[HN Gopher] Does generative AI facilitate investor trading? Evid...
___________________________________________________________________
Does generative AI facilitate investor trading? Evidence from
ChatGPT outages
Author : anonu
Score : 80 points
Date : 2024-07-14 00:37 UTC (3 days ago)
(HTM) web link (papers.ssrn.com)
(TXT) w3m dump (papers.ssrn.com)
| 2Gkashmiri wrote:
| How would this work in the market, I am Curious to try it
| mebutnotme wrote:
| It could be news parsing, where it can turn news article text
| into data points that an investment model could imply market
| signals from.
|
| Something similar could happen with transcripts of company
| investor calls, or any other text/audio companies put out. With
| the right prompt you can turn text into signals, the trick
| would be getting that prompt right (and prompt engineering to
| limit hallucinations as much as possible).
| DowagerDave wrote:
| all those data sources are trailing though. Most
| intentionally try to (over simplistically) explain; few to
| none predict.
| candiddevmike wrote:
| Sentiment analysis is AI now?
| ben_w wrote:
| Always was
| potatoman22 wrote:
| This is a classic example of "AI is whatever hasn't been
| done yet." https://en.wikipedia.org/wiki/AI_effect
| Legend2440 wrote:
| Sentiment analysis has always been done with ML models.
| It's one of those problems that can really only be solved
| with data.
| 0xdeadbeefbabe wrote:
| AI is bullish now
| sfink wrote:
| > the trick would be getting that prompt right (and prompt
| engineering to limit hallucinations as much as possible).
|
| Why do you have to get it right? Don't you just have to get
| it the same as everyone else, slightly faster?
| hluska wrote:
| That is the definition of 'right' in a trading context.
| kjkjadksj wrote:
| Or you can just not use ai and build your own sentiment
| analysis metric that is actually going to work. E.g. maybe
| you make a word cloud of a bunch of recent news articles, you
| identify certain candidate words as positive sentiment or
| negative sentiment. Now you can score any new article by
| counting pattern matches with either list. Not really much
| initial work to set up, and you probably get much higher
| quality results than what hallucinating ai would get you, if
| you happened to find that magic prompt that is.
| aweektogetit wrote:
| Could see it being used for data analysis when doing sourcing,
| plenty of AI-powered datascience notebooks out there
| joering2 wrote:
| I think lately results of chat went downhill. I got used to
| asking "are you sure about that" after each chat answer to my
| question, and to most of the time the answer is "you are right,
| here is how I was wrong".
| jazzyjackson wrote:
| chatgpt has never stood up for itself, if you express doubt it
| will defer and say it must by mistaken even when it's correct
| sitkack wrote:
| That is because they turned the sycophancy up to 11. You have
| to turn it down, if you are a serious LLM user and you don't
| have your own system prompt, you are using it wrong.
| babypuncher wrote:
| This doesn't change how frequently the LLM provides a wrong
| answer, just how confident it is when you question it.
| Zambyte wrote:
| For what it's worth, certain system prompts _can_ improve
| output quality.
| 12_throw_away wrote:
| Such as?
| Zambyte wrote:
| CoT prompting
|
| https://arxiv.org/abs/2407.01687
| jazzyjackson wrote:
| I havent found any serious LLMs to use :)
| kelseyfrog wrote:
| It makes it difficult to get less-biased responses. I find
| myself asking an affirmative-worded question and its
| affirmative-worded inverse to subtract out the sycophancy.
| It's a poor workaround.
| simonw wrote:
| That's not a recent thing - most of the LLMs I've tried have an
| annoying tendency to agree with you if you question them, even
| if they weren't actually wrong. The term for it is
| "sycophancy", I think there are a few papers that talk about
| it.
| amerkhalid wrote:
| This reminds me of some communications training where they
| were saying that the most things are not worth disagreeing on
| and just agree with whatever.
|
| And I think I read in some parenting books too that it is not
| worth correcting kids over things that don't matter like if
| they say dinosaurs and mammoth were alive during same time.
| Over-correcting hurts their confidence.
|
| Maybe LLM been trained on these ideas or maybe it is start of
| science fiction nightmare and AI really think that it is not
| worth arguing with mere humans.
| ec109685 wrote:
| Ask a question like "Why do EVs use less electricity per mile
| when it is 100 than 70 degrees outside" and then reverse the
| question.
|
| You'll get two answers that totally contradict each other.
|
| "Electric vehicles (EVs) use less electricity per mile in
| warmer temperatures (around 100 degrees) compared to cooler
| temperatures (around 70 degrees) primarily due to the following
| reasons:
|
| 1. *Battery Efficiency*: EV batteries operate more efficiently
| in warmer temperatures. Cold weather can reduce the battery's
| ability to hold a charge and increase internal resistance,
| leading to higher energy consumption.
