Skip to content

40 Comments

  1. uno
    July 10, 2020 @ 8:21 am

    Nice. 0 DTE -16D seems to be the sweetspot from risk-reward point of view. Is 0 DTE is opening a day before expiration before market close? BigErn does 3 DTE

    Reply

  2. uno
    July 10, 2020 @ 8:23 am

    Forgot to ask is this backtest applicable to SPX?

    Reply

    • spintwig.com
      July 10, 2020 @ 11:12 am

      0 DTE is opening the day of expiration. Indeed, he does.

      Yes, SPX results should be very similar.

      Reply

      • Matty G
        July 10, 2020 @ 2:53 pm

        Isn’t SPX quite a bit more dangerous because the options expire on a Thursday but are priced based on Friday’s opening?

        Reply

        • Russell
          July 10, 2020 @ 4:44 pm

          I think AM-settlement is for SPX options rather than the SPX weeklies (SPXW) which are PM-settled. Maybe I’m misunderstanding?

          Reply

  3. Steve
    July 10, 2020 @ 3:56 pm

    How did the algorithm work on the 0 DTE? For instance, if you were looking for a 16D and it wasn’t available because of the high Gamma did you go out to the next expiration and look there up to the 3 DTE or was 3 DTE only for Fridays? Since you were only holding one position at a time, is it safe to assume you waited until near close on the next day to look for your next position?

    On the 0 DTE, 16D and 30D what was your average premium collected?

    Reply

    • spintwig.com
      July 11, 2020 @ 11:53 am

      When looking for a 16D position, the tolerance is +/- 6 delta. So anywhere between 10-22, closest to 16. Tolerances are different for different delta targets.

      The approach is to first look at DTE targets then look at a delta targets. If a 0DTE position with a delta between 10-22 doesn’t exist, then look at a 1DTE position with the a delta in the range and choosing the one closest to 16D. Repeat this process until all DTE targets within the range have been explored. If after all DTE positions have been explored and there is not a suitable position, take no action – a new position is not opened that day.

      There are scenarios where more than one position is open at a time. For example, if on Friday only a 3DTE / Monday-expiring position satisfies the requirements, such a position is opened. Then on Monday the 0DTE same-day expiring positing satisfies the requirements, that position is opened. At this time there are now two open positions.

      Avg premium collected is in the “Premium Capture“ chart and table in the results section.

      Reply

  4. Eric
    July 14, 2020 @ 4:30 pm

    I’m curious if the results would be similar using put credit spreads with the 0 DTE strategies. For someone with a small account this is more appealing to me and I’m wondering what the downsides of doing a spread would be opposed to just selling a put. Is this something you would look into in a future study?

    Reply

  5. _Eric_
    July 21, 2020 @ 2:55 pm

    Hi. Awesome stuff! I was wondering if you could provide more context around the drawdown percentages. I want to make sure I understand this correctly and put a dollar amount to the drawdown.

    Lets say the SPY is trading for 326. You are running a campaign, selling the 0 DTE, 16 Delta puts everyday. Would the notional value be $32,600 per put sold? Since the max drawdown for 0DTE, 16D is 9.36%, does this equate to a draw down of $3051.36 per option sold?

    Thanks!

    Reply

    • spintwig.com
      July 22, 2020 @ 1:06 pm

      Yes, notional on a 326 SPY put would be $32,600.

      Drawdowns are measured as a portfolio performance metric as opposed to an option position performance metric. That is, 9.36% max drawdown is the difference between the portfolio value at the bottom of the drawdown and portfolio value at the peak from which the portfolio fell.

      There are many drawdowns as the portfolio ebbs and flows. 9.36% happened to be the largest drawdown in this example.

      Because the drawdown is relative (i.e. a percentage) and not absolute (i.e. a specific dollar amount), it’s not possible to put a specific dollar amount to it given the data available. For example, a 10% drawdown would be viewed as more severe than an 8% drawdown. However, more dollars in an absolute sense could be lost on an 8% drawdown.

