A synthetic control causal analysis of Musk's tweet on Tesla stock price | Hacker News Reader

A synthetic control causal analysis of Musk's tweet on Tesla stock price

alexpghayes.com

105 pointsalexpghayes73 comments

Despite the oddly enjoyable humor from the situation, performing an analysis like this is important to do because the vast majority of finance headlines are just instantly made up after something occurs, and they often mistakenly imply causality when there is no evidence of it, or sometimes even evidence again it.

Don't let the Bayesian rhetoric deceive you — this analysis succumbs to the same pitfalls of mistakenly implying causality.

That's true, there's a lot lacking here, including looking at TSLA's volatility. But, it's much better than providing zero evidence imo.

it's hilarious, not because musk shot himself in the foot (in the long-run he likely didn't), but because he tweaked the whole gambling industry otherwise known as the stock market, day traders and algorithms alike, in 7 words (9 actually, but whatevs) on a popularity contest, aka social, network. he's too cool for school.

> it's hilarious, not because musk shot himself in the foot (in the long-run he likely didn't), but because he tweaked the whole gambling industry otherwise known as the stock market, day traders and algorithms alike, in 7 words (9 actually, but whatevs) on a popularity contest, aka social, network. he's too cool for school.

I hope he burned a bunch of the WSB troll's positions they paid with their stimulus money with that tweet. I'm going to go check the subreddit now to see the dismay and Robinhood screen caps in all red now [1] [2].

I also find it amusing, but then again I also found the 'Bankwapt' tweet funny, too. Anyone who bought that dip and rode it to 9XX did also.

1: https://www.reddit.com/r/wallstreetbets/comments/gbod91/too_...

2 https://www.reddit.com/r/wallstreetbets/comments/gbsjhe/didn...

First of all, for every buyer there's a seller, so for each person that lost money on Tesla, somebody made money (relative to the benchmark).

Secondly, I don't know why you think only "day traders and algorithms" were affected by this tweet. I assure you, retail passive money (e.g. pension funds) also have exposure to Tesla.

Thirdly, why so gleeful about someone's perceived loss?

1) there are winners and losers in gambling too. 99.9% of trades are gambles, not value investments. you can more charitably call it providing liquidity to the value investor, but that's only true of the better known ~20% of equities.

2) retail passive funds are long-term investors, and pricing blips like this will be washed away in the long run. and it's likely a tiny exposure relative to the overall size of the fund.

3) it's an ingenious and mirthful move to target the greedy vs. true investors. if there aren't negative repercussions to the gambling, you'd lose the drive to optimize mispricings and arbitrage opportunities out of the system.

edit: note that the likely purpose of opening up the market to "unsophisticated" investors is to provide added volume and liquidity to "sophisticated" investors, not to provide economic opportunities to the unsophisticated.

Ha. But this approach seems really problematic since it's based on the assumption that algorithms are good at predicting the counterfactual, while the reality is that the stock market is by nature extremely unpredictable.

Example 1: You could imagine the US gov announces a national quarantine, and Elon tweets this in response 1min later. The tweet didn't cause the stock drop, the quarantine did (a classic C caused both A and B problem). But this algorithm wouldn't know that.

Example 2: Imagine a court case ruling against Tesla is announced, and Elon immediately tweets. Same issue, but harder to detect. In the quarantine case you could assume Elon doesn't influence stocks besides Tesla, and observe that all of them are crashing.

Example 3: A really good earnings report for Tesla is announced, at the same time Musk tweets it's overpriced. The price cancels these out and stays steady. Now the algorithm gives a false negative, since actually the tweet prevented the price from going up.

In your case 1, doesnt the decorrelation from the index deactivate any effect that the gov quarentine would have?

The algorithm isn't so much predicting anything here though, it's retroactively applying prices based on known correlations with the S&P 500.

That's a prediction.

For Example 1, wouldn’t the entire S&P index fall, this showing Musk isn’t responsible,

This is bad analysis. It's not even close to absolute.

He's not accounting for the volatility of the stock. That's the reason that premiums on tsla options have been and are so incredibly high.

He didn't even mention that the SPY is much less volatile over its entire history than TSLA individual stock considering it's an index of 500 large/mid cap stocks. Of course it's going to be less volatile.

Did he tank it? Maybe, but every tech stock had a dip around 11am. If he tanked it then the sentiment analysis bots tracking his tweets need to be tweaked a bit.

AMZN's dip [0] looks nothing like TSLA's [1] today though.

Founders commenting on the valuation of their company is absolutely tradeable news to day traders.

I would argue that this tweet directly affected TSLA tanking in price immediately after it landed.

[0] https://www.google.com/search?client=firefox-b-e&q=amazon+st...

[1] https://www.google.com/search?client=firefox-b-e&q=tesla+sto...

AMZN is an outlier today because they reported earnings last night. Better to compare it to an index.

... and also skewed by the WSJ article on unfair third-party seller treatment that contradicts previous house Judicial Committee testimony and may lead to new hearings...

The problem with muddy water is it's muddy.

The DOW looks almost identical to AMZN.

I agree. Synthetic controls can be powerful but they are not magic wand for causal analysis.

The fit on the left seems too good to be true. It could be that it using the outcome (Tesla's stock price) as part of the calibration, which biases your synthetic control. In ML speak it is an overfitted model. http://www.oekonometrie.uni-saarland.de/papers/SCM_Predictor....

Also, in the package documention it says "The model also assumes that the relationship between covariates and treated time series, as established during the pre-period, remains stable throughout the post-period". In this case the covariate is the S&P500, so clearly the relationship didn't stay stable. As you mention, there was a dip in every tech company.

This doesn't mean that Musk's tweet didn't affect the price, but this is no proof.

Does anyone know if Twitter are allowed to legally frontrun the markets before making tweets publicly visible?

Almost certainly not, sounds like textbook insider trading to me?

Probably wouldn’t be. In any regard could sell an early access feed if they wanted and it’d be legit.

https://developer.twitter.com/en/enterprise

They could do this, but then Twitter almost certainly wouldn't be considered a valid instrument for releasing financial data (as it is right now):

https://www.sec.gov/news/press-release/2013-2013-51htm

THAT is a very interesting press release (and itself raises the slightly circular question: are SEC press releases considered a valid instrument for releasing legal guidance on how to comply with SEC regulations?).

But: If the SEC regard Twitter as, to all intents and purposes, a neutral 'wire service', then that means the SEC allows Twitter users to assume - regardless of what Twitter's own ToS say - that Twitter distributes information fairly and impartially.

Weirdly, it's not clear that the SEC has any way to regulate Twitter to ensure that they actually meet that assumption. If they were to discover, for example, that Twitter DID route interesting tweets to their in-house investment management team before publication, it's not obvious that the SEC could do much other than issue a press release saying 'sorry, looks like our advice about this was wrong, Twitter turns out NOT to be an impartial information distribution channel, please stop assuming it is.'

I wonder to what extent the existence of that SEC guidance affects how Twitter and Facebook behave? Do they consider, when deciding to implement features that modify the flow of postings from publisher to receiver, whether they are creating a mechanism which contravenes the SEC's understanding that they implement a 'fair disclosure' mechanism? Should they? Should there be regulatory compliance reasons that oblige them to?

And, per the SEC's guidance, this also applies to "social media outlets like Facebook and Twitter", whatever that means. Baidu, maybe? Is TikTok a legitimate channel to release material information? Do TikTok's engineers know that?

This world we've built... it's really weird.

This is, in fact, the exact opposite of insider trading.

No one working at Twitter is also working at Tesla. Ergo, no insiders.

Anything that Elon sends to Twitter is instantly considered public knowledge and its data belongs to Twitter. If they delay his tweets to make trades that's well within their rights.

actually, I think this is exactly what misappropriation theory is supposed to account for. can you explain more around why it is well within their rights? I don't think it is.

Are journalists and editors?

You could make some pretty big moves just with a dramatic headline and a buried lede

I don't know if it's explicitly part of the regulations but as a former "insider" at a large publicly traded valley tech company, I can say that it's almost certainly against the rather all-encompassing wording of the shit I had to sign that enumerated all the "Thou Shalt Nots" from the regs (as well as watch the insufferably annoying 'compliance training' videos (with interactive quizzes!)

Basically, it went far beyond all information about my employer to include ANY "non-public information" about any OTHER company that I might become aware of through my employment, whether directly or indirectly, including (but not limited to) any product plans, business strategies, etc (or lack thereof), I was prohibited from trading on or disclosing to any other person, yada, yada, yada.

Since I had, at least, theoretical access to tons of customer data on our servers as well as forecasts based future contract negotiations with our many thousands of huge public co customers... per my attorney's advice, I just held all my company RSUs and put the rest of my portfolio on auto-pilot in a Boggle-headish standard blend of cheap ETFs and never traded any individual stocks. This happened to work out extremely well for me financially as well as simplifying my life and reducing my exposure to lawsuits and regulator scrutiny. It was a combo of luck and the fact that for ten-year time periods, that strategy is actually pretty good.

The rules that govern what individual employees are allowed to do privately with the information they come across in the course of their work which belongs to their employer are very different to the rules that govern what the company itself is allowed to do with the information it has access to.

Of course you couldn't personally beneficially trade on information you accessed in the course of your employment (although primarily that would be a breach of your employment terms, not necessarily a breach of the law...)

But that says nothing whatsoever about what you can do with that information to benefit the company.

Employees at Twitter would go to jail if they used Twitters information to trade. Journalists have gone to jail for sharing information about stories they are writing to third parties.

But it is possibly legal for Twitter the organisation to do that or sell tweet "firehoses" to third parties.

I would not be surprised if they had deal with with traders that gave them a firehose of tweets milliseconds before everyone else.

There was an interesting case in the UK where a company that provided the audio feed for the Bank of England announcements fed the feed to hedge funds before everyone else heard it.

https://news.sky.com/story/bank-of-england-cuts-audio-feed-o...

Edit: this is my uneducated opinion not legal advice.

They would? Are you sure about this? Can you cite an authority on it? I don't know either way, except that I know from years of reading Matt Levine that the laws on insider trading are a lot more complicated than you'd assume.

I love Matt's views on the subject. At its core, as I understand it, the issue is who owns the information, and is Twitter licensed to distribute it as they see fit.

https://www.quora.com/Securities-Law-Is-it-legal-for-a-journ...

I agree that a citation is needed ;)

But in this case I think the case history points to OP being right, though you'd likely have to have a pretty egregious and clear pattern of such activity, and it would have to be deemed material information:

https://www.reuters.com/article/us-sec-capitalone-insidertra...

The Vitaly Korchevsky indictment probably comes close to demonstrating a model of what this would look like:

https://www.justice.gov/usao-nj/file/765216/download

That Bank of England case seems to have been handled as a breach of the terms of the contract between the Bank and the supplier, not as a criminal matter. The referral to the Financial Conduct Authority might suggest that the hedge funds that purchased access to the feed might have breached some regulations by using a service whose existence was not sanctioned by the Bank... but it's definitely not clear to me that any of the same sort of relationships as exist between the BoE, regulated hedge funds, and a tech company with a supply contract with the Bank, also exist in any meaningful way between Elon Musk, Twitter, and Tesla...

It's not immediately clear to me that they can't.

Twitter makes NO guarantees that they will deliver every tweet to all consumers simultaneously, or even at all. They don't guarantee they won't hold the tweet up while they decide how to process it - they certainly analyze tweets for content and automatically hold some for manual review, for example. So why on earth couldn't they accept the tweet, process the information in it, and make corporate decisions about how to, say, adjust their investment portfolio... before they put the tweet's contents out on the wire for the world? Is there anything in the ToS that says they can't act on the contents of tweets you send them? You shared the information with Twitter with the intent of publishing it to your followers - but the agent you handed the information to first was Twitter.

There's a sci-fi black mirror kind of scenario in there, too... It's quite possible that if the president of the US were to declare war, the first place that information would reach outside the White House and Pentagon would be a Twitter HTTP service.

Think about the responsibility the software that has to handle, route, and process that request bears. Think about the threat model the systems that software is part of need to consider, if that really is potentially one way it could be used.

Worse still, it might not even be the 'send tweet' handler that gets notice first; the twitter client into which the message is being typed, sending back analytic data to help optimize its autocomplete suggestions or pushing data to the server so it will be ready to offer an appropriate gif to accompany the tweet, might be sending back the draft text of the announcement before the president hits 'Send'.

So here's your story writing prompt:

A Twitter engineer reviewing logs trying to track down a bug in the gif autosuggest algorithm chances across a series of requests from the president's iPhone in the last two minutes, containing text from a tweet being composed, that look for all the world like the drafting and redrafting of an announcement that the country is at war...

What does Twitter do?

This is remarkable. I would love to see this in black mirror

Block the tweet? It's like the USSR guy who decided that the early warning signal was incorrect, and prevented WW3.

This idea of unilateral "freedom of speech" is beyond goofy - some actions that deliberately cause massive amounts of distress, even in the short-term, must be attenuated in some manner.

If Trump announces the US is at war, for most purposes in the short term (until congress steps in to assert their power) the US is at war. It is not a false alarm.

If twitter had some reason to believe that the tweet was a false alarm about going to war, and not an official announcement by the US military's commander in chief, that might be a reasonable to take. Perhaps if they also believed the account had been hacked or that a staffer had been duped into thinking that they were supposed to announce a war when they weren't. That's not the hypothetical situation here though.

Sounds like a good cliffhanger leading into the third act, though.... :)

IANAL, but it strikes me as 100% illegal for Twitter to trade on their tweets before they are released (at least for the class of tweets which contain corporate announcements - tweets from e.g. the persident might be treated differently).

Some of those tweets can be resonably be construed as being MNPI, so trading on it before it's published on the platform is most likely illegal.

Can newspapers/journalists trade on a piece of news before they break it?

Arguably the information is not public until it is available to more than just twitter employees. And trading on non public info can be securities fraud.

If they were doing that, then they are not doing a very good job at it.

Did Elon also cause AMZN stock to drop around the same time? And FB? And MSFT? And UBER? That’s a whole lot of clout, unless I’m misunderstanding something here...

No, and I don't believe the author claimed this.

I believe GP's point is that this is coincidental broad dip, not TSLA dip though of course the downside with rhetorical questions like that is that you have to guesss at meaning.

WTF, you're just spreading nonsense. Take a look at this image and see if you can tell just looking at it which of them is tsla and at what time musk commented: https://imgur.com/OwwKVyf that includes every one of the stocks you mentioned and there is one clear outlier. No other stock has a clear change at that time, and the behavior of the one stock is clearly different than the rest.

I do not doubt you could perform this same analysis limited to tech companies and get the same result (though comparing tesla to these tech companies is rather irrationally aspirational).

Your chart tells me TSLA is a more volatile stock. Everyone knows TSLA is as volatile as a cryptocurrency. Why do you want to believe in god-like tweet powers so much?

"Of course, you probably shouldn’t believe these estimates since I didn’t generate them with neural net."

The current status of "AI" summed up.

I especially like the reply to Musk's tweet: "dude, I just lost $10k because of this tweet".

So you've invested at least $100k on the highly volatile stock of a luxury car company, overpriced by any reasonable standard, in the middle of a world pandemic with half of the world in lockdown and a huge economic crisis looming. Headed by a brilliant but slightly deranged CEO with a passion for tweeting anything that passes through his head.

I mean, you've been looking for it, didn't you? You wouldn't have complained if Musk had tweeted some bs about Teslas getting warp drive next years via OTA update.

It's more likely this particular trader was selling options on Tesla stock that were only relevant when the stock price moved dramatically.

maybe she bought the stock years ago

Then she didn't lose anything, and in fact has made quite a lot!

It looks like the guy who tweeted that is, by his own profile picture, clearly a man.

If he doesn't like Elon or Tesla's board of directors, that is fine. Everyone is entitled to their opinion.

But it is like buying a utility stock and then complaining that they are boring and slow to innovate. It may be valid criticism, but what did you think you were buying.

It's far more likely that he held some options that became worthless due to the sudden drop in the stock price. I love Elon and his vision but tweeting that was irresponsible.

> overpriced by any reasonable standard

Except of course the standard that has historically proven to be the best one, ie the market.

I have some stock to but I'm not an idiot and check every day and freak out how much up and down and freak out about it, its a deliberate long decision. Elon has made comments like this before, quite often in fact.

> the standard that has historically proven to be the best one, i e. the market.

Well, the market isn't the best standard, it's the only one. The market is the price. That doesn't mean it's reasonable (otherwise bubbles wouldn't exist).

Well, you an also look at expert predictions and many other things. Generally the market performs better then those.

So on avg, if you disagree with the market you are likely wrong.

In what sense the market performs better than expert predictions? Isn't it like saying that the weather performs better than the forecasts?

This doesn't actually prove anything, right?

Isn't this a textbook case of how a confounding variable could explain the effect?[1]

It's also a bit disingenuous to claim "before the tweet, Tesla stock prices are well-predicted by the S&P 500" -- anyone can take a quick glance at 1 or 5 year charts of SP500 vs TSLA and disabuse themselves of that notion quite quickly.

[1]: https://en.wikipedia.org/wiki/Confounding

So from the "facts" Ive read on the internet, a reputable source, Musks girlfriend is mad at him. Maybe he has a self destructive streak. Maybe he is bored and wants a new problem to solve. Maybe he has substance abuse issues.

That said, I dont know what compels people to tweet things that just end up complicating their lives. Twitter seems to be a helluva drug to some people. Maybe the fact that my dormant account has like 4 followers is a blessing...

For people like Musk, attention is the drug. Twitter is just the conduit.

Now I wanna see your topology assignment

(not joking; it sounds like a fun way to procrastinate)

Event studies have been around since the late 1960s.

Elon downgrades the stock and market agrees with him. That's my analysis.

Whatever Musk tweets, be it the stock price is too high or too low, doesn’t affect its true value one bit. OTOH, you can be pretty sure he’s doing everything he can to increase the value of Tesla though.

Value is whatever price a buyer will pay for something at the time of the transaction. There is no such thing as 'true value' in capital markets.

How much whiskey is there in the world right now and how can we get it delivered to Tesla’s compliance department?

> How much whiskey is there in the world right now and how can we get it delivered to Tesla’s compliance department?

Wishing alcoholism on our enemies now are we?

Alcoholism? Enemies? You got all that from a joke about how much stress Elon creates for his own people with his inanity?

I'm really, really hoping that Elon actually asked a securities attorney and found out that if he is not short any Tesla shares, that he can just tweet that without repercussions.

I mostly hope this because I'd really like the option to die on Mars (by which I mean, of extreme old age in a comfortable bed in a luxuriously scenic habitat). Thus, I choose to believe that Elon's occasional manic episodes are just stress and sleep deprivation from obsessive overwork.

Any serious analysis would find that the fate of Elon Musk has nothing to do with your likelihood of dying on Mars.

Reply on news.ycombinator.com