Hedge fund and tech questions that are rarely asked

COWEN: Given all the data that search companies and some of the other major tech companies have, why aren’t they bundled with hedge funds?

SCHMIDT: What do you mean by bundled?

COWEN: Well, literally in the same company. You’d have a tech company and a hedge fund, and there would be a synergy because the hedge fund would use the data generated by the tech company for investment. So the hedge fund would have that data first. We don’t see that in the market.


COWEN: The major tech companies have done very well, of course, but if we imagine some world in the future where some tech companies are at or near insolvency, and if we think maybe they have a fiduciary responsibility to sell off the information they hold on people, is that a regulatory problem we will need to address?

Obviously, a successful tech company is not going to do that. They would wreck their franchise.

SCHMIDT: Yeah, so the problem that you’re posing is, we have a company that has a great deal of useful information that’s also bankrupt.

COWEN: Right.

Those were my questions to Eric Schmidt.


We can flip the first question around and ask why don't hedge funds collect data for trading by offering free software like search and email to users? That wouldn't seem like a good strategy. So, the answer probably is that collecting data directly the traditional way is more efficient than collecting data indirectly by producing free software. Google's data may not be as useful for trading as Cowen's question implies.

The second question is more interesting. How do we know the most valuable use of data isn't more nefarious than the original intended use and why wouldn't economic forces guide such data towards its highest value use? Schmidt answers that the highest value use would be that the data "can be monetized in a good way...that’s valuable and serves...customers’ interest." That's not obvious to me.

On second thought, maybe it is obvious in a slightly different sense. Maybe, gathering data for a particular purpose is not that easy. One has to do a lot of work to gather, process, and organize data before it will be useful for any given purpose. So, once a company purposefully optimizes its data collection and organization for targeting ads, it may be unlikely that the data will inadvertently also be useful for some other purpose. Maybe, there are fewer synergies in data than one might think.

Re the first question, a while ago in 1999 there was an American Express I think offer to provide free stock market trades to all people who provided detailed financial information like their net worth. I got into trouble with my 1% family (even though they are cheapskates) since I naively filled out the questionnaire, then got a call from somebody from Chicago as I recall, who sounded Hispanic, even like a lowly call center person. Freaked my family out as they rightly thought this guy was a gang member (which he sounded like) and I learned my lesson. So companies do this sort of thing all the time.

Re the second question, bankrupt tech companies are often left with only their patents as an asset, which they sell to patent trolls or to their former competitors, to keep the patents out of the hands of trolls. Very not uncommon.

My uneducated answers to those questions:

#1: Maybe the hedge fund people and the tech people just haven't thought of it yet. Or maybe the information that tech companies gather isn't useful enough (yet) to hedge funds. Perhaps I don't appreciate the value of the information though.

Either way, I don't buy Schmidt's answer of (IIRC) "people wouldn't trust the tech company who do that and that lack of trust would ruin their business". I already don't trust them.

#2: "Is that a regulatory problem we will need to address?" Probably should. Schmidt's answer (IIRC again) was something along the lines of "this kind of tech company wouldn't go under". That struck me as kind of foolish. Never underestimate the ability of people to screw things up.

Recently a computer retailer (NCIX) here in Canada went under and their servers were auctioned off... and those servers contained the information of all its customers, including credit card information. That data was then re-sold to presumably nefarious actors. In that sense #2 has already happened, just on a small scale.

Hedge funds are actively buying exotic data streams, just in ways that aren't quite linked to the consumer. Think about the movement tracking devices on container ships or on semis in the US, or FarmLogs in ag. The data are being consumed, but probably from off-brand tech companies and not FAANG

#1 is a great question. Follow up: why is Google amassing so much capital instead of returning it to shareholders?

Libertarian attitudes and tech make for weird questions. Maybe the hedge funds should pair with phone companies. Record all the calls 'mine' the data for information. Would we need a regulatory framework to correct that? Why would that be different then mining emails and web traffic.


Why weren't hedge funds listening in to all those calls to Charles Schwab?

Because it’s illegal? Obviously?