      Suppose a 10% drawdown occurred early when the portfolio was, say, 10k which results in a 1k loss. The portfolio eventually recovers and grows to 100k then experiences an 8% drawdown. This results in an 8k loss – more dollars were lost but it represents a smaller percentage.

      Reply

  6. _Eric_
    July 22, 2020 @ 5:21 pm

    Thanks for the detailed reply. I’m trying to figure out how I would size this strategy. I guess a better question would have been, if notional value is 32,600, how many puts are you selling per 32,600 in your test? i.e. What is the leverage factor?

    Sticking w/ the 0 DTE, 16 delta:

    I see the minimum starting capital was 11,000. I’m assuming this minimum number is the margin required + a drawdown buffer that would allow you to keep trading. Is it safe to say this is 3x to 5x leverage? When would you add more contracts? every time the account increased by 11 k?

    Is 11k the starting number that the CAGR was calculated on?

    Thanks and sorry for all of the questions. Maybe I’m just having a dense day.

    Reply

    • spintwig.com
      July 23, 2020 @ 11:38 am

      Correct – assuming a start in Feb 16 2018, $11k was the minimum amount of capital needed to write a 0-3 DTE short put every day without blowing up [within the assumptions of the backtest].

      Max margin utilization is essentially 100% or 5x leverage, whereas the average margin utilization is essentially 30% or 1.5x leverage. From time to time there would be 3 concurrent positions open. More often then not, there would be only 1-2 positions open at any given time with the 0-3 DTE strategy. This is why the max margin utilization and average margin utilization are so far apart for this particular strategy.

      Yes, the CAGR was calculated on the 11k starting capital.

      Reply

  7. TFJ
    July 27, 2020 @ 12:15 am

    I’m having trouble understanding your 45 DTE curves. What is the difference between the 5D-50 (red here) and 5D (purple)?

    Reply

  8. RI
    August 2, 2020 @ 5:28 pm

    For 0DTE – 5D, is there a reason the premium capture is around 56% even with a 98.5% win-rate – does the 1.5% loss rate reduce the premium capture so significantly?

    Reply

    • spintwig.com
      August 2, 2020 @ 5:57 pm

      Yes, the 1.5% of trades that were losers offset nearly half of the profits earned from all the profitable trades.

      Reply

  9. Joe Jefferson
    August 4, 2020 @ 12:25 pm

    I see “Timing 3:46pm ET” — does that mean for the 0 DTE strategy, the following statements are true?

    On Friday, 3:46PM a short put is opened with a Monday expiry
    On Monday, 3:46PM a short put is opened with a Wednesday expiry
    On Tuesday, 3:46PM a short put is opened with a Wednesday expiry
    On Wednesday, 3:46PM a short put is opened with a Friday expiry

    etc…

    Forgive me if I just missed it in your amazing writeup!

    Reply

    • Joe Jefferson
      August 4, 2020 @ 12:26 pm

      Oh man, the formatting did not come through, sorry!

      Reply

      • spintwig.com
        August 6, 2020 @ 3:46 pm

        No worries – it came through formatted when I received the email alert 🙂

        For the 0 DTE strategy, a position is opened Mon @ 3:46pm with a Mon expiration (it’s essentially a 14-minute option, ignoring after-hours trading).

        There are no Tuesday-expiration options so on Tues a position is opened 3:46pm and are generally a Wed expiration. I use tolerances on the delta and DTE targets to help ensure a position is opened daily. This mitigates much of the timing luck issue while also capturing the “intent” or “spirit” of the backtest.

        Wed comes around and a new position is opened at 3:46pm for expiration in 14 min (and there’s also Tuesday’s position that’s expiring).

        Suppose Friday comes around and there’s no position that can be opened within the specified delta target range. I allow up to 3 DTE of tolerance so the strategy can open a position that expires Monday. Also, various holidays and quadruple witching can force a non-same-day-expiring position to be opened.

