Charting Is Mostly Superstition 2020-08-23T20:26:15.474Z
Market Misconceptions 2020-08-20T04:46:21.991Z
The Wrong Side of Risk 2020-08-16T03:50:03.900Z
How to Lose a Fair Game 2020-08-14T18:41:13.638Z
Repeat Until Broke 2020-08-13T07:19:30.156Z
You Need More Money 2020-08-12T06:16:09.827Z
When is the right time to hole up? 2020-03-14T21:42:39.523Z
Do 24-hour hand sanitizers actually work? 2020-03-01T20:41:43.288Z
gilch's Shortform 2019-08-26T01:50:09.933Z
Stupid Questions June 2017 2017-06-10T18:32:57.352Z
Stupid Questions May 2017 2017-04-25T20:28:53.797Z
Open thread, Apr. 24 - Apr. 30, 2017 2017-04-24T19:43:36.697Z
Open thread, Apr. 17 - Apr. 23, 2017 2017-04-18T02:47:46.389Z
Cheating Omega 2017-04-13T03:39:10.943Z


Comment by gilch on The tech left behind · 2020-11-18T02:33:34.001Z · LW · GW

The Pimsleur series of language courses are just audio, and they use spaced repetition (among other research-backed techniques) without a computer. They've got an app now, but the original tapes would work on a Walkman. You're supposed to do one lesson per day. They've scheduled the material to bring vocabulary words up when you're about to forget them.

Comment by gilch on The tech left behind · 2020-11-18T02:23:01.194Z · LW · GW

I think the EBR-II reactor was a notable example. The government cut funding three years before the completion of the program. Its design is what we now call an "integral fast reactor". Its passive safety features demonstrated that it literally cannot melt down. An IFR design would also produce much less waste than a conventional light-water reactor.

Comment by gilch on The tech left behind · 2020-11-18T02:11:24.996Z · LW · GW

The Smalltalk programming language and environment was revolutionary at the time and still highly influential to this day. Lots of later languages have copied some of its features, and none of them really got it right.

The grammar is extremely simple and easy to pick up compared to most industry languages. A famous small program demonstrating all of the language (but not the library) fits on a postcard.

Using the debugger, you can catch an exception, walk up the stack, and correct, recompile and swap in individual methods while the program is still running. You can save the entire state of the program at any time and resume it at a later time, even on another machine. You need an entire OS in a VM to do this in almost any other language.

The tight feedback loops you get from its interactive programming stye is arguably superior to almost anything else we have today, although e.g. Python or ClojureScript can approach this level of interactivity, it isn't their default.

Smalltalk's first stable release was in 1980 and we still haven't caught up to its level in industry. It's hard to understand exactly how this happened historically, but it seems to be path dependence based on some combination of (relatively) poor marketing, early mistakes in design, and the limitations of older hardware that could barely handle those features when the industry was first taking off.

But there are open-source Smalltalks now, most deriving from Squeak. Pharo, Dolphin, and Cuis are notable. There is even a VM written in JavaScript so you can try it in your web browser.

Comment by gilch on The tech left behind · 2020-11-18T01:37:20.815Z · LW · GW

Aerospike rockets are supposed to be much more fuel-efficient in atmosphere than are the conventional bell nozzles.

There are good reasons "rocket science" has become a synonym for "difficult". Nobody wants to take a chance on unproven technology when designing rockets is already hard enough. Not even Elon Musk, at least so far.

Comment by gilch on A review of Where Is My Flying Car? by J. Storrs Hall · 2020-11-07T22:38:44.322Z · LW · GW

The rise of the flying car industry

Comment by gilch on Where do (did?) stable, cooperative institutions come from? · 2020-11-05T02:14:30.666Z · LW · GW

According to Strauss–Howe generational theory, history is made of cycling Saecula each divided into four generational Turnings: the High, the Awakening, the Unraveling, and the Crisis, in that order, which tend to last approximately 20 years while the next social generation of people advances in age and takes over their role in society. Each Turning is provoked by the flaws of the last.

We are now approaching the end of the current Crisis Turning (since about 2005). The Crisis is the Turning when institutions are at their nadir, due to their long Unraveling in the previous Turning (1982–2004), when individualism was at its peak. Institutions will be destroyed and rebuilt for the next Turning. The current Millennial Saeculum will end in approximately 2025 and usher in the next Saeculum. The coming High will be when collective institutions will be strongest and individualism will be at its weakest.

The fourness of the Turnings also pattern match onto other things. Strauss and Howe associate each social generation with the Prophet, Nomad, Hero, and Artist archetypes, in that order. (The Millenials are Heroes, Generation X are Nomads, etc.) I also wonder if this fourness also maps onto Simulacra Levels, after reading Zvi's The Four Children of the Seder as the Simulacra Levels, which also has a generational feel.