Rand Paul threw a fit in the senate about metadata and the NSA. Too complex and abstract for people to get upset about. Obviously recording their calls would be enough to bring out the pitchforks.

That is what we were talking about, yes, and the absence of similar laws for things like Google search queries.


Google search != phone calls.

They’re not even in the same ballpark.

Feel free to use Duck Duck Go? Problem solved?


Or elsewhere,

"There is, however, an important exception to these laws: provider exception. Under the provider exception, these laws do not apply to "the person or entity providing a wire or electronic communications service." This exception, for example, allows various free email providers (Gmail, Yahoo Mail, etc.) to process user emails to display contextual advertising."


Nothing says evidence like a dead citation link.

There’s a difference between reading emails and scanning for keywords for ads.

But if it bothers you, you can always not use Gmail?

It works for me, maybe you are blocked.

Given that I have installed and operated enterprise servers, there is a great deal I could do. I could rent naked metal and run a full stack of my choice.

Would that solve the wider American privacy problem?

You probably expect me to be and "I've got mine" kind of guy.

'So the hedge fund would have that data first.'

Or you could just cut out the middleman, as the Microsoft embrace and extend strategy (which is now a generation old, by the way) shows. And in the case of alphabet, they are already involved in a number of fields - again, without needing the overhead of a hedge fund.

Do you think Gates, Bezos, Zuckerberg et al are interested in sharing a penny with anyone else?

'but if we imagine some world in the future where some tech companies are at or near insolvency, and if we think maybe they have a fiduciary responsibility to sell off the information they hold on people, is that a regulatory problem we will need to address'

The 90s just flew by in a daze, didn't they? Or possibly even 2015 - 'After filing for bankruptcy in early February, RadioShack is currently making its way through the painful process of figuring out how creditors will be paid back -- auctioning off real estate and trademarks. Also on the list is more than 13 million e-mail addresses and 65 million customer names and physical addresses -- as well as potential information about customers shopping habits.


The potential sale of customer information is among the aspects being challenged, prompting complaints from the state of Texas and AT&T.

AT&T is arguing that some of the data that RadioShack is trying to sell actually belongs to the mobile phone giant. AT&T worked with the retailer to market some of its phones and should not be able to auction off the information about those sales, the company says. The data should be destroyed, AT&T says, otherwise it may fall into the hands of its competitors.


Back in 2000, after the dot-com bubble burst, the FTC sued to stop Toysmart.com from selling off customer data in violation of its privacy policy. In that case, the FTC said that data in question included names, addresses, billing information, shopping preferences and family profiles, including the names and birth dates of children. The data was eventually destroyed. The FTC has also sent letters about proposed data sales in the bankruptcies of other companies, including Borders, XY Magazine, and ConnectEDU.

In most cases, companies are caught violating their own privacy policies, said David Vladeck, the former director of FTC's consumer protection bureau. "The agency's view is that whatever promises are made at the time of data collection would be binding on their successors even if it was a successor who simply bought assets at a bankruptcy auction," he said.

So what would happen if one of the large tech companies fell into bankruptcy and tried to sell off customer data?

Both Google and Facebook have contemplated that possibility and leave open the possibility that customer data could be sold.' https://www.washingtonpost.com/news/the-switch/wp/2015/03/26/bankrupt-radioshack-wants-to-sell-off-user-data-but-the-bigger-risk-is-if-a-facebook-or-google-goes-bust/??noredirect=on

As it turns out, this was the end result in that now 3 year old and undoubtedly obscure bankruptcy case - 'RadioShack struck a deal with a coalition of 38 state attorneys general to destroy most of RadioShack's consumer data, and stipulated that no credit or debit card account numbers, social security numbers, dates of birth or even phone numbers would be transferred.


Most of the assets, including some limited customer information, were purchased by General Wireless, a subsidiary of RadioShack's largest shareholder, which intends to keep 1,750 of the stores open with the RadioShack name and operate its online business. General Wireless agreed not to sell the customer data it is buying to a third party, and to comply with RadioShack's previous privacy promises.' https://money.cnn.com/2015/06/10/news/companies/radioshack-customer-data-sale/index.html

From the wording, one assumes that General Wireless can still send out a Radio Shack newsletter using e-mail, for example.

