad IV/RV rule is not enough on its own.
Setup Sell a 25-delta put and a 25-delta call. Buy a 10-delta put and a 10-delta call as wings.
Exposure A short strangle with defined risk. The wings cap the loss on each side and cost premium.
In the sample At midpoint the condor was flat. After costs it lost about 0.1% per trade. Four legs, crossed twice, is the whole difference. The win rate was a little over half. On the 133 matched windows it ranked fourth of six.
Test next The section 7 volatility-carry condor is the obvious follow-up: same legs, entered only when IV is rich against RV. In this sample it did not help.
Setup Buy the 25-delta put and call. Sell the 10-delta put and call as wings. The same four legs as the iron condor, with signs reversed.
Exposure A long strangle with a capped payoff. You gain from a move out of the middle region, up to the wing width.
In the sample The midpoint result is the exact opposite of the iron condor's. After costs both lose. The reverse condor had the smallest worst trade in this section, because the wings limit what can go wrong, and it won fewer than four trades in ten.
Test next Compare with the uncapped long strangle on shared windows and count how many of the strangle's profitable months the wings cut short.
Setup Sell the ATM call and put. Buy 10-delta wings on each side. Wing widths can differ.
Exposure A short straddle with defined risk. Maximum profit sits at the body strike; the payoff falls away fast on either side.
In the sample Worse than the iron condor on every measure: lower win rate, larger mean loss, larger worst trade. The body sits exactly where the underlying starts, so almost any move hurts, and the position is expensive to open and close.
Test next Measure how often the underlying stayed within one wing width of the body after ten sessions. That is the frequency the structure needs.
Setup Buy the ATM call and put. Sell 10-delta wings. The mirror of the iron butterfly.
Exposure A long straddle with a capped payoff, financed in part by the wings.
In the sample Positive at midpoint, negative after costs, the mirror image of the iron butterfly. The cap trimmed the large moves that a plain straddle relies on, and the extra two legs cost more to trade than they saved.
Test next Compare with the long straddle on shared windows. Count the months where the move exceeded the wing width; those are the months the cap gave away.
Iron structures mix puts and calls. The structures in this section use one right throughout: a call butterfly is all calls, a put condor is all puts. Expiration diagrams can look the same as their iron cousins, but entry prices, liquidity and exercise behaviour differ. Strikes must exist on the listed grid at the required spacing; where they do not, the window is excluded rather than approximated.
The four symmetric structures here have the lowest win rates of any profile with more than a handful of trades, and the four lowest Sharpe ratios of any monthly strategy. That is mostly a mismatch between the structure and the holding period. A debit butterfly earns most of its value in the final days before expiration, when the short body options decay. Closing after ten sessions of a 35-day option leaves most of that behind, and three or four legs of spread costs take the rest.
Setup Buy one call about 3% below spot, sell two ATM calls, buy one call about 3% above spot. Equal spacing, 1:-2:1.
Exposure A debit trade that pays most if the underlying finishes at the body strike at expiration. Loss is limited to the debit.
In the sample Near flat at midpoint, clearly negative after costs, and profitable on about one trade in six. The gap between those two numbers is four legs of spread.
Test next Hold to expiration instead of ten sessions, on the same windows. That shows how much of the payoff the early exit left behind. OIC long call butterfly
Setup The same 1:-2:1 construction with puts: buy one put below, sell two ATM, buy one above.
Exposure The same expiration shape as the call butterfly, built from puts.
In the sample A slightly better mean than the call butterfly, a lower win rate and a smaller worst trade. The two are close enough that the difference is more likely to be which windows were eligible than anything about puts versus calls.
Test next Compare call and put butterflies only on windows where both had valid strikes and quotes. Then compare their entry debits leg by leg.
Setup Buy a call two widths below spot, sell one call one width below, sell one call one width above, buy one call two widths above. Width targets 1.5% of spot.
Exposure A butterfly with the body split in two, giving a wider flat-topped payoff region at expiration.
In the sample Fewer than one trade in ten was profitable. The mean after costs was about the same as the butterflies. The wider profit zone did not help within ten sessions, because the position had not yet earned its decay.
Test next As for the butterfly: change the exit to expiration on the same windows. Then compare the two widths.
Setup The same four-strike construction with puts.
Exposure The same as the call condor, built from puts. Not the same legs as an iron condor.
In the sample The lowest win rate in the article, at 5.0%. The worst trade, -4.99%, is larger than any 1.5% width could explain for a defined-risk debit structure. It is a March 2020 QQQ trade, and the four legs were quoted with spreads of about three to four dollars each at entry. Defined risk on paper was not defined risk at the prices in the archive.
Test next Open the ledger at the worst trade and reconstruct the four quotes. Then add a maximum-spread rule at entry and see what the sample looks like without those months. The historical chain has a maxSpreadPct filter for exactly this.
Setup A 1:-2:1 call butterfly with unequal wings. The lower wing targets 3% of spot; the upper wing is twice as wide.
Exposure The wide upper wing tilts the payoff. Loss is larger on a rally through the upper region than on a decline. Whether the trade opens for a credit depends on prices, not on the name.
In the sample The worst mean in this section. The wide upper wing sat on the side the market moved most often, and the two short calls in the body lost as the underlying rallied through them. The win rate was higher than the symmetric butterflies, but the losses were larger.
Test next Report up-move and down-move windows separately. A single average hides which tail paid for the structure's better region.
Setup A 1:-2:1 put butterfly with the wide wing below the body. The upper wing targets 3% of spot; the lower wing is twice as wide.
Exposure The mirror tilt: larger loss on a deep decline, less exposure to a rally.
In the sample Positive at midpoint and the best of the six in this section after costs, because the wide wing sat on the side the market visited less often. The mean was still negative once the four legs were paid for.
Test next Pair this with the call broken wing on shared windows. Together they show how much of each result was the market's direction over the period.
These positions hold options with different expirations. The front option targets 35 days, the back option 65. Both legs are marked at the ten-session exit, so the result depends on where the underlying is and on how each expiration repriced. There is no single expiration payoff to draw.
Calendars need a valid quote on the same strike in two expirations, so their eligible samples are smaller than the verticals'.
Setup Sell the ATM call in the front expiration and buy the ATM call at the same strike in the back expiration.
Exposure You gain if the underlying stays near the strike and if back-month volatility rises relative to the front. You lose on a large move in either direction.
In the sample Near flat at midpoint, clearly negative after costs, with the largest cost drag of the six headline structures. Fewer than one trade in three won. Time decay in the front option did not arrive fast enough in ten sessions to cover two legs of spread.
Test next Section 7's term-structure filter enters this calendar only when back-month IV is higher than front-month IV. It found four eligible windows and lost all four. Start there and ask why the filter almost never fired.
Setup Sell the ATM put in the front expiration and buy the ATM put at the same strike in the back expiration.
Exposure The same maturity bet as the call calendar, built with puts. Exercise and dividend considerations differ.
In the sample Almost the same mean as the call calendar, a slightly better win rate and a larger worst trade. The two calendars sit almost on top of each other, which is what put-call parity predicts when the strike is the same.
Test next Compare the two on shared windows and decompose each result into front-leg and back-leg P&L. The ledger has both.
Setup Sell a 25-delta call in the front expiration and buy a 50-delta call in the back expiration. Different strikes and different dates.
Exposure A calendar with a bullish tilt. The back call carries more delta than the front, so the position gains on a moderate rise and loses on a decline.
In the sample The only calendar-family profile with a positive mean after costs, and one of the higher win rates in the article. The reason is the tilt: the position was net long delta in a market that rose. The worst trade was a decline, and it was large.
Test next Hedge the entry delta with shares and rerun. What remains is the calendar component, which is what the strategy name claims to be trading.
Setup Sell a 25-delta put in the front expiration and buy a 50-delta put in the back expiration.
Exposure A calendar with a bearish tilt.
In the sample The mirror of the call diagonal's story: net short delta in a rising market, with a negative mean and a win rate below 40%. The two diagonals together show that the tilt, not the calendar, drove both results.
Test next Same as the call diagonal: hedge the entry delta and look at what is left.
Setup A call diagonal and a put diagonal together: sell the 25-delta front call and put, buy the 50-delta back call and put.
Exposure The two tilts cancel at entry, leaving a position that is roughly neutral and long the back expiration against the front.
