RichardGibbons

I mean that I'm not buying VIX calls because the VIX is at 62, and it's unlikely to get much beyond 80. What's more, the VIX options are European exercise, which means that they converge to intrinsic value only right at expiration, which makes them a bad speculative bet right now (because they won't move much compared to how much an American exercise option would.) Betting on the VIX going down is perhaps a more reasonable bet, but it's also hard to make that bet because the timing's tricky. And if you go way out in time--like a year out--the European exercise makes it hard to take profits quickly. Oddly enough, I had VIX $16 Mar 18 calls that I bought for $1.90, and sold for break even a week and a half later in late Feb, right before the market plummeted. I was getting cute, hoping the VIX would fall a point so I could rebuy slightly farther out in time for the same price. Didn't happen. I wouldn't have held all those calls to expiry regardless, but on the day of expiry, the VIX was above 80. But a potential (80-16)/1.90 = 33 bagger was there. Oops. (New rule: if the market's obviously going to crash because of a massive pandemic, don't trade away your hedge in order to pick up pennies.)

I have been buying SPY puts (in tiny quantities... not a serious hedge) because that's where the liquidity is. I'm not doing the VIX because it's too high to make bearish bets, and I'm not doing the Dow because I'd rather not bet on 30 stocks.

Interesting article. If the authors are right, it's a game-changer. What's the push-back to this? What are they missing? The thing they're missing is that their model doesn't match the evidence. In the UK, they're testing 8000 people a day, but until yesterday were finding fewer than 1000 cases per day. Presumably they're testing the people most likely to have COVID-19, yet far fewer than half the tests come back positive. The only way that half the population could be infected and yet only a small percentage test positive is if either the tests are far more likely to have a false negative than a correct positive, or if the people they're testing are far less likely to have the disease than everyone else. By far, the most reasonable hypothesis is that their model sucks.

Thanks for your honest response--I really do appreciate it when people are willing to act with intellectual honesty. So much of that is missing in the world today. That said, your 86% asymptomatic rate isn't credible. The most credible number I've read is ~30% (though at this point, I'd be delighted if it were 86%. Asymptomatic people are bad early in a pandemic, and good when it's out of control.) In any case, we'll have to keep watching New York to see when they get overloaded. New York's normal capacity is about 60K hospital beds and 3000 ICU beds. So, I'll assume that if we go over, say, 70k simultaneous hospitalized COVID-19 patients or 3300 simultaneous COVID-19 ICU patients, that means that you'll have recognized that "this isn't a big deal, there are already hundreds of thousands if not millions infected and there's been no problems" analysis was incorrect. It's nice to have goal posts planted firmly in the ground, and clearly if COVID-19 patients alone--without even taking into account people in the hospital for other reasons--exceed hospital capacity, then it was a pretty big deal. All that said, I'm pretty surprised that you don't understand why people would care much more about an epidemic than heart disease deaths.

Wow. I find it bizarre that USA is on track to be the worst hit country on earth and several people on this board are absolutely convinced that the federal leadership has nothing to do with that.

Hey Orthopa, you said that when you realized that you were wrong, you'd eat crow. I don't care about you eating crow, but took that as a statement that you'll be intellectually honest enough to change your hypothesis when the evidence contradicts your view. So, I'm curious when you'll decide that your hypothesis is incorrect. Quite a while ago, you seemed to believe COVID-19 would likely not be a big deal because you believed that there were already millions of cases in the USA (but almost none of it had showed up at the hospitals)? Have the exploding number of cases in New York changed your perspective yet? (Note that right now, New York state has 5% of the worldwide diagnosed COVID-19 cases, with 2,635 patients bad enough to be in hospitals, 621 in the ICU, and 114 deaths.) If you haven't changed your mind yet, do you have any thoughts about evidence that would make you change your mind?

The irony is that axing the CDC expert probably wouldn't have made a difference, as Trump ignored everything until March anyway. There's no reason to believe that an additional CDC expert would have resulted in Trump being less stupid or pigheaded.

So, I'm curious how you resolved the cognitive dissonance between being a passionate Trump supporter on China with the belief that the government should try to keep away from business. It feels like, if America wants to go this path of eliminating China from supply chains, it requires either massive tariffs or extreme regulations--basically the government completely destroying the supply chain of many businesses. (FWIW, I agree with you that Western countries ought to disassociate from China for all the reasons you say, but I can say that because I'm fine with government interfering with business for the greater good. But I don't see how one solves this problem if one believes the non-interference of government in business is the greater good. Can you enlighten me?)

Yep. By my calculations, it'd be roughly 3 months assuming 20K people infected now. (20K * 2^14) = 327M. We'd need 14 doublings to infect everyone in the USA, and we double every 6 days, so that's 6 * 14 = 84, or 12 weeks. Plus, say, a month for the disease to go from incubating to healed. Of course, this would result in lots of unnecessary deaths, but it's probably optimal for the economy (barring second order effects like a revolution from people being annoyed at the widespread deaths and the government seemingly not doing anything.)

I try to be evidence-based (don't always succeed), and since that post I've got substantially more evidence. Cobafdek's explanation makes sense to me. Essentially, when it comes to math, until I have evidence to the contrary, it sounds like I should assume that a typical doctor is roughly as knowledgeable as someone who's never taken a math course in college. And that's fine. If you're a surgeon, I imagine I as a patient gain more from you doing another medical course than a statistics course. It just means that, I should have little confidence in your analysis of math things like statistical sampling and exponential growth, just like you have no confidence in me for medical things. (I have a math degree and a couple computer science degrees. I'm completely ignorant of how to operate on people, and basically trust my doctor 98% for anything medical. So you're welcome to call me ignorant too.) Sure. While people talk about average rate to double of about 6 days, there's high variance in the number of people infected by any given person. The typical infected person infects 2 others, but one person in South Korea seemed to have infected a thousand people. So extrapolating doubling from tiny numbers doesn't work. Once you get a bunch of cases (say, 40), the infection rate converges on the average, and you can start to play the 6 day doubling game. (Which, as a percentage of the population starts to take social distancing seriously, should increase, decreasing infection rate.) For the USA, it's also worth noting that deciding who to test based on different criteria than other countries will affect the results of tests, and that seems to be true in the USA. (e.g. suppose you test one person, and find them positive, and then never test anyone again, identifying no more positive people. Is it reasonable to conclude that the virus has stopped spreading? USA isn't quite this extreme, but it's doing way fewer tests, particularly per capita, than pretty well every other developed country.) LOL, for a few years in middle school, one of my nicknames was Gibble Dick. That was unfortunate. :)

Yeah, basically one of several reasons South Korea had fewer problems was because the did a bunch of testing early on, basically the stuff some people were saying was a waste of time (which probably is a waste of time now). That said, the most fascinating thing of this thread today is that a professed MD on an investing board seems to believe there's little difference between cases doubling in 3 days versus 6 days. If we're talking 55 days since the first case, that's the difference between 2^9 = 512 cases and 2^18 = 262,000 cases. Is it really possible to become and MD without a basic understanding of exponential growth and no understanding of statistics/sampling theory? (This is a serious question, not rhetorical, because I don't know the answer and I'm curious if such big holes are normal in doctors' education.) (Cobafdek: this is why I'm at the 2% rather than the 70% number for orthopa's theory. If someone doesn't understanding even the most basic concepts of exponential growth or statistics--ideas you'd learn in your first year courses--them, when it comes to a pandemic, their hypotheses about the meaning of anything they observe are likely to be worthless.)

I'd put the chance that Orthopa is right at less than 2%. But I've been continuing to mull the issue, since it's an interesting thought experiment--how can Orthopa's data point be reconciled with the 100 data points that contradict his claims? One way it could be true is if the area around Orthopa previously had a coronavirus that infected people and gave them some sort of heightened immunity compared to everywhere else. Or, maybe the COVID-19 came early to his region, but was a mutated version that happens to have a much lower rate of serious consequences. That said, I think both of these cases are super low probability, that it's much more likely that multitude of experts saying "this is a big deal" are right, and Orthopa isn't. In fact, in my case, I view the evidence Orthopa's brought to support his argument as weakening his argument since it's showing that he's confidently making large, unwarranted leaps to support his thesis. To me, this increases the chance that he's a guy who's comfortable squeezing evidence into odd shapes in order to support his conclusions. (e.g. a few days ago, 2 cases was enough for him to extrapolate conclusions about 100,000 people infected. Today, he's saying that 140,000 tests isn't enough to extrapolate anything.) At this point, I'm curious whether he's a troll or just completely locked into an incorrect mental model. I still lean toward the latter.

Over the last couple of months, I've bought calls on the VIX and sold puts on SVXY, all profitably. The problem with the VIX calls is that they're European exercise, so I ended up selling them for less than I could have because there was still a few weeks of time value left in them. The issue with VXX puts are that when the VIX is high, the futures flip from contango to backwardization, which basically means that until the VIX gets below, say 22, there will be a built in upward bias to the VXX (kind of like the VXX is short something that has time decay.) So it's harder to make money there than you might think. Also, the return of the VIX to sub-20 will likely be slower than you think. Recently, I've been limiting my options bets to 1-2 week options on SPY. They've also all been profitable, but I think my overall strategy would be unprofitable, since I've been flipping them for small profits before I get massive wins (and one 100% loss would wipe out all my wins.) Still, if I believed the virus were to be proven no big deal in 3 weeks, I'd buy at the money calls about 3 weeks out. (Because with implied volatility skew, the price you pay right now for 3 weeks isn't that much more than the price you pay for 1 week.)

Yeah, Dalal, I know it wasn't random testing. But right now, they have identified 2 cases that meet the criteria. If they test 10,000, they can see whether or not the "2 cases ==> 100,000 cases" is reasonable. (i.e. if they test 10,000 and find no cases, they'll have enough evidence to prove that whoever wrote the paper saying "two cases implies 100,000 cases" made a colossal error.) I suppose it would also say something about the number of people who have contracted COVID-19, but my point was just that it can used to refute or bolster the paper's bizarre "2 cases implies 1%" theory.

Thinking about Ohio a bit more, this is actually a great time to test that 2 leads to 1% infection rate hypothesis. If they randomly test 10K people, they can get a 95% confidence interval on the 1%, providing decent evidence of the original claim. If they test 20K people, they can do better than a 1% confidence interval. Too bad that during pandemics, nobody's thinking about expending medical resources to test theoretical hypotheses. :)

Well, Ohio says if you can identify two cases where you can't identify the cause then it means you have 100,000 cases. That may be true, but it could also mean that the 2017 Morbidity and Mortality Weekly report doesn't actually know what it's talking about, or that it's being misunderstood. To me, to extrapolate 2 to 100,000 for any disease seems like a huge leap, and I'd have trouble designing an experiment to prove that hypothesis, so that makes me skeptical. It would be interesting to know the reasoning behind that conclusion. Regardless, if it's true, I agree with you, Orthopa, that it's a reason to be extremely optimistic. So I've been trying to answer my own question, thinking about scenarios where there would be low fatalities in the USA, and high fatalities in Iran/China/Italy. I wonder if it could be the intersection of COVID-19 with the flu or some other disease. Like, suppose there's another disease that has greatly weakened the immune systems of people in specific geographic regions, but not easily contagious across broader geographies. I have no idea if that's possible, but it would explain the discrepancy. The hypothesis I've seen on the low death rate in Korea is that patient 31 attended a massive gathering of a Christian group that focused all its recruitment on a young demographic. That resulted in the infections of thousands in the group. So, there was massive testing of a large group of young people, ensuring that positive tests came from a population where the mean age was far younger than the general population, a population where the death rate is close to negligible. It's worth noting that, if your hypothesis is accurate, Orthopa, there is probably a fast 5-10 bagger to be made in call options with an expiration about 3 weeks from now.

Orthopa, the problem is that what you're saying doesn't seem to align with evidence. It seems fairly clear that a bunch of people have died in Italy, Iran, and China as a result of COVID-19. You seem to be claim is that millions in the USA have been infected a long enough time ago that we'd already be seeing lots of deaths if COVID-19 were a big deal. But USA has not seen lots of deaths. So, to be credible, you need to make it simple for us to understand this disconnect. Are Americans just more robust than the Italians, Iran, or Chinese? Do Americans have some sort of herd immunity that makes them less likely to die? Are Italy, Iran, and China simply pretending to have all these deaths, when really, they don't? Is there something about American culture that allows millions to catch COVID-19, but nobody to die? If you don't have some explanation for this disconnect between your hypothesis of millions infected but nobody dying, the most reasonable thing for people to believe is that your hypothesis is wrong. Particularly considering that there doesn't actually appear to be any evidence for your hypothesis except "some people got sick this flu season and didn't die, and it's conceivable that those people had COVID-19". (That said, I don't think you're ignorant. I think you've got the "I'm smart and know a lot about the topic, so my hypothesis unsupported by evidence must be right, and I'll defend it unto death" thing going. Pretty well all smart people make that mistake occasionally.)

I don't. I just assume if the quantity I have invested in dividend-paying investments is greater than the amount I have in margin that I'll be fine. (That said, it's never even close to being an issue, since I rarely use margin and when I do, it's easily less than the 5% of my portfolio range.)

Pushing the infection rate curve down could make a big difference if seasonality matters for this disease. Viruses tend to not like warm summer weather. If that's true of Covid19, delaying infection be even a few weeks could make a massive difference if the disease effectively becomes far less infectious during the summer. It would mean less strain on healthcare resources and would provide much longer runway to work out treatments/vaccines. I wonder if the people who don't see value in preventative measures simply don't understand exponential growth (compared to little growth in healthcare capacity). But maybe I'm wrong, because that would be an odd thing for people on investment site not to understand.

I can't! I think both countries should be limiting travel internationally for at least a few weeks. Citizens returning from overseas (particularly problem areas) should be quarantined and tested. Yeah, me too. They should be blocking international travel. BC has had better testing than the USA, but the federal government doesn't seem to be able to walk (deal with blockades) and chew gum at the same time (implement anti-coronavirus strategies). I think the problem is that Trudeau generally cares much more about people's perceptions of him than concrete solutions to real problems. No_free_lunch, you believing that criticism of the US response indicates support for the Canadian response is most likely a result of your bias.

Yeah. It isn't. In the Canadian system, it's all about getting diagnosed, going to your appointments, and getting the help you need as you need it without being bankrupted. That said, the parking at hospitals can be pricey--like $15 a day!

I'd say that America at this point has one of the best western cultures for encouraging the spread of the disease. Large structural barriers in place to prevent testing Monetary incentive not to be tested Potentially massive personal costs/bankruptcy should one be hospitalized "Continue working while sick" culture Leaders who will spread disinformation about the disease, and be believed by a significant percentage of the population Once individual cases are identified, I'd expect a solid medical response in those cases, but I suspect the factors above mean that the USA is pretty likely to get something like Wuhan where hospitals are overwhelmed. And then I think from there, second-order effects will cause bad things to happen to the economy. From an investment perspective, the thing to worry about isn't a loss of a quarter or a year's discounted cash flow. It's the second-order effects like debt rollovers, supply chain disruption, and layoffs due to the slowing economy leading to unemployment spikes, consumer fear, and demand disintegration.

I'm curious if the researchers realized that this wasn't risk-neutral because of opportunity costs, or if the writer just wanted to make things "simple". If you're in a company with a high ROE, then a 25% chance of a 300% return after 3 years (or 75% of nothing after one year) isn't even close to a break even proposition.

I did Andrew Ng's machine learning courses at Coursera. It was well worth it. I went in with little knowledge about the topic, and, after doing the courses, I feel like I have enough knowledge of the topic to muck around on my own. So, it was basically exactly what I was hoping for. The main thing that it was lacking--compared to a regular course--would've been a large project. Pretty well all the "homework" could be done in an hour or so, but was so targeted toward individual lessons. There wasn't really a big project to bind it all together.

Yeah, it seems pretty clear to me that AI companies have worse economics that SaaS, but SaaS have better economics than any other business. That said, I think there's a chance that an AI company evolves that has a stronger moat and better economics than any SaaS business, but I'd expect the average SaaS business to have way better economics than the average AI business.