Given that framework, strong institutions develop in response to a crisis, when culture tilts toward collectivism. When that collective culture unravels into individualism, the institutions decay.

[Epistemic status: stab-in-the-dark pattern matching.]

Comment by gilch on Daniel Kokotajlo's Shortform · 2020-11-04T23:05:44.154Z · LW · GW

Reminds me of cookie cutters from The Diamond Age

A cookie-cutter was shaped like an aspirin tablet, except that the top and bottom were domed more to withstand ambient pressure; for like most other nanotechnological devices a cookie-cutter was filled with vacuum. Inside were two centrifuges, rotating on the same axis but in opposite directions, preventing the unit from acting like a gyroscope. The device could be triggered in various ways; the most primitive were simple seven-minute time bombs.

Detonation dissolved the bonds holding the centrifuges together so that each of a thousand or so ballisticules suddenly flew outward. The enclosing shell shattered easily, and each ballisticule kicked up a shock wave, doing surprisingly little damage at first, tracing narrow linear disturbances and occasionally taking a chip out of a bone. But soon they slowed to near the speed of sound, where shock wave piled on top of shock wave to produce a sonic boom. Then all the damage happened at once. Depending on the initial speed of the centrifuge, this could happen at varying distances from the detonation point; most everything inside the radius was undamaged but everything near it was pulped; hence "cookie-cutter." The victim then made a loud noise like the crack of a whip, as a few fragments exited his or her flesh and dropped through the sound barrier in air. Startled witnesses would turn just in time to see the victim flushing bright pink. Bloodred crescents would suddenly appear all over the body; these marked the geometric intersection of detonation surfaces with skin and were a boon to forensic types, who cloud thereby identify the type of cookie-cutter by comparing the marks against a handy pocket reference card. The victim was just a big leaky sack of undifferentiated gore at this point and, of course, never survived.

Comment by gilch on Non Polemic: How do you personally deal with "irrational" people? · 2020-11-04T05:03:31.176Z · LW · GW

Let us know how it goes.

Comment by gilch on Non Polemic: How do you personally deal with "irrational" people? · 2020-11-03T18:23:09.473Z · LW · GW

The reasonable man adapts himself to the world: the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man.

—George Bernard Shaw

I'm all for having an accurate map, and that does mean updating that map. But don't let that stop you from trying to alter the territory—and actually fixing problems.

If the world fails to meet your expectations, sometimes the problem is with the world.

Comment by gilch on Non Polemic: How do you personally deal with "irrational" people? · 2020-11-02T19:04:43.741Z · LW · GW

Have you seen Street Epistemology yet? It's an effective way of leading irrational people to notice their own contradictions, but it does take some patience.

Comment by gilch on Message Length · 2020-10-23T04:04:22.922Z · LW · GW

Toolz has a sliding window.

Hy's partition can also do a sliding window like Clojure's can, by setting the step argument.

Comment by gilch on A tale from Communist China · 2020-10-22T21:40:11.462Z · LW · GW

Suppose you think there's a stock bubble; the temptation is to buy, hold as it rises, and then sell at the top of the bubble before everything comes crashing down. But enough people are trying to do that that you need some special skill in order to be in the leading edge, able to sell when there's a buyer. It's much less risky to just sell the stocks as soon as you think there's a bubble, which foregoes any additional gains but means you avoid the crash entirely (by taking it on voluntarily, sort of).

The correct move in this situation isn't to get out early, but to leverage down. By keeping a balanced fraction in cash, you both accumulate the gains as the bubble inflates, and survive the inevitable crash—and it's likely you can come out ahead on net. The appropriate ratio depends on the amount of volatility you're expecting.

I'm not sure how to generalize this insight, but maybe there's a similar move available? A winter home in Costa Rica, perhaps? Even if that means a smaller summer home here.

Comment by gilch on What reacts would you like to be able to give on posts? (emoticons, cognicons, and more) · 2020-10-05T05:12:11.915Z · LW · GW

Are you imagining these reacts for top-level posts only like tags, or also for comments and for replies to comments like karma? I feel like the latter would be more useful. Shortform, for example, kind of makes comments act like small top-level posts. Lowering the bar for participation seems more important the deeper a reply is nested.

Comment by gilch on What reacts would you like to be able to give on posts? (emoticons, cognicons, and more) · 2020-10-05T05:08:01.604Z · LW · GW

But what's the benefit of having a small set of 'standard reacts' instead of allowing/requiring that users express their thoughts in feelings in text?

Would you rather see a dozen replies to a post that simply say "Updated."? Neither does anyone else, so they don't, which means we're deprived of the useful feedback. Maybe one person replies "Updated." and they get the karma for being first instead of (or in addition to) the original. That doesn't seem fair. Or maybe they get downvoted for a reply that was too short, even if a lot of people agree. With an "updated" react, this just works.

This site in particular might be better because we don't provide low-effort options for providing (effectively very-low-information) feedback.

I kind of think this is a good point, but there are tradeoffs. We want to lower the bar to providing feedback so we can get more feedback, but not so much that we disincentivize the discourse. I feel this tradeoff is probably worth it, but that's a guess. If someone has something they're willing to say, how often is a react going to prevent it? They'd probably do both, like how often people will downvote and also explain why.

Comment by gilch on What reacts would you like to be able to give on posts? (emoticons, cognicons, and more) · 2020-10-05T04:31:47.577Z · LW · GW

Social media tends to amplify the controversial, while hiding the consensus. Controversy engages more participation, and thus generates more noise, making it more noticeable. Maybe the controversial bits really are more important to talk about, but because of the availability heuristic, this gives everyone a distorted view.

I'm not entirely sure what to do about it, but more fluid feedback might be useful. Karma helps a lot, by giving consensus a way to get noticed, but it's not perfect.

Something like Arbital's probability distributions that could be aggregated from multiple users might be more informative than a simple "agree/disagree" reaction. You could then say how confidently you agree/disagree. But a post may have multiple claims, so we'd need some way to separate them out. Maybe you could excerpt a quote and use that as a "reply", but one that others can add to with their own estimates. With a good interface, this would be only slightly more effort than clicking agree/disagree, much less than a full reply.

Some way to express what degree of surprise (maybe as deciban odds compared to your priors) might be more informative than an "updated" reaction. These could probably also be aggregated.

Comment by gilch on What reacts would you like to be able to give on posts? (emoticons, cognicons, and more) · 2020-10-05T04:12:16.322Z · LW · GW

Karma currently conflates multiple possible reations into a single datum. An upvote currently could mean "me too" or "I updated" or "I agree" or "others should read this", or "LOL", etc. A downvote could mean "fallacy" or "poor quality" or "disagree", etc. It's hard for posters and readers to discern intent.

Facebook and Github (and probably others) now have a small number of Emoji reactions, instead of just +/- or "Like". This is a more fine-grained feedback mechanism than a simple vote, but still easy (and familiar) enough for everyone to understand and use.

These give us additional low-effort feedback mechanisms that normally wouldn't be worth a written reply. Facebook's selection isn't appropriate for Less Wrong, but we could choose some better ones. They also need not be icons. Words or short phrases (i.e. "tags") will do. We'd probably want to pull memes about good discourse and their failure modes from the Sequences, e.g. #applause-light or #updated, etc., as well as all the well-known logical fallacies and biases. (These could also be links to the wiki for the uninitiated.)

For maximum flexibility, we could also give users the option to reply with a custom tag. Popular options could be added to the default suggestions, either automatically, or curated by the Sunshine Regiment. My concern with this approach is that custom tags may not have clear meanings, but a curated set could be linked to agreed-upon definitions in the wiki.

Tags also give us a good mechanism for straw polls. I've seen karma used this way on the old Less wrong, with a yes/no question asking for karma votes, (and a reply by the same author to counter balance it with opposite votes). The new weighted karma kind of destroys this feature. But with tags, we could vote with #yes #no or even #A #B #C #D, etc. for multiple-choice questions.


This was from before we had tags for posts. Calling these reactions "tags" might be confusing now. I'm also not sure how I feel about using logical fallacies per se now.

Comment by gilch on What reacts would you like to be able to give on posts? (emoticons, cognicons, and more) · 2020-10-05T04:04:52.513Z · LW · GW

While I like the basic set, I have mixed feelings about the fallacies. I don't know which 25 you're using, but many of the so-called "logical fallacies", while fallacious for deductive syllogisms, are nonetheless correct Bayesian inferences when working inductively in the real world.

I don't see many syllogisms in LessWrong comments. We've moved past that. Being able to slap down good arguments for technicalities that don't even apply in that context seems like a problem.

For example, on priors, you are better off trusting experts in their field than laymen. But this is called the "argument from authority fallacy". The correct counter is Argument Screens Off Authority, which needs a more specific situation to work.

Another example, if I'm debating the existence of God with a theist, and point out that the Bible was written by ignorant goat herders who were deeply confused about reality, he might counter that I've committed "the genetic fallacy" and the provenance of his evidence has no logical bearing on the merits of his claim. And yet, for evidence to count, it must be entangled, by links of cause and effect to what it is evidence of.

But I also like the idea of quick feedback so we don't need to re-hash points of broad agreement, both for newcomers and to hone our reasoning skills.

I'm not sure what to do about this. There may be certain LessWrong memes or concepts we could catalogue like this to serve a similar function—to point out obvious missteps.

Comment by gilch on What reacts would you like to be able to give on posts? (emoticons, cognicons, and more) · 2020-10-05T03:42:56.108Z · LW · GW

While I'm opposed to adding all the Slack emojis (Because there are too many and they're unclear, and yet you can't always find the one you need.) I like this basic set. The "delta" seems especially important. But they overlap with karma, especially the +1/-1 bit. We should have one or the other, and we already have karma. I think karma is also sometimes used to mean the +/- (agree/disagree), but I also like the idea of separating these concerns.

Comment by gilch on Honoring Petrov Day on LessWrong, in 2020 · 2020-09-30T20:00:23.184Z · LW · GW

In Petrov's case in particular, the new satellite-based early warning system was unproven so he didn't completely trust it, and he didn't believe a US first strike would use only one missile, or later, only four more, instead of hundreds. Furthermore, ground radar didn't confirm. And, of course, attacking on a false alarm would be suicidal because he believed the Enemy would push the button, so striking first "just in case", failed his cost-benefit analysis.

It was not "just" a commitment to pacifism.

Comment by gilch on Honoring Petrov Day on LessWrong, in 2020 · 2020-09-27T18:09:34.368Z · LW · GW

Game theory is very much applicable to the real world. Imperfect information is just a different game. You are correct that assuming perfect information is a simplification. But assuming imperfect information, what does that change?

You want to lie to the Enemy, convince them that you will always push the button if they cross the line, then never actually do it, and the Enemy knows this!

Sometimes all available options are risky. Betting your life on a coin flip is not generally a good idea, but if the only alternative is a lottery ticket, the coin flip looks pretty good. If the Enemy knows there's a significant chance that you won't press the button, in a sufficiently desperate situation, the Enemy might bet on that and strike first. But if the Enemy knows self-destruction is assured, then striking first looks like a bad option.

What possible reason could Petrov or those in similar situations have had for not pushing the button? Maybe he believed that the US would retaliate and kill his family at home, and that deterred him. In other words, he believed his enemy would push the button.

Applied to the real world, game theory is not just about how to play the games. It's also about the effects of changing the rules.

Comment by gilch on Honoring Petrov Day on LessWrong, in 2020 · 2020-09-26T20:18:02.015Z · LW · GW

I think this is the wrong lesson. If the Enemy knows you have precommitted to never press the button, then they are not deterred from striking first. MAD is game theory. In order to not blow up the world, you have to be willing to blow up the world. It's a Newcomblike problem: It feels like there are two decisions to be made, but there is only one, in advance.

Comment by gilch on Charting Is Mostly Superstition · 2020-09-24T04:13:52.060Z · LW · GW

Yes, I'm trading several of those right now, including a couple that I've hinted at in the sequence so far.

Comment by gilch on The new Editor · 2020-09-23T23:13:20.818Z · LW · GW

LessWrong will host our images now? How do we add images using markdown? It didn't work when I tried drag and drop or pasting.

Comment by gilch on Charting Is Mostly Superstition · 2020-09-23T17:02:31.074Z · LW · GW

I only demonstrated with a spreadsheet because I expect that's something most of the audience would understand. But spreadsheets have trouble handling the amount of data we need to work with. As I said, I would probably be using Python myself. Anaconda comes with enough to do this stuff. Pandas/Numpy, matplotlib, scipy, statsmodels. I kind of prefer HoloViews for charting. I also use the AlphaVantage API to get data sometimes.

I have looked at Quantopian a bit, but haven't really gotten into it yet. From the documentation, it does look like it can do the stuff I'm talking about.

Comment by gilch on Market Misconceptions · 2020-09-23T16:51:08.403Z · LW · GW

I am using the term loosely. Alpha and Beta come from the CAPM, where they have a very specific meaning. Perhaps I should have said "risk premium" instead of "beta". That's closer to what I mean. And by "alpha" I mean what's left after subtracting out the risk premium and luck: the skill of the trader.

Comment by gilch on Market Misconceptions · 2020-09-23T16:44:43.162Z · LW · GW

Turn the question around: assume your goal is to lose money. If a market was 100% efficient, the price moves would be 100% unpredictable, i.e. random. You're just as likely to make money as to lose it trading such a market. How could you lose money if you tried? Think about it.

I can only think of a few ways.

The first, obvious, way is to trade a lot. You lose commissions and spread each time you trade. Even if some of your trades happen to make a lot of money, you can lose more by overtrading. How quickly this happens depends on the size of the spread and commissions, and the frequency of your trades. You lose more slowly this way if you trade less often. Hence, DON'T PANIC.

The second way is leverage. If 100% of your portfolio value is within the expected move of your trade, then you stand to make a lot of money on the upside, but you lose everything on the downside. Even if you happen to start in a winning streak, you are almost certain to eventually lose it all. But 100% is not required. As long as you're over the Kelly fraction, you'll lose the entire bankroll eventually. Hence, Don't Bet the Farm.

The third way is more subtle. Even assuming an efficient market, the buy-and-hold strategy can work over time due to the risk premium. The random walk has a negative skew with a positive drift. You'll hit a few home runs when the market crashes, but, over the long term, a short-and-hold strategy will lose money. Hence, Don't Fight the Wave.

I can think of numerous individual strategies for losing money, but they all seem to fall into one of these three categories, assuming an efficient market.

There is a fourth way to lose money trading, but it breaks one of our assumptions: trade in an inefficient market. Alpha is difficult to find. You're probably no more likely to stumble upon a negative alpha trade than a positive one. Find some alpha, and then make the opposite bet.

Comment by gilch on Donald Hobson's Shortform · 2020-09-10T06:13:27.351Z · LW · GW

In what sense is a disconnected number line "after" the one with the zero on it?

Comment by gilch on Luna First, But Not To Live There · 2020-09-10T01:59:56.440Z · LW · GW

Very much disagree on space colonies as hedge against human extinction. I could write a more detailed critique, but the bottom line is there is no x-risk severe enough to wipe out all (not merely 99.999%) humans on Earth but at the same time not severe enough to also wipe out all moon/Mars colonies.

I think this is a good point. Civilization may eventually recover from some catastrophic risks (e.g. nuclear war). And some risks are so severe that even Mars would not be safe (e.g. UFAI).

But are there no risks that could wipe out humanity on Earth that wouldn't also kill a Mars colony? A comet impacting the Earth might be at the right scale for that. Or maybe a runaway greenhouse effect triggered by our carbon emissions.

And what do you think about using space colonies as a hedge against the collapse of civilization (rather than extinction)?

Comment by gilch on The Box Spread Trick: Get rich slightly faster · 2020-09-06T02:46:35.062Z · LW · GW

I think the specific choice of SPX (the S&P 500) here doesn't matter too much, but presumably it'll be good because it's something that gets traded on a lot.

The liquidity isn't the only reason. SPX has European-style options. If you tried this on SPY instead, you'd be exposed to early-assignment risk. SPX is not the only index with European options, but it is the most liquid.

I'm unlikely to make use of this myself, not living in the US and not having that kind of money to invest.

You can adjust the width of the boxes to borrow a smaller amount. And the XSP mini options are one-tenth the size of the SPX options. They're not as liquid though. Some brokers take international customers.

Comment by gilch on How to Lose a Fair Game · 2020-08-27T21:56:17.633Z · LW · GW

Sure, the macroeconomic regime was different then; but in a world of negative interest rates, are you sure it won't change again?

Nope. We are not trying to avoid all risk. We're trying to get exposed to risk so we can get paid for it. The right side is uncomfortable. Taking on the risk of bonds crashing, in the appropriate amount, so we can get paid for it, is exactly the point of adding leveraged bonds to a risk premium portfolio. If it weren't risky, there wouldn't be a risk premium for holding it. People on the margins have been forecasting the end of the bond bull market for years. If you had listened to them then, you'd have given up the returns up till now.

If you play the game long enough, risks will eventually bite you. You will have drawdowns. Don't Bet the Farm. But the market pays you extra for it. You'll still eventually come out ahead if you size your exposure appropriately.

Comment by gilch on What posts on finance would your find helpful or interesting? · 2020-08-27T01:23:34.957Z · LW · GW

Some brokers are available internationally. I have heard of people using Interactive Brokers outside the US. CFDs are illegal in the US, but an important investment class elsewhere.

I've been using Alpha Vantage and Yahoo finance for historical data. I have heard Stooq is good for UK market data.

I've started using Oanda and Zorro on an EC2 VM to autotrade Forex. Zorro is also supposed to work with Interactive Brokers (among others). You can try Zorro for free if your account size is small enough.

I have heard there are other options. Lots of brokers have an API. It's probably not worth your time to write your own backtester/robotrader when you can use one off-the-shelf. You still have to code your strategy though.

Comment by gilch on What posts on finance would your find helpful or interesting? · 2020-08-27T01:02:36.203Z · LW · GW

[I'm not your financial advisor. I don't know your financial situation, and this is not financial advice. Leverage is a double-edged sword. Don't Bet the Farm.]

Depends on what you need it for. 3x ETFs are probably plenty for a basic risk premium portfolio. LEAPS might have more variety, but you have to buy 100 share increments (which makes it harder to balance in a small account), and you have to deal with vega risk and occasional rolling hassle and costs.

Portfolio margin is probably the best way overall. But I haven't seen it available for less than a $100,000 account. You can get almost 2x on an ordinary Reg-T margin account though. Brokers may charge outrageous amounts of interest for margin loans. You can sort of work around this by financing with box spreads.

Outside the US, you can probably get 5x on CFDs. I can't trade them, so I haven't looked into them too much.

Leverage up to 20x is easy to get for Forex (and maybe 50x for the major pairs), because Forex volatility is so low, you need a lot of leverage to bother with it. But the market is mature and efficient enough that I don't recommend trading it without some automation. Grinding out an edge that small is really tedious if you have to do it manually.

Comment by gilch on What posts on finance would your find helpful or interesting? · 2020-08-27T00:32:40.925Z · LW · GW

Leverage is the ratio of the amount of the investment you control to the amount of money you need to control it. Buying the investment outright is 1x leverage.

Wise use of leverage is probably required for a good return. I talked about this a bit in How to Lose a Fair Game.

Leveraging up means you are increasing the ratio, which usually means you are borrowing money to buy investments.

The common way to do it is a margin account. You buy stock, and use the stock as collateral for a loan from your broker, which you use to buy more stock (not necessarily the same one). You can profit twice as much from a favorable move, but you also lose twice as much from an unfavorable one. That's 2x leverage, you can use less. The broker also charges interest. The loan is fully collateralized, so if the shares you're putting up lose value, you have to pay back the excess loan. This situation is called a "margin call". You can also get leverage using options or buying shares in a fund that is itself leveraged in some way.

Leveraging down means the opposite, usually you keep some cash in reserve, and keep the ratio of cash to investment value balanced. For example, a portfolio of 50% cash and 50% stock has 0.5x leverage, assuming you keep it balanced as the value of the stock changes.

Comment by gilch on What posts on finance would your find helpful or interesting? · 2020-08-26T23:25:31.582Z · LW · GW

Short selling means that you borrow shares of stock to sell to someone else. You probably can't do this in a basic account or a retirement account (with some minor exceptions). Your broker will find someone on your behalf. Some stocks are hard to borrow. Your broker might not be able to find enough of them.

You're still on the hook for any dividends and have to compensate the original owner for any dividends they miss. Again, the broker should handle this automatically. If the market price falls, you can then cover your short position for less than you sold it for, which is a profit. The opposite case would be a loss. If you do have a margin account, your broker may also loan your shares to someone else to sell. Some brokers compensate you for this. If your shares are loaned out you still get the dividends, but they don't count as dividends for tax purposes.

Comment by gilch on Charting Is Mostly Superstition · 2020-08-25T19:04:23.091Z · LW · GW

Maybe the argument is something like financial markets have so much noise that you're more likely to accidentally overfit to noise rather than find real patterns that let you infer a useful model

That is pretty much my argument, yes.

Emotional trading is such a danger that I also want my trading to be in principle something that I could program into a computer, even if I do, in practice, execute the trades manually. This isn't compatible with "eyeballing it".

I do look at price charts. I use Heikin Ashi candles, Bollinger bands, probability cones, moving averages, volatility graphs, and I even eyeball support and resistance levels. But I mostly don't expect to predict the future with these. It's more about noticing when my initial assumptions have been violated, by past behavior.

but if that's the case that's a problem everywhere, and you just have to get more aggressive about dealing with it

I thought that's what I was doing. What else would you suggest?

up to some limit where there's simply not enough signal to determine anything useful.

I think this is the case in some markets. You can still sometimes get enough signal for alphas that affect multiple assets in similar ways, by trading ensembles.

Comment by gilch on The Wrong Side of Risk · 2020-08-24T20:02:34.904Z · LW · GW

Right. A call is an option contract. It's a different instrument than the underlying shares of stock. You can be short a call option and not short shares. You're still executing a bearish play on the stock, but you're not shorting the stock.

If you get assigned on your call (unless it's cash-settled), you will have to produce the shares to give to the contract holder. If you didn't already have the shares, you will end up with a short position in the stock, which you will eventually have to cover by buying stock.

You can construct a synthetic short stock position using options, by selling a call, but you also have to buy a put at the same strike and expiry. This will behave very similarly to (typically) 100 short shares.

Comment by gilch on What posts on finance would your find helpful or interesting? · 2020-08-24T19:52:20.128Z · LW · GW

There are lots of apps for that. Search for "demo account" or something. The ones with a real stockbroker are going to be the most realistic, and will probably use real market data delayed by 15 minutes. My favorite trading application so far is the desktop version of thinkorswim (TD Ameritrade owns it at the moment, but Schwab might be buying them.) It's scriptable and has some nice analysis tools.

Comment by gilch on How to Lose a Fair Game · 2020-08-24T19:32:51.441Z · LW · GW

Leveraging up, on its own, is never going to improve your Sharpe ratio. But you can use leverage to get a better return if your portfolio is below your risk tolerance, at least until you hit the optimal bet size.

Volatility does have bad effects on leverage, which is why my somewhat safer system reduces the leverage for an asset when its vol is high, even below 1x, if necessary, by holding cash. The Kelly strategy means there is always an optimal amount of leverage, and it's unlikely to be exactly 1x.

Comment by gilch on Charting Is Mostly Superstition · 2020-08-24T17:20:04.487Z · LW · GW

If the EMH is true,

First, I don't believe the EMH is true. It's an approximation at best.

I would expect every effect that's in the literature soon to be removed from the market by market participants.

I only object to "soon". Alphas do decay and you have to keep finding new ones, but it can take years before it stops working.

Why do you believe that won't happen?

Alphas don't persist simply because of secrecy. That's a misconception. There are cases published in the academic literature that still show up in recent price data. Longer-lived alphas often persist because exploiting them is somehow distasteful (e.g. a skewed risk profile, capital constraints, awkward setup). But even for short-lived alphas, new cases of known inefficiencies can reoccur for known reasons, and if you get there first, you can be the one to exploit them. (E.g. The pump-and-dump fade.) Not all markets are liquid enough to be efficient, even approximately. Sometimes there are more whales than sharks.

Comment by gilch on The Wrong Side of Risk · 2020-08-24T17:04:15.904Z · LW · GW

No, it really doesn't. You can make money from an out-of-the-money short call even if the stock goes up, as long as it stays below your strike price. You can even make money if it exceeds your strike price and then falls back below before expiration (assuming you don't get assigned early, which can happen if there's a dividend that exceeds the time value of the option). And even if it expires a little above your strike, if you took in enough premium, you could still come out ahead on net after subtracting your loss from covering your assigned short position. You don't have to be right about the market direction to make money. It's enough to not be terribly wrong.

My main point here was that you collect the risk premium for selling the call, but you pay the premium for shorting the stock.

Comment by gilch on How to Lose a Fair Game · 2020-08-24T16:39:54.331Z · LW · GW

Is there an underlying frequent rebalancing or other effect that I am missing?

Rebalancing is required. My example rebalanced after a 10% deviation, but you'd get very similar performance by rebalancing monthly.

My understanding was that leveraged ETFs are only useful for day trading and not for long-term buy-and-hold

That does seem to be a common misconception, but it's easily shown to be false. A (e.g.) 3x daily ETF tracks at 3x per day. Over longer periods, due to compounding, this accelerates your returns (even your negative returns), so don't expect performance to be 3x over periods longer than a day. Results can be even higher than 3x for instruments that tend to trend up in the long term, which stocks and bonds do.

Comment by gilch on Tips/tricks/notes on optimizing investments · 2020-08-24T07:40:48.036Z · LW · GW

Well, it's obviously not going to be the midpoint when it fills because you can only buy at someone's ask or sell at someone's bid. But with a limit order, you can be the best bid or ask.

I don't usually chase it. If you're buying a call and the market drops, you get a fill. If it rallies, maybe you wait 15 minutes and adjust, or try again tomorrow. For LEAPS it wouldn't be unreasonable to try for a few days.

The market is usually calmer in the middle of the trading day, maybe because the big players are eating lunch, although it can get chaotic again near the close.

Except for a very liquid underlying near the money, you're almost always trading options with a market maker. The market maker will set the bid and ask based on his models. If he gets a fill, and can't get enough of the opposite side, he'll buy or sell shares of the underlying to neutralize his delta. He's not making directional bets, just making money on the spreads. If you offer a trade near the midpoint, but even slightly in the market-maker's favor, he'll usually trade with you, when he gets around to it. This could easily take fifteen minutes. He also doesn't like you narrowing the spread on him, because that means someone might trade with you directly and he doesn't get his cut, so if he can handle your volume, he'll just take your order off the book.

Comment by gilch on Charting Is Mostly Superstition · 2020-08-24T01:17:20.108Z · LW · GW

I disagree. I do, of course, run a backtest before trading, and ultimately with simulated fees and spread, yes. But when looking for new alphas, I don't do that first.

I don't just eye the histograms. I compute summary statistics. I also plot two at once to see where they differ, and maybe plot a bell curve on top of the histogram to see where it differs.

It's so easy to get in a cycle of optimizing backtests but this will almost always overfit. I want to be sure the effect exists first, in isolation, before I try to trade it.

We're trying to reverse-engineer the behavior of a "random variable" (hopefully plus some exploitable non-randomness) from its outputs. Maybe this is easier with an immature, less-efficient market like crypto (in which case, I want to figure out how to trade it too), but there is so much noise in what I'm trading, that the exploitable effects are very hard to see in a backtest. The signal-to-noise ratio is too low.

I think the best way to illustrate this would be with a simulation. Model the price as a random walk with an appropriate distribution (normal-ish, depending on how realistic you want to be), plus some signal (in the simple case of collecting the risk premium, just a constant drift) and backtest that. You'll find that with realistic parameters, the returns from a backtest are highly variable, depending on your random seed. I just can't trust it.

I don't know if you've read HPMOR yet, but remember the scene with the 2-4-6 test? Read it if you haven't yet, I don't want to spoil it for anyone.

It's a good illustration of what's required of scientific thinking. How do you go about building an accurate model of something? What cognitive biases get in the way?

Suppose I ran one of those simulations, and gave you the output as price data. Remember, this isn't a stock, and I know exactly what the underlying distribution is, because I made it on a computer. Maybe I added some exploitable non-randomness. How confident would you be in knowing that distribution from running a backtest? How would you find the exploit? How confident could you be in your bet size? If I then took your strategy and ran it using fresh data from the simulator (with the same distribution), how confident could you be that your strategy would perform well?

You don't have to be 100% accurate to make money, but you have to be accurate enough, and more accurate is better. When the noise is very loud, it takes a lot of data to infer a distribution like that with much accuracy. Often more than is available from a single instrument.

Comment by gilch on What posts on finance would your find helpful or interesting? · 2020-08-23T21:40:33.126Z · LW · GW

I don't understand Titan's strategy well enough to know if it would work, but a Sharpe of .77 is totally achievable though various means. It's really not that hard to beat the index in terms of Sharpe ratio. Diversification really is a free lunch. The S&P 500 is American large caps. You can diversify a lot more than that. That doesn't seem to be what Titan is doing though.

I don't buy the EMH, because alpha exists, but yes, that is what the EMH would imply: you can only outperform due to luck.

If you model an asset price as a random walk with drift, and then try to compute the Sharpe ratios, you will get different answers at different times due to luck. Same process. Could it vary between .77 and .51? Yes. Easily.

Comment by gilch on What posts on finance would your find helpful or interesting? · 2020-08-23T20:41:01.913Z · LW · GW

One thing I'd really recommend to get started is to open a demo (or "paper") account with a stockbroker. Get familiar with the mechanics of actually ordering trades with their software. Then you can experiment with what you're learning. Don't worry about blowing up your account, because it's not real money. You can just reset it.

Comment by gilch on Tips/tricks/notes on optimizing investments · 2020-08-22T19:18:30.468Z · LW · GW

I do think the basic idea of reducing exposure is valid. I don't assume a constant return distribution, so in principle, the optimal Kelly fraction for an asset can vary over time. Risk-parity portfolios like the one I described in How to Lose a Fair Game did seem to generate a little alpha due to the volatility adjustments when I backtested them.

Comment by gilch on Tips/tricks/notes on optimizing investments · 2020-08-22T19:08:46.841Z · LW · GW

Some of these answers are still applicable outside the US. Every country has different tax laws though. There are also differences in what kinds of investments are accessible. I've heard that some Europeans aren't allowed to trade American ETFs, for example. I'm not sure of the details though. On the other hand, they can trade CFDs, which are illegal in the US.

Comment by gilch on Tips/tricks/notes on optimizing investments · 2020-08-22T18:57:05.021Z · LW · GW

I guess technically it's actually "expose you to gamma risk"

Yeah, that sounds right. But gammas can turn into delta as the market moves. If you do box with American options and get assigned early, the shares (or short shares) will hedge you for a while because they'll have a similar contribution to your overall portfolio delta as the option they replaced, but it's not going to have the same behavior as an option when the price moves. So you'd want to close and reposition before that happens, which, of course, requires capital and commissions.

Hopefully they'll be smart enough to sell your index ETFs because that's much more liquid?

You would think. Sometimes you get liquidated by an algorithm though. I've heard that Interactive Broker's liquidation algorithms are especially aggressive, which is part of how they can offer such competitive margin loan rates. (They also have a "liquidate last" feature that lets you protect some positions from the algorithm for longer. Definitely use that for the boxes.)

The above is purely theoretically though.

Yes. I have no first-hand experience with this. I have heard things on forums from people, but I can't call that a reliable source.

Comment by gilch on Tips/tricks/notes on optimizing investments · 2020-08-22T18:44:27.544Z · LW · GW

Tax Lien certificates. Basically, you're giving an extension to someone who is delinquent on their property taxes, and ensuring that the local government, who probably very much needs predictable funds, collects them in a timely manner.

Some of these are cheap, in the hundred dollar range, which makes it easier to get started even if you don't have a lot of money to invest. Terms and availability depend on the area you buy them from. Interest rates can be very high, around 20% in some areas. In some cases (likely foreclosures), you can have a good chance of becoming the owner (or part owner) of the property, which can be massively profitable (but also a hassle).

On the other hand, some property is not that valuable, so you need to do some research. The lack of secondary markets for these makes them rather hard to sell early. And if you don't live in an area that offers good terms, you may have to travel to find the good deals, which is an expense. Some counties do offer auctions online, but you'd still need to do some research on the property.

Comment by gilch on Tips/tricks/notes on optimizing investments · 2020-08-22T18:36:43.790Z · LW · GW

Peer-to-peer loans (e.g. LendingClub). I wouldn't suggest starting with this unless you're already investing in the usual instruments. These are a more exotic investment for extra diversification. These have a high risk of default (don't bet the farm), but also high interest. You get to be the credit card company.