In the sample A negative mean and the lowest win rate in the section. The worst trade was small for a four-leg structure. Costs did most of the damage: the midpoint result was close to flat.
Test next Track the position's delta through the ten sessions. The neutrality at entry does not last once the underlying moves through a short strike.
Here the quantities matter as much as the strikes. A ratio spread sells two contracts for every one it buys and keeps uncovered exposure on one side. A backspread reverses the quantities and can lose in the middle of the price range. None of these constructions targets neutrality or requires a credit at entry.
Two mirror pairs live in this section. The call ratio spread and call backspread are the same legs with opposite signs, and so are the put ratio spread and put backspread. Their midpoint results are exact opposites. After costs, three of the four lose and the fourth barely clears zero.
Setup Buy a 25-delta call and sell a 25-delta put, same expiration.
Exposure Synthetic bullish exposure with a gap in the middle. Above the call strike you participate; below the put strike you own the decline. The net premium depends on skew and is usually a small debit or credit.
In the sample The best mean in this section and one of the best in the article, for the same reason as the long call: the market rose. The put side had a bad month, and the worst trade shows what a short 25-delta put does in a selloff.
Test next Compare with 100 shares and with the long call on shared windows. The risk reversal is a cheaper way to hold much of the shares' exposure; the ledger shows what that discount cost in the worst months.
Setup Buy a 25-delta put and sell a 25-delta call.
Exposure Synthetic bearish exposure. Unlimited loss on a rally above the call strike.
In the sample The mean of -0.61% is the worst among the monthly profiles with more than a handful of trades, and the drawdown of -46.86% is one of the two deepest in the dataset. A standing short-upside position through 2020 to 2025 was expensive. The number to remember is not the average but what the short call did in the strongest rally months.
Test next Run it only on dates where a stated bearish signal fires and compare with unconditional entries. Without a signal, this is a bet against the drift.
Setup Buy one 50-delta call and sell two 25-delta calls.
Exposure Profit on a moderate rise toward the short strike. Above it, the extra short call loses without limit.
In the sample Slightly negative at midpoint, clearly negative after costs, with a worst trade that shows the uncovered call at work in a strong rally. The median was near zero: most months did little, and the tail did the damage.
Test next Plot the result against the ten-session move. The loss region above the short strike is the whole risk; find out how often the sample went there.
Setup Buy one 50-delta put and sell two 25-delta puts.
Exposure Profit on a moderate decline toward the short strike. Below it, the extra short put loses down to zero.
In the sample The only profile in this section with a positive mean after costs, and only barely. At midpoint it was clearly positive: selling two 25-delta puts collected skew premium in a market that rarely fell far. The March 2020 windows were excluded by the quote rules for this profile, so the sample never tested it in a crash. The worst trade was contained by the sample, not by the structure.
Test next Find the excluded windows in the selection ledger and price the position through them by hand. A ratio put spread's worst case is a crash, and this sample skipped the one it had.
Setup Sell one 50-delta call and buy two 25-delta calls. The call ratio spread with signs reversed.
Exposure Profit on a large rally. Loss in the region between the strikes, where the short call has gained and the long calls have not yet caught up.
In the sample The exact mirror of the call ratio spread at midpoint, and a loser after costs. Fewer than four trades in ten won. The market rose often, but mostly not far enough within ten sessions to get past the loss region.
Test next Compare exits at ten sessions and at expiration. The backspread's loss region is deepest at expiration, so the exit rule changes the result more than for most structures.
Setup Sell one 50-delta put and buy two 25-delta puts.
Exposure Profit on a large decline. Loss between the strikes on a moderate decline. A small profit or loss on a rally, depending on the entry credit.
In the sample The mirror of the put ratio spread, and the second-worst mean in this section. Fewer than three trades in ten won. It is bought protection against a crash, and the sample period charged for it every month the crash did not come.
Test next Attach it to a stock portfolio and measure drawdown reduction in the worst windows. That is the job it is built for; a standalone mean does not measure it.
A signal is an entry rule, not a payoff. These five profiles apply a fixed volatility measurement to a structure defined elsewhere in this article. They are research proxies with thresholds frozen before the sample was run. They do not replay FlashAlpha's production endpoint scores. The live strategy endpoints cover the same signal families: /v1/strategies/vol-carry, /v1/strategies/skew, /v1/strategies/tail-pricing and /v1/strategies/term-structure.