|
| 2. *Thermal Management Systems*: In cooler temperatures, the
| EV's thermal management system works harder to keep the battery
| at an optimal operating temperature, consuming more energy.
| Conversely, in warmer weather, less energy is required to
| maintain this temperature.
|
| 3. *Reduced Heating Requirements*: When it's around 100
| degrees, the need for cabin heating is eliminated, whereas at
| 70 degrees, some heating might still be required for comfort,
| consuming additional energy.
|
| 4. *Air Conditioning Efficiency*: While air conditioning is
| used more in hotter weather, modern EVs are designed to be
| highly efficient in cooling compared to heating, leading to
| lower overall energy consumption even when the AC is in use.
|
| These factors combined result in lower electricity consumption
| per mile in warmer conditions compared to cooler ones."
|
| Reversed:
|
| " Electric vehicles (EVs) use more electricity per mile in
| extreme temperatures (both hot and cold) due to the increased
| energy demands of their climate control systems. Here's why:
|
| 1. *Air Conditioning in Hot Weather:* When it's 100 degrees
| outside, the EV's air conditioning system has to work harder to
| cool the cabin to a comfortable temperature. This requires
| additional energy from the battery.
|
| 2. *Battery Cooling:* High temperatures can affect the
| performance and longevity of the battery. To protect the
| battery, the EV's thermal management system will actively cool
| the battery, consuming more electricity.
|
| 3. *Increased Resistance:* High temperatures can slightly
| increase the electrical resistance in the battery and other
| components, which can lead to marginally higher energy
| consumption.
|
| In contrast, at a moderate temperature like 70 degrees, the
| demands on the air conditioning and thermal management systems
| are much lower, resulting in more efficient energy use."
| Workaccount2 wrote:
| This type of behavior seems like it should be something that
| should be (relatively) easy to fix with some kind of external
| intervention layer. Google has been able to answer queries
| without using LLMs for some time now, so maybe some kind of
| pre-check followed by a prompt injection to stiff-arm the LLM
| in the right direction.
| adw wrote:
| Whenever you're building some ad-hoc non-differentiable
| circumstantial hacks around your learned model (regardless
| of whether it's an LLM or not), two alarms should go off in
| your head;
|
| a) Could this be made differentiable? End-to-end systems
| are nearly always stronger than systems made up of multiple
| models or models-plus-heuristics;
|
| b) is it appropriate to be using a statistical model here
| at all?
|
| External intervention layers are generally severe technical
| debt.
| arrowsmith wrote:
| What does it mean to "make something differentiable"?
| Arn_Thor wrote:
| I don't understand how this is a surprise to so many people.
| The thing can't reason. If you point out an error, even if it's
| real, the GPT doesn't suddenly "see" the error. It constructs a
| reply of the likeliest series of words in a conversation where
| someone told it "you did X wrong". Its training material
| probably contains more forum posts admitting a screw-up than a
| "nuh-huh" response. And on top of that it is trained to be
| subservient.
|
| Because it doesn't _understand_ the mistake in the way a human
| would, it can't react reliably appropriately
| exitb wrote:
| Hypothetically speaking, if a chat provider used a Golden Gate
| Claude [1] technique to essentially create and pump a meme stock,
| is that a fraud?
|
| [1] https://www.anthropic.com/news/golden-gate-claude
| sixhobbits wrote:
| Love this idea. Would be great to see how it applies to twitter
| (e.g. reduction of tweets during GPT outages) but I guess that
| twitter data is hard to get now
| ProllyInfamous wrote:
| Background: I am personally a long-term value investor ( _a la_
| Sir Templeton _or_ Buffet +), but annually breakfast with a
| retired "numbers guy" (who's currently building his short-term
| trading strategy). For decades he worked finance for among the
| largest accounting firms in the world...
|
| Since 2016, our mutual friend and I have been adamently
| suggesting he look into "this bitcoin thing" -- his response was
| typically [disinterested].
|
| At a recent breakfast, it was explained how "talking with ChatGPT
| finally convinced [him] that _you guys have been correct about
| Bitcoin all along_. "
|
| When asked "what did ChatGPT explain to you that we two humans
| could not?!"
|
| His response was simply "I never got embarrassed asking the
| computer _any_ finance questions, even stupid /simple ones; I've
| always been _the finance guy with answers_ so I forgot how to be
| humble in unexplored topics. "
|
| Pretty damn humbling, indeed; now if only he had listened back in
| 2016 =D
|
| +: "Time _in_ the market beats _timING_ the market. "
| ipython wrote:
| I would be curious to hear what information it (ChatGPT)
| provided that changed his mind?
| DowagerDave wrote:
| it sounds like it's more the non-human / non-embarrassing
| setup than the actual results. The same way I swear at tools
| or yell at clouds.
| fwip wrote:
| Another datapoint that shows that LLMs will simply repeat what
| everyone is saying and happily give the exact wrong answer.
| snapcaster wrote:
| Can't have been trained on hackernews then, this is the most
| anti-crypto space I'm aware of besides /r/buttcoin. i'm also
| softly anti crypto so not criticizing, just pointing out
| you're literally in an environment where the opinion you're
| talking about gets eviscerated every time so it can't be what
| "everyone is saying"
| svara wrote:
| HN just loves novelty, once crypto made mainstream news HN
| fell out of love.
|
| The pattern has applied to Tesla/Musk, cryptocurrency,
| language models and probably much more I'm not aware of.
| passwordoops wrote:
| I agree about the love of novelty, but two of your
| examples (Tesla/Musk, LLMs) are also falling out of favor
| in mainstream news. Especially Elon since Twitter. LLMs
| too (or at least the hype surrounding them as AGI) have
| seen mainstream panning of late, in parallel to
| mainstream outlets
| rafaelmn wrote:
| I don't think it's about novelty - heck people here will
| praise Unix and C regularly.
|
| It's more that crypto initially had potential, at this
| point it's been tried and failed to deliver - at least
| from the value perspective of people here.
| staplers wrote:
| at this point it's been tried and failed to deliver
|
| A trillion dollars of value suggests otherwise.
| BoorishBears wrote:
| Imagine if dollar bills appreciated in value thousands of
| times over by hiding them under your pillow.
| rafaelmn wrote:
| I mean drug trade is also huge but you won't find many
| people praising it here (the second part of the sentence
| you quoted)
| DowagerDave wrote:
| fell out of love because it was no longer novel, or
| because the novel wore off and they realized it wasn't
| anything special?
| chant4747 wrote:
| > now if only he had listened back in 2016 =D
|
| So your investments could be used to screw other people out of
| their money who are similarly financially illiterate/insecure?
|
| It's kind of crazy to me that there are somehow still staunch
| cryptocurrency proponents out there who have somehow ignored
| all the negative press it has received.
| snapcaster wrote:
| Even if you're very conventionally minded, it shouldn't be
| that crazy to you that not everyone adopts dominant media
| narratives
| christianqchung wrote:
| By dominant media narratives you mean a self evident train
| of scams and security incidents? There's no conspiracy or
| undue effort to make crypto look bad, that happened in a
| decentralized manner.
| snapcaster wrote:
| You were the one that brought up the media (nobody said
| conspiracy?)
| christianqchung wrote:
| Media narratives and conspiracies are often used
| interchangeably, I can't read your mind so I included
| both. Also in your original comment you brought up the
| media, it's right there, so...
| snapcaster wrote:
| dude the comment i responded to is from you saying
| "cryptocurrency proponents out there who have somehow
| ignored all the negative press it has received."
| chant4747 wrote:
| In fairness, that was actually me. May want to double
| check the usernames. Both start with ch - easy mistake to
| make.
|
| I would have participated more in the discussion but HN
| is telling me I'm posting too fast so I'm guessing a
| moderator didn't like some of my harsh tone.
| snapcaster wrote:
| haha okay that explains it, yes i just didn't read and
| assumed it was same person
| codetrotter wrote:
| > a self evident train of scams and security incidents
|
| Because a bunch of bad people jumped on the bandwagon to
| exploit others.
|
| Bitcoin != "crypto"
|
| Bitcoin was the original and it's still here. And it will
| continue to be here.
|
| The goal of bitcoin is not to "make money". Bitcoin is
| money. A new kind of money. One that no central entity
| can control. One of which there will only ever be
| 21,000,000. No one can print more Bitcoin. Unlike fiat,
| which the government and the ruling class use to screw
| people over right in front of their faces all in broad
| daylight.
|
| Bitcoin can not be diluted. Bitcoin can not be inflated.
| Bitcoin can not be controlled. Bitcoin can not be
| stopped.
|
| Fiat is weak money. Fiat is bad money. Bitcoin is strong
| money. Bitcoin is good money.
|
| Bitcoin is a force of nature.
| aradox66 wrote:
| OTOH, Bitcoin can be subject to liquidity crises which no
| entity could intervene to resolve.
| snapcaster wrote:
| Is that true? I don't doubt it but can you describe what
| that would look like in practice?