        The methodology you cited is BigERN’s strategy – I backtested that as a guest post on his site: https://earlyretirementnow.com/2020/06/17/passive-income-through-option-writing-part-5/

        Reply

        • AP
          April 8, 2021 @ 6:46 pm

          Not following the criteria for non-expiration days, ie: opening on Tuesday (since no expiration) for Wednesday expiration.

          I can’t imagine you would be opening 16D/30D/50D trades on Tuesday at 3:46 for Wed EOD expiration. With 5x leverage a single sell off Wednesday morning would wipe out years of returns. Today, a 17D trade for Wednesday expiration for SPX is at 4075 (SPX closed at 4097.22). Even a 1% movement down (to 4056.25) would lead to an overnight loss of ~1,875.22 and that’s just from a 1% movement. Make the movement larger and the loss multiplies to 10k+ before you know it. If this is the case and somehow in the end the profits do end up being higher than losses, the bulk of the profit seems to come from the overnight trades rather than true 0 DTE expiration because a 16D trade for overnight is going to provide a significantly higher premium than a 16D trade for expiration in 15 minutes.

          I’m also not exactly following how you can maximize margin utility to 100%. If you’re at Tuesday, you open a position for Wednesday expiration while using 100% margin you would have to close the position before 3:46 at potentially unfavorable prices to be able to maintain 100% margin utility.

          Maybe I’m just dumb and don’t understand (very possible.)

          Reply

          • spintwig.com
            April 9, 2021 @ 12:36 pm

            The strategy seeks to open a position daily in order to mitigate timing luck. In doing so, not all positions are exactly 0 DTE (same-day expiration). For example, a position opened on Tuesday will be associated with a Wednesday expiration. Thus, the position will be 1 DTE.

            Hindsight bias was used to identify the minimum amount of starting capital such that the max margin utilization attribute would not exceed 100%. Optimizing to maximize *average* margin utilization by tuning position size and trade timing at each individual trade is possible per the mechanics you mention re: exit existing position at 3:46pm but would be an even more egregious use of hindsight bias than optimizing for max margin utilization.

            True. In practice one will not know what the future holds so it wouldn’t be advisable to run a strategy with a max (for average) margin utilization target of 100% lest there be a sell off.

            Reply

  10. William
    August 8, 2020 @ 6:07 am

    Hi Spintwig,

    Thanks for the great work as always, really appreciated the insight. However, I suspect the outperformance could be due to the timing of the position opening – the last 14 minutes of the day could be more tranquil than other periods in the day.

    Would it be possible to run the 0-DTE backtest by opening positions at the start of the trading day?

    Reply

    • spintwig.com
      August 9, 2020 @ 11:32 am

      Great point William! This is a fine example of timing luck potentially impacting results.

      Short answer is unfortunately “not at this time.” It turns out my data sources only offer a single price snapshot for a given day. There are data sources that have 1-minute granularity but they’re expensive to acquire and crunch. Perhaps one day this will be a backtest configuration I can accommodate.

      Reply

  11. Semper
    November 2, 2020 @ 10:50 pm

    Does 25% profit apply to if it declines by a value of 25% profit as well too?

    Reply

    • spintwig.com
      November 3, 2020 @ 4:21 am

      Positions were not exited based on losses. An exit on a 25% decline in value would be synonymous to a 0.25x stop loss.

      Reply

  12. Ed
    November 20, 2020 @ 7:35 am

    Just came across this – thank you for the great work. I haven’t seen anyone really talk about this strategy or similar elsewhere (other than ERN blog). Is be interested to hear if you have any thoughts about the mechanism or an explanation for the success of the shorter Dte strategies?

    I am somewhat new to options trading so forgive me if this makes little sense, but do you think it could have to do with the expansion in implied volatility during turbulent markets? Ie the time value of longer dated options rises far more than it does for shorter dated ones, causing bigger drawdowns and margin hits.