Each filter reduces the sample to the windows where it fired. A better conditional mean on fewer trades is not the same as a better strategy: the time spent waiting and the windows the filter skipped are part of the result. Signal decisions ledger
Setup Enter the 25/10-delta bull put credit spread only when the signal-day ATM implied volatility is at least 1.20 times the trailing 20-day realized volatility.
Exposure The same as the bull put credit spread, on fewer dates.
In the sample The filter fired on 73 windows. On those, the mean was +0.17% against -0.01% for the unfiltered spread, the worst trade was -2.24% against -7.05%, and the Sharpe of 1.18 is the best in the dataset. The filter skipped the unfiltered spread's worst window. That is the most encouraging signal result in the dataset. It is also 73 trades on one fixed threshold.
Test next Build the control: the same spread on the windows where the filter did not fire. If the control is as good, the filter is selecting time, not edge. Then vary the threshold and see how fast the result moves.
Setup The iron condor, entered only when ATM IV is at least 1.20 times trailing RV.
Exposure The same as the iron condor, on fewer dates.
In the sample The filter did not help here. The conditional mean was worse than the unfiltered condor's, on about a third of the windows. A rule that improved the one-sided put spread did nothing for the two-sided range trade in this sample.
Test next Split the filtered windows by which side of the condor lost. If the call side is doing the damage, the IV/RV ratio is measuring the wrong thing for the up-move risk.
Setup When the 25-delta put IV and 25-delta call IV differ by at least 0.03 (three volatility points), sell a 25/10-delta credit spread on the richer side.
Exposure Usually the bull put spread, because equity index puts are almost always richer than calls. Occasionally the bear call spread.
In the sample The result is close to the unfiltered bull put spread on both mean and win rate, and the worst trade is the same trade. The filter fired on most windows and did not screen out the month that mattered.
Test next Raise the threshold until the filter fires on a minority of windows and see whether the selected set differs from the unfiltered spread in any way other than size.
Setup When 10-delta put IV exceeds ATM put IV by at least 0.05, enter the 25/10-delta bull put credit spread.
Exposure The same as the bull put spread. The expensive 10-delta put is the wing you buy, not the option you sell, so this is a slope condition on entry, not a trade that shorts the expensive tail.
In the sample Almost identical to the skew-conditioned vertical and to the unfiltered spread: same worst trade, similar mean, similar win rate. A steep put wing was the normal state of the market across these six years, so the condition rarely excluded anything.
Test next Rebuild the trade around the signal. If the 10-delta put is expensive, a test of that claim sells it, with the 25-delta put bought as a hedge. That is a different spread from the one in this profile, with a different risk.
Setup When the back-month 50-delta call IV exceeds the front-month 50-delta call IV by at least 0.02, enter the call calendar.
Exposure The same as the call calendar. The rule buys the higher-IV back option and sells the lower-IV front option, so it is not obviously a buy-cheap, sell-rich trade.
In the sample Four trades, four losses. The condition almost never held on the monthly signal dates in this period. The four rows are a demonstration of the rule and its data needs, not a result.
Test next Change the condition to the one calendar sellers actually want, front IV above back IV, and count how many windows that yields. If it is still a handful, the monthly schedule is the constraint.
The earnings cohort covers four 2025 announcements each for AAPL, MSFT and AMZN. Event dates were checked against issuer releases after the fact. Each profile has between six and twelve completed trades. That is enough to show the accounting and far too few to show an edge.
Except for the buildup trade, positions enter one observed session before the announcement and exit one session after it, so the holding period always spans the event without assuming a release time. The front expiration targets seven calendar days after the event; the back expiration targets 35. The AAPL end-of-day view has no quote-size columns, so earnings trades do not enforce displayed depth. For the comparison every one of these profiles needs, the implied move and the historical crush, FlashAlpha's earnings endpoints (/v1/earnings/expected-move, /v1/earnings/iv-crush, /v1/earnings/history) cover any listed name. Event dates and issuer sources
Setup Buy the ATM call and put the session before the announcement. Close the session after.
Exposure You gain if the stock moves more than the straddle cost. You lose the volatility premium that comes out of the options once the event is known, plus the move you did not get.
In the sample The worst mean in the article. Three of twelve events moved enough to pay. The other nine lost the volatility premium and then some. Buying the event was expensive on these twelve dates.
Test next Record the implied move at entry beside the realized move for each event. The question is not whether the stock moved but whether it moved more than it was priced to.
Setup Sell a 25-delta put and a 25-delta call the session before the announcement. Close the session after.
Exposure You keep the premium if the stock stays between the strikes. Loss is unlimited above the call and large below the put. There is no margin model.
In the sample Ten of eleven events paid, and the mean was the highest in the article. The one loss was small. Eleven events on three of the most liquid stocks in the market, in one year, is not a sample that can show what happens on the twelfth.
Test next Extend to a full year of earnings across a broad universe before drawing any conclusion. Then add the margin and liquidation model that a live short strangle needs.
Setup The short strangle with 10-delta wings bought on each side.
Exposure Defined risk on both sides, paid for with part of the premium.
In the sample Nine of eleven events paid. The mean was less than half the short strangle's on the same events: the wings cost premium and, with only small moves in this sample, never paid it back. That gap is the insurance premium for the event that did not happen.
Test next Compare the two on the same eleven events, event by event. Then price the wings against the strangle's theoretical loss on a 10% gap.
Setup Sell the ATM call in the first expiration after the event and buy the same strike about 35 days out.
Exposure You gain if the front option's volatility premium collapses more than the back option's after the event, and the stock stays near the strike.
In the sample Six completed events, all losses, and six exclusions. A positive midpoint mean became a negative after-cost mean: the two legs on a single-stock chain cost more to cross than the strategy made. Start with the exclusions before reading anything into the six.
Test next Inspect why half the events were excluded. If it is quote quality on the front week, the strategy may not be tradable at the prices in the archive.
Setup Sell a 25-delta front call and buy a 50-delta back call across the announcement.
Exposure A bullish calendar. The result depends mostly on the direction and size of the gap.
In the sample Four of twelve events paid, with a mean close to the earnings calendar's and a much worse median. The stocks did not gap up often enough to reward the tilt.
Test next Compare with the earnings calendar on matched events, with the entry delta hedged. That isolates the maturity trade from the direction bet.
Setup Buy the ATM straddle ten sessions before the announcement and sell it the session before. Never hold through the event.
Exposure You gain if implied volatility rises into the event by more than time decay takes away, or if the stock moves in the run-up.
In the sample Near flat at midpoint and negative after costs. Four of twelve events paid. Implied volatility does tend to rise into earnings, but in this sample it did not rise enough to beat ten sessions of decay and two crossings of the spread.
Test next Record entry and exit IV for both legs. The trade is a bet on the IV path, so measure the IV path directly. The historical chain carries per-contract implied volatility at every minute, so the path is there to read.
Setup Sell the ATM straddle the session before the announcement and buy it back the session after. This is the short straddle held through the event, not a trade entered after it.
Exposure You keep the volatility premium that leaves the options after the event, less whatever the stock moved.
In the sample Eight of twelve events paid, and the mean was positive. The worst trade was a large one: a single gap took back a large multiple of the average win. Selling the event paid on these twelve dates, and the twelve dates include the shape of how it stops paying.
Test next Same as the short strangle: a broad universe, a full year, and a margin model. Then compare the straddle with the strangle on matched events to price the strike gap.
Beyond the 50: structures and programs not in this dataset
No list captures every options strategy. Calls, puts, shares and cash combine across strikes, maturities, quantities and entry conditions. What matters is whether a new name changes the payoff, the signal or the management rule, or only the packaging. None of the following is backtested in this release, and none inherits a result from a similarly named profile above.
Strategy or variant
What changes
What a test has to add
Naked call
Sell a call with no paired put or protective wing
Unbounded upside exposure, margin and liquidation
Naked put
Sell a put without reserving full strike cash
Capital, margin and funding; the cash-secured result does not carry over
Synthetic long stock
Buy a call and sell a put at the same strike and expiration
Financing, dividends, exercise and a comparison with real shares
Synthetic short stock
Sell a call and buy a put at the same strike and expiration
Borrow economics, financing and assignment
Covered put
Short shares and sell a put against them
Stock borrow, dividends owed and upside exposure from the short shares
Covered strangle
Hold shares, sell a covered call and sell an additional put
Funding rate, settlement design and early exercise on American options
Buffered or capital-protected overlay
Cash or bonds plus a defined option package
Product-specific financing, payoff boundaries, fees and reset rules
Some programs need a richer dataset than end-of-day quotes:
Managed spreads and condors. Profit targets, stops, delta adjustments and rolls make a different strategy from a fixed exit. "Close at half the credit" needs an intraday price path to say whether the target or the stop was hit first.
0DTE and opening-range trades. A same-day expiry is a maturity choice, not a strategy. Intraday quotes, signal timestamps and fast-changing exposure are the whole test. The Historical API's minute-level chain is built for this.
Gamma scalping and delta-hedged volatility. A hedged option position's result depends on the hedge schedule, the price path and the cost of every hedge. The unhedged straddle rows above do not measure it.
Dispersion and correlation. Index options against constituent options bring basket weights, membership changes and many simultaneous markets. A single-ETF condor is not a substitute.
Dealer-positioning, gamma-flip and flow signals. These select or manage ordinary structures. Testing them needs time-stamped signal history and an entry after the signal was available, which is what point-in-time replay provides.
VIX options and options on futures. Different settlement, exercise and exposure. Start from the contract specification, not from ETF results.
Use an established reference to check the exact legs behind a label. The OIC catalogue is a good start for conventional structures. OIC strategy catalogue
Backtest it yourself
Every trade in this article was priced from FlashAlpha's options archive: end-of-day bid and ask quotes for the contracts in each month's chain, with the deltas used to select strikes. The Historical API serves the same archive at one-minute resolution, back to January 3, 2017 for SPY, QQQ, IWM and eleven other symbols, with 8+ years of history on 54 symbols and 204 symbols in total.
This study
Three ETFs, one entry a month, close to close
End-of-day chain, contracts picked by delta
Fixed ten-session exit, no signal timing
Five volatility filters as fixed research proxies
With the Historical API
Any minute since 3 January 2017: enter at 10:30, exit at 15:30, or test any schedule
GET /v1/optionquote/{symbol}?at= returns every leg's bid, ask, implied volatility, greeks and open interest at that minute
204 symbols, 54 with 8+ years of history
The signals themselves, replayed leak-free: /v1/strategies/vol-carry, skew, tail-pricing, term-structure and /v1/vrp/{symbol}?at=
Earnings: /v1/earnings/expected-move, iv-crush and history for the implied move and the crush distribution
The Historical API is part of the Alpha plan, from $1,199/mo, with unlimited daily requests on the same key as the live API.
Schedule. Select contracts from the first observed trading day's end-of-day chain each month. Enter at the next observed close. Exit ten trading sessions later, before expiration. No stops, targets, rolls or hedges.
Expirations. Front options target 35 calendar days, within 30 to 45. The back option in calendars and diagonals targets 65 days, within 55 to 80.
Strikes. Delta targets of 50, 25 and 10 mean the nearest archived absolute delta in the chosen expiration, rejected if more than 0.10 away. ATM means the strike nearest spot on the signal day.
Quotes. Every selected contract needs positive, uncrossed bid and ask quotes on the signal, entry and exit days, a signal-day spread of at most the larger of $0.10 or 30% of midpoint, and displayed size covering the trade. If a later quote is missing, the trade is excluded and recorded, not replaced.
Dividends. Windows that span an ex-dividend date are excluded for stock overlays and any short call. Early assignment is not modeled anywhere.
Data. FlashAlpha's options archive, read directly: end-of-day bid, ask and delta for every listed contract in each month's chain, underlying closes and ex-dividend dates. The archive holds minute-level option quotes back to January 2017 and is the same store the Historical API serves.
Size and costs. One contract per option leg. Stock overlays hold 100 shares. The after-cost scenario buys at ask, sells at bid, and pays $0.65 per contract per side. Stock legs cross the underlying bid/ask with no commission.
Signals. Realized volatility is the sample standard deviation of 20 unadjusted log returns, annualized by the square root of 252. Implied volatility is stored as a decimal, so a 0.03 difference is three volatility points.
Risk metrics. Each trade's result is a percentage of the entry value of 100 shares. For each strategy, the months are lined up and each month's figure is the mean across the ETFs traded that month (one to three). Sharpe is the mean of that monthly series divided by its standard deviation, times the square root of 12, with no risk-free rate subtracted. Max drawdown is the largest peak-to-trough fall in the running total of the monthly series, not compounded. Profit factor is the sum of winning trades divided by the sum of losing trades. The 100-share benchmark uses the underlying's midpoint price change on the same windows with no costs. All of it is in risk_metrics.csv and recomputed by the article builder from the per-trade ledgers.
Uncertainty. The 95% intervals in the controlled comparison use 1,000 bootstrap resamples of complete calendar months, keeping the three ETFs within a month together.
Of 216 scheduled ETF-month windows, 133 survived for all six headline structures. The three ETFs contribute unequal counts: SPY 54, QQQ 39, IWM 40. Requiring valid entry and exit quotes can bias the surviving sample. The expanded release holds 7,265 option-trade simulations and 106 wheel cycles across 42 monthly profiles, seven earnings profiles and the wheel. Full methodology · Construction rules for all 50
What is not modeled. This is a retrospective study designed in 2026 with no untouched holdout. It does not model early assignment, financing, margin, collateral yield, taxes, market impact or the daily path of an account. End-of-day quotes are aggregated views without a contributing timestamp. Treat every number as a historical quote simulation with an audit trail, not as a track record.
Downloads
The packs include raw query responses, selected legs, quote lineage, trade results, exclusions, calculation code and offline reproduction instructions. File hashes let you check you are working from the same inputs.
Suggested citation: FlashAlpha Research. Options strategies ranked: a reference guide with 50 light-backtest profiles. 15 September 2026, updated 16 September 2026. Based on flashalpha-options-taster-2020-2025-v1 (audit revision 1.1) and flashalpha-options-expansion-v2.
Questions readers ask
Which options strategy performed best?
On the controlled comparison, the long call had the highest average after-cost result of the six structures tested on the same 133 windows. On the full monthly table, the covered call had the highest mean on its own sample. Both are findings about this period and this metric, not a prediction.
What does the percentage mean?
Every result is a percentage of the value of 100 shares of the ETF when the trade was opened. A covered call at +0.76% on a $500 ETF made about $380 on a $50,000 position. The figure is not a return on the option premium, the margin or an account. The downloadable ledgers keep the raw figures in basis points, and one basis point is 0.01%, so 76.3 basis points is the same +0.76%.
Which strategy had the best Sharpe ratio?
Among the monthly strategies, the vol-carry filtered put spread at 1.18, on 73 trades. The bullish risk reversal and the covered call follow. On the six-strategy matched windows, 100 shares of the ETF with no options scored 0.66, above every option structure tested there.
Which strategy had the highest win rate?
Among monthly profiles, the cash-secured put at 74.1%. Among the six on matched windows, the bull put credit spread at 72.9%. Both rank well below the top by average result, because their losing trades are large.
Are all 50 strategies profitable?
No. Most of the 50 profiles show a negative mean after costs on their own samples. The data shows what each rule did on these quotes over these years. It does not show a persistent edge for any of them.
Is the wheel the same as a cash-secured put?
No. The wheel starts with a cash-secured put but continues into share ownership and covered calls after assignment. Its result depends on how long it holds shares and what they do while held, which is why it has its own ledger and its own scale.
Do these results validate FlashAlpha's strategy endpoints?
No. The five signal-filter profiles use published proxy formulas with fixed thresholds. Validating an endpoint would mean replaying its exact production definition with historical inputs, which is what the Historical API is for.
Where does the data come from?
FlashAlpha's options archive: minute-level option quotes, implied volatility and greeks for 204 symbols, back to January 3, 2017 for SPY, QQQ, IWM and eleven others. This study used its end-of-day view. The Historical API serves the same archive at minute resolution, on the Alpha plan.
Where should the next hour of research go?
Pick one family, open its ledger, sort by loss and read the dates. Then change one rule and rerun. A matched comparison of two rules beats another leaderboard.