| AlexandrB wrote:
| > Bitcoin can not be controlled.
|
| > The world's top Bitcoin mining pools mainly come from
| China, with five pools being responsible for more than
| half of the cryptocurrency's total hash.
|
| https://www.statista.com/statistics/731416/market-share-
| of-m...
| staplers wrote:
| They dont control anything. If they left the network
| would readjust difficulty after 10 minutes and continue
| on the same way.
|
| Seriously educate yourself. You're being the guy in the
| op comment.
| JustLurking2022 wrote:
| It's not about them leaving the network, it's a matter of
| whether they could be coerced by an authoritarian
| government to alter transactions.
| astrange wrote:
| > One of which there will only ever be 21,000,000.
|
| This is a good property of something you want to be
| scarce, but it's a terrible property for money, whose
| purpose is to exist in the correct amount to keep stable
| money velocity in a growing economy. (Or a shrinking
| one.)
| tiku wrote:
| We can divide bitcoin into smaller pieces,
| smaddox wrote:
| Except anyone can start a parallel chain, e.g. DogeCoin.
| The only thing that makes BitCoin special is that it was
| the first and is still the most well known. It's value is
| entirely based on network effects and speculation. It's a
| speculative store of value which can be digitally
| transfered for a modest fee.
|
| Fiat has the force of the state behind it. This has pros
| and cons, depending on perspective and use. As a
| currency, it's superior in many ways. As a store of
| value, it's in many ways inferior to BitCoin... Until
| it's not.
| dorkwood wrote:
| > His response was simply "I never got embarrassed asking the
| computer any finance questions, even stupid/simple ones; I've
| always been the finance guy with answers so I forgot how to be
| humble in unexplored topics."
|
| I think I'm the opposite. My feeling is that anything I type
| into ChatGPT gets stored in a database, ready to be leaked at
| some future date, or read by prying eyes without my knowledge.
| I'm more careful about what I type into it than what I say out
| loud in front of my friends.
|
| Or maybe I'm just paranoid?
| theanti9 wrote:
| https://rabbitu.de/articles/security-disclosure-1
|
| Doesn't it count as paranoia if it already happened?
| malcolmgreaves wrote:
| Wow! Great disclosure! I find it hard to read though, since
| the text doesn't properly capitalize the start of
| sentences. Messes up with the natural language parser I
| have in my head :)
| TeMPOraL wrote:
| Except it doesn't; that's not OpenAI, that's some random
| get-rich-quick startup built on top of OpenAI API. Never
| trust those.
| 01HNNWZ0MV43FF wrote:
| That's true, although the pain of giving up privacy is less
| sharp than the pain of social embarrassment.
|
| Sorta like not wearing a helmet on a motorcycle. Or not using
| condoms. Or not getting vaccinated.
| kjkjadksj wrote:
| Time in the markets beats timing the market if you actually
| have capital to throw around to gleam a decent 5% yearly
| return. If you don't have much capital, you have to time the
| market like hell to build up your capital to a level where 5%
| yoy actually is a decent amount of money.
| huijzer wrote:
| 5% is decent? The index made about 10% on average, Peter
| Lynch about 30%, and Buffett in the '60s was at 40-50%.
| kjkjadksj wrote:
| 5% return is a pretty common estimate for passive
| investing. I certainly didn't come up with it, smarter
| people than me did and spread it all over financial news
| media.
| 01HNNWZ0MV43FF wrote:
| If timing the market works at a small scale, why stop?
|
| If it worked, wouldn't everyone do it?
| wongarsu wrote:
| Your risk tolerance changes. If you are living on
| $4000/month and have $500 to invest, you may as well bet
| those $500 on black. If you have a sizable investment that
| you hope will pay for your retirement you might want to go
| for the solid 10% yoy. Similarly if you manage other
| people's money they might be unhappy with you making 50/50
| bets with that money.
|
| It's the same logic that causes people to buy lottery
| tickets. Winning big in the lottery is unlikely, but
| getting rich by saving two lottery tickets worth of money
| each month is impossible. The risky strategy with the worse
| expected outcome is the only one that leads to your goal.
| kjkjadksj wrote:
| It does work at small scale and people don't stop, make
| careers and billions out of it, see high frequency trading.
| You can play market timer yourself just by using some game
| theory and keeping up with the cart.
| DeltaCoast wrote:
| This looks so fun! I had not considered getting an apple watch
| until now. I mostly just use my phone for spotify and maps.
| LeoPanthera wrote:
| Wrong thread, probably.
___________________________________________________________________
(page generated 2024-07-17 23:10 UTC)