Yeah, slept on this too, sorry if I rant... Bottom line if you plan to buy options and use any target based strategy you're very likely to pay a net premium at your loss. In this case the strategy also goes against the general decay direction somewhat so it's like betting against the market by buying a put while also betting on it going up at the same time buying a call... So you need a major swing in your favor i.e. unexpected IV surge, which for UVXY is a VoV in increase. The largest I've seen is a 4X increase btw, during COVID, and premium is already pretty hefty on these. But the rareness takes care of that. Maybe you could consider buying a ratio to gain from more mild swings, or doing a risk reversal or something like that? UVXY (well, the 5 LongVol ETFs that matter) are my specialty btw And note UVXY had a major change in structure just 3 weeks after volmagedon it went from 2x to 1.5x feb 2018. So be wary backtesting this
ORATS.com has a backtester that has UVXY. We do not have your parameter to hedge. For splits, our backtester exits the position the day before and gets back in the day after. This is to avoid holding non liquid post-split options, and waiting until the new series of options is posted after the split.
I looked at ORATS back when I started but decided it may be too expensive vs the detail I wanted. I may need to reconsider. When doing a split what is the assumption about price?. Is the mark for execution just (bid + ask) /2? Does it use a last trade if there is one? Both of these will lead to inaccuracies as rolling a trade this far out on most tickers will cause huge spread slippage. I like the idea of skipping over a split day otherwise. Wasn't ORATS the big player / one of the first to offer backtests?
Still digesting your thoughts, but I’m wondering first off if the long call hedge is the problem here, as I have made some good returns in the past holding UVXY long puts for long periods of time. I know this sounds like curve-fitting, but what if I do nothing but hold long UVXY puts bought on the day of issuance, splitting the bid ask, and held until expiration day and then auto-exercise, presuming they are in the money. Can you also tell me how you are backtesting? The software or program.
Yes the call side is a loser and the put side gains a little. What do you mean by curve fitting? I believe the gain was about 10%>. Interestingly the weeklies are much closer to 0 expected profit. My strategy is more short underlying/short calls... But not exclusively. It depends. The backtesting is 100% my own creation. It's 5TB of data using the NBBO feed of every historical bid/ask data by the minute going back as far as 10 years. All expirations and strikes. The code is python mostly but the logic and data curation is what matters. For instance all strategies go through sizing and monte Carlo optimization automatically. I am forced to be a little cryptic because the sellers of that data are really uptight as data sharing is highly regulated for a fee, but I can share summaries of research results.
Really, only 10 percent per two year trade! I must have been lucky with the trades that closed in January and are slated to end this coming January. Must have missed or timed the VIX spikes just right. I was joking about curve-fitting meaning optimizing a strategy to meet a predetermined goal. I’ve been burned shorting calls. I just wasn’t patient or bold enough to hold though the VIX spikes. As for short stock, IB has annoyed me with UVXY in the past by rating borrow rates, closing positions, etc, so I avoid that.
So, I deal with this all the time and I do not pretend to be a master shorting vol but there is a few more optimal strategies. If you look at history the levels are very predictable. Remember when you are selling LongVol ETFs you are essentially selling SPX puts far otm, rolling then every day to maintain a ~30-60 days to expiration. So decay is a constant for that reason (contango on VX futures). Then you have decay from the daily resetting. Shorting calls adds even more decay. So yeah you are going pay dearly in asymmetrical losses. The reason people lose is simply sizing, getting forced out of positions margin called and all. I hold about 10X the position size to not run out of funds and even that is tough. If vol goes over 100% I have to hedge. Hedging is key to keep margin fixed. I only hold -200 shares short UVIX and -100 UVXY currently. In a surge I may buy VXX put spreads or butterfly's. There's too much to talk about there.
I think the reason I prefer the long put UVXY LEAPS and maybe also long SVXY is the defined loss on a spike without hedging. Totally get the position-sizing and risk-management. See you at the next spike.
50% off for ET subscribers. Yes, we have been in the backtesting game for 14 years. https://orats.com/university/custom-backtesting Slippage Slippage is the extra amount paid or the smaller amount received for an option bought or sold compared to the middle of the options bid-ask. We use a percentage of the options bid-ask spread to represent this amount and the percent slippage is different depending on how many legs there are in the trade. The slippage formula to buy is: Bid + (Ask - Bid) * slippage% The slippage formula to sell is: Ask - (Ask - Bid) * slippage% The following table explains the default method to calculate slippage on entering and exiting a trade: # of Legs Slippage % bid x ask Buy Trade Price Sell Trade Price 1 .75 1.20 x 1.40 1.35 = 1.20 + (1.40-1.20) * .75 1.25 = 1.40 - (1.40-1.20) * .75 2 .66 5.10 x 5.90 5.628 = 5.10 + (5.90-5.10) * .66 5.372 = 5.90 - (5.90-5.10) * .66 3 .56 4.20 x 5.20 4.76 = 4.20 + (5.20-4.20) * .56 4.64 = 5.20 - (5.20-4.20) * .56 4 .53 8.50 x 10.30 9.454 = 8.50 + (10.30-8.50) * .53 9.346 = 10.30 - (10.30-8.50) * .53
That didn't format nicely..can you post the table as an image? What is the slippage % , how do you calculate it or does the user enter it? Would be nice to see in practice and match up trade execution to the NBBO quote and have it tell us the reality, especially as the past was less liquid, potentially creating false opportunities in backtest