"so the problem that you’re posing is, we have a company that has a great deal of useful information that’s also bankrupt.......That’s not a scenario that’s likely because, if it has a great deal of useful information, then that can be monetized in a good way, in a way that’s valuable and serves their customers’ interest. So I think it’s an oxymoron. I think it just is not going to happen."

It seems bad management doesn't exist for Mr. Schmidt. Not amazing at all that the CEO is overconfident. However, this may be precisely the future of Myspace which Tyler can see it and perhaps Mr. Schmidt too, but he doesn't want to discuss it on public.

I also liked the part about "glue people". A successful software company is more than a group of over-caffeinated engineers. Some lower-level Google employers have troubles understanding this.

You all have small penises.

What Cowen omitted (because he assumes readers know) is that hedge funds are replacing analysts with computer engineers and quants, whose market predictions are not based on fundamentals of the companies but market (e.g., trading) data (that identify, for example, trends). I sometimes ridicule economists as today's soothsayers, but it's today's computer engineers and quants who are the real soothsayers. I should point out that this new approach to market predictions requires enormous computer capacity, enormous space in which to house the computers, and enormous amounts of power to run the computers. Would HAL, who died a week or two ago, be disappointed that His descendants are devoted to generating profits rather than conquests in space?

Question: will the hedge fund soothsayers follow the patterns they identify in the data or will the hedge fund soothsayers manipulate the data to create the patterns? https://www.nytimes.com/2018/11/19/science/artificial-intelligence-deepfakes-fake-news.html

If Eric Schmidt thinks any business model is resilient to bankruptcy, he's fooling us.

Hedge funds already acquire data in huge volumes. In some cases they have a lot more diverse data than tech companies.

They also buy stolen data illegally — it is quite easy to acquire data from Chinese tech companies and manufacturers, for example, on black markets.

What does Google have that is actually unique? Search data is great for targeted ads but doesn't tell you reliable things about future demand. You can learn a lot more about future demand with satellite imagery, port data, Chinese warehouse data, currency flows, etc. — all of which hedge funds access readily.

I'd say Amazon has the most valuable data, but they are already vertically integrating into the markets where they see wide margins. Why set up a hedge fund when they can invest in those markets directly? A very good hedge fund can return 15-30%/year (unreliably). Amazon has a much high return than that in terms of share value which suggests the are spending their capital better.


While I don't want to let the big companies off the hook, there is no question that second-tier companies are under much less scrutiny and share their data much more freely.

(I can't believe rayward didn't say "planatir" above, or for that matter, "Cambridge Analytics.")

Or this company - 'At i360 it's our goal to help our customers understand and leverage data to its full potential, taking it from informative to actionable for results they can count on.'

And what data might that be? 'Our extensive database of 290 million consumers and voters enables us to to predict behavior and deliver true data-driven strategies with individual-level insights.'


(And don't forget this - 'It’s no secret that cookie targeting is no longer a sufficient targeting tactic on its own. While it does enable you to reach a consistent audience through your offline and online outreach, it’s scale is limited by match rates and expiring IDs. And on top of that, the difficulty of matching cookies across screens presents another challenge.

That’s why i360 is taking advantage of the latest technologies in the digital space including mobile ID matching, direct-to-partner matching, and device graphing.')

Yeah, it is easy to forget that consumer tracking is a whole industry now, with engineers constantly working to improve it.

"I'd say Amazon has the most valuable data, but they are already vertically integrating into the markets where they see wide margins. Why set up a hedge fund when they can invest in those markets directly? A very good hedge fund can return 15-30%/year (unreliably). Amazon has a much high return than that in terms of share value which suggests the are spending their capital better."

Very good point.

Possible answer to #1: Maybe markets are efficient and having that information is worth nothing? ie, it's worth less than the financial professionals would pay for it (not worth sharing any piