Hacker News Reader

Individual stocks are surprisingly random across three month stretches even in bull markets. 75% is significantly outperforming in terms of random individual stock picks over the last 10 years.

The real issue is people wouldn’t bring it up if the number wasn’t “shocking”. Which self selects for outliers.

Right, stocks [which are just proxies for companies] are completely random because of <random bs paper from E. Fama from 51 years ago>. Because, of course, how would it ever be possible to predict how a company might perform? That must surely be impossible, because TradFi economics has decided it must be. And then people just continually parrot the same stuff they've heard and it becomes more and more accepted as fact.

I just hate hearing these same narratives being repeated because... they've already been repeated. Why must stock returns be random? It's predicting whether a company is going to do well or not, which has nothing to do with randomness. And yet people (like this commenter) just keep repeating this stuff, absolutely endlessly. They don't even know why they're doing it (try asking him) -- they just keep doing it.

> It's predicting whether a company is going to do well or not, which has nothing to do with randomness.

You're right that that has little to do with randomness. But it also has little to do with investing.

You're misinterpreting what random paper says.

Random paper (Fama et al.) doesn't say it's impossible to predict how well a company performs. It says why do you think you can predict better than the 1000 Ph.D. employed by trading co. inc. when all of you have access to the same public information? So the price of the company's stock is going to converge to some number that already reflects the known information.

Not sure where you're getting the stock returns are random bit either.

The “random” bit seems to be coming from Retric’s comment to which they are replying.

Despite the downvotes, I am in generally in support of GP.

If you read hn a lot and built a lot over the past decade it hasn’t been hard to predict better than “a 1000 phds” which companies would win.

NVIDIA, apple, even small plays like twilio. It was obvious these companies had big leads in stuff that made money.

Go back and read the Cloudflare IPO filing thread here. That opened at what, $16?

You read the R2 thread from a few days ago it is again obvious AWS will bleed some of the most influential engineers when this becomes generally available. I don’t have a PhD mind you, but I would presume this will lead to entire companies moving to CF for some primitives, eventually CF pick up entire accounts.

Do PhDs at trading companies get annoyed because some AWS services still will only respond with plain/text in api responses containing json? That they have to write a custom parser as a result?

Or that the UI for AWS SNS in setting up an HTTPS subscriber kinda sucks? Like it was mostly done but worked so people got pulled elsewhere?

Do wall street PhDs realize the bundling of expensive AWS cloud primitives to use other AWS cloud products is anti-customer, and that maybe this is going to translate to devs voting with their feet when CF makes this move?

Heck, even the R2 announcement makes the egress fees issue clear in plain English. There are tons of comments validating that pain. Business people shouldn’t need a lot of convincing.

Do the above observations and experiences working with AWS and reading others experiences add up to more knowledge than 1000 PhDs?

I wouldn’t have thought so, but the value of an academic degree has been on a downward spiral for some time.

so i assume you’re a billionaire fund manager?

> You read the R2 thread from a few days ago it is again obvious AWS will bleed some of the most influential engineers when this becomes generally available.

Is it obvious? How likely is it that you're right - is the chance 100%? Is it even possible that you're wrong and this don't happen?

A company doing well isn’t indicative of the stock price doing the same. For example Cloudflare has one of the highest P/Es on the market, meaning its future growth has been accounted for by investors. So executing extremely well doesn’t cut it. You have to find new areas of growth that investors haven’t anticipated or grow your existing lines more quickly than investors anticipate.

hey, go trade, then! you can just long these equities yourself and make a lot of money if you're right. and there's a lot of very smart and very sophisticated firms that would love to be your counterparty, so you'll get great execution!

Everything is obvious in the stock market with survivorship bias but in reality there's just as many "sure fire wins" that don't succeed as there are ones that do.

Besides, successful trading is more about getting the timing right than it is about picking which companies will eventually increase in value.

>It says why do you think you can predict better than the 1000 Ph.D. employed by trading co. inc. when all of you have access to the same public information?

well one reason would be that there is something about the market at time X that means conventional wisdom is faulty, since the 1000 Ph.D. employed by trading co. inc will be following the conventional wisdom there is a chance to outwit - years leading up to 2007 housing crash spring to mind as an example where this can happen.

Housing looked obvious, but when exactly would it pop. As the saying goes “The market can stay irrational a lot longer than you can stay solvent!”

Sure, if you’re correct it’s easy to say the logic was sound, but you never learn the odds just what happened.

The value of a stock is based on the overall markets prediction of how that company will perform. If you can predict how the value of a stock will change, you are not predicting how the company will perform, but are predicting how the company will perform relative to the overall market's prediction.

That market has a lot of very smart people spending a lot of time making their predictions. If you can consistently identify stocks that are undervalued then either you are smarter than the rest of the market, or you know something the rest of the market doesn't.

It is possible to be smarter than the rest of the market. Particularly if you are investing in relatively low value companies and doing significant due diligence yourself. However, if you work in Congress, it is much more likely that you got your edge by knowing something that the rest of the market doesn't, and we have generally decided that it is illegal to make trades based on that information.

This isn't necessarily true. Not everyone has the same goals, risk tolerance, and is working with the same amount of capital in the market.

If you perceive some pattern in stock prices, you can bet on the continuation of the pattern. If there were commonly discernible patterns to prices, everyone with cursory knowledge of finance would become magically rich.

This can't happen because prices rise as a consequence of buying and fall as a consequence of selling. People who bet on a pattern destroy it.

The consequent randomness is easily verifiable. Try running some stock market data through a machine learning library, you won't get any interesting results.

Have you tried? You would be surprised at what can be achieved on smaller time frames.

But there is a thriving industry of spotting patterns which other people haven't yet thought of. And IIRC there is some evidence that even after these patterns are discovered in the academic literature, they persist for a while, just with lowered returns. The logic of what you're saying is true in principle, but in practice markets are not quite so perfect.

It wasn't my intention to give the impression that markets are totally efficient, just that the forces at play tilt the system toward increasingly random behavior. I was responding to the proposition that markets are considered random because "TradFi economics has decided it must be".

It’s not whether you predict the company trajectory (possible) but whether you predict it better than other people who are also trying to predict. It turns out the average market prediction has low bias (stock go up or down).

There’s very strong data on call purchases and other derivatives short of buying the specific stock, immediately before good news. You can run into this data via reading academic stuff efficient market hypothesis, as the front running damages the theory a bit and has led to shifts in how to explain EMH (semi-EMH, etc).

This data has been around for years

If you rarely trade then start buying short dated OOM calls just before news is a easy way to get a call from the SEC.

Not sure what you’re getting at. Data is there in Econ circles for a long time. Check it out

Right, whether or not 75% is suspicious depends entirely on the price action of the stock. If it has been trending up, there will be relatively fewer 3-month periods where the stock is down than if it is flat or trending down. SFBS is up from 13.25 to 79.61 and CNBKA is up from 32.95 to 115.22 since 2014.

If the price movement was completely random with no bias up or down, and you entered trades at completely random times, then you would expect 50% of trades to be profitable 3 months later. So 50% is our baseline. Then consider that these stocks have been going up. THEN consider that you don't have to enter trades at completely random times; if you buy during a dip, the price merely needs to return to the mean for you to be up.

A blind chimp could have a 90% sucess rate if the endpoint is holding for 3 months in the period 2014 to 2021. What isnide information are they suggesting that the CEO of a phone services company has?

> If you pick any random stock at any time and hold it for three months, what are the odds that it will be worth more?

High, very high if you filter down to the most likely stocks that would be traded, apple etc.

Statistically speaking I'd say those numbers more than warrant further investigation.

75% success rate for one trader is suspiciously high. Generally most active traders under-perform the market they're trading in, even a bull one.

> Generally most active traders under-perform the market they're trading in, even a bull one.

But if the market has been going up, even underperforming it means you can still make a profit (unless you significantly underperform). The quoted text in the gp is just talking about profit, not comparing to the market.

> 72 of them, or 75%, have shown a profit three months later.

That is surprisingly close to the number Matt Levine reported on some time ago. A group was trading almost exclusively on insider information, and (IIRC) 77% of their trades made a profit. It's a massive edge over other market participants, but only goes to show that even with that kind of information at your disposal, you are still not guaranteed to make money on every trade.

I wish I could find the article. The overlap between "Matt Levine article" and "insider trading" sets is large enough to make any search for a particular piece of news a needle/haystack problem.

EDIT: to clarify - when you are trading on inside information with the goal of making a quick buck, you are not interested in how the company will do in short term. You are trying to predict how the market in aggregate will react to that same information, once it becomes public.