    Reply

    • spintwig.com
      November 22, 2020 @ 10:26 am

      Thanks for stopping by! I don’t have a definitive reason, but my speculation is that

      1) the 0DTE play is less saturated and thus has less structural arb happening to keep prices exactly where they should be (as compared to, say, 30DTE or 7DTE horizons) and

      2) if the trade goes “bad” in that realized vol exceeds implied vol, the amount of vol that can happen in a day is less than that of a week or 45DTE. Thus, losses are capped to a single day’s worst move. Yes, longe-dated options are priced efficiently with more premium to account for this and have a longer runway before the position goes ITM. However, if a substantial fat tail happens like Feb/March 2020 then the 0DTE options can only lose for a day and the next trade takes advantage of the now-significantly-different greeks.

      Reply

  13. Kuba
    January 4, 2021 @ 1:13 pm

    Can similar results be applied to 0DTE 16Delta credit spreads? Or how they change?

    Reply

    • spintwig.com
      January 6, 2021 @ 2:29 pm

      That’s a good question. I haven’t yet looked at the viability of 0DTE spreads. It’s on the roadmap but remains a little ways away due to the QQQ 7DTE leveraged backtest that next, followed by a segment on client backtests that were approved to be published.

      Reply

      • Kuba
        January 6, 2021 @ 2:42 pm

        If I can help you somehow, it would be my pleasure.

        Reply

        • spintwig.com
          January 7, 2021 @ 2:06 pm

          Help me keep up with all the custom backtest requests so I can allocate more time to public-facing content 🙂

          Reply

  14. M
    January 10, 2022 @ 6:33 pm

    I’m kinda torn on what the reality is – the Tasty Trade program seems to be selling 45 DTE, and exiting at 21 DTE.

    The data that I’ve seen between you and ERN indicates 0 DTE seems to be best in terms of return / volatility.

    I’m leaning more towards you guys and selling as close to 0 DTE, bc it seems like you have the data while Tasty Trades will throw up some abstract chart that says theoretically what they think the option greeks do, without showing us the numbers to back up their back tests.

    Reply

    • spintwig.com
      January 14, 2022 @ 9:27 am

      There is a newer study that compares 7, 45, 90, 180 and 365 DTE (LEAPS) at https://spintwig.com/spx-short-put-7-365-dte-leaps-s1-signal-options-backtest/

      Early management isn’t explored but it does compare all the durations from Jan 2007 through Oct 2021. This should help with the analysis. CAGR is greatest and annual vol is lowest for the unmanaged 7 DTE strats. I personally trade 0 DTE and BigERN does 2-3 DTE.

      Agree that the tasty trade “research” is vague and lacking in details, has limited explanation of methodology, and no transparency (source data not readily available).

      Reply

      • M
        January 16, 2022 @ 3:38 pm

        Thank you.
        Question: when you are trading 0 DTE, are you doing daily, or are you following the S1 = True signal? I understand that daily seems to have a higher CAGR, but s1 = true seems to have lower vol, so I’m curious about your personal preference for your own program

        Reply

        • spintwig.com
          January 17, 2022 @ 2:06 pm

          When I’m trading, I trade most expiration days – namely M/W/F. Sometimes I sit out for a while. I sat out nearly all of Dec + the first week of Jan. Last week was profitable.

          If I was doing longer durations I’d stick with s1, hands down. For 0 DTE specifically, the jury is still out. There is far less data for 0 DTE strats, between lack of intraday data (I only have a single price snapshot per day as opposed to 1-minute granularity) and lack of historical data (Wednesday and Monday-expiring SPX options didn’t exist until 2016).

          I don’t trade longer durations because 1) it’s not a good fit for my trading style / personality (the human factor will get in the way too much), and 2) I have no desire or need to hold a leveraged position overnight and don’t want that kind of risk on the books.

          Reply

  15. Cesar Espinoza
    September 8, 2022 @ 4:50 am

    Any particular reason you don’t test deltas above 50? Wouldn’t that be useful?

    Reply

    • spintwig.com
      September 8, 2022 @ 10:37 am

      Good question. Strategy performance, when delta targets are above 50, are primarily driven by intrinsic value (i.e. value of the underlying). The higher the delta, the closer the PnL curve gets to that of holding long underlying (minus frictions).

      Reply

Leave a Reply

Your email address will not be published. Required fields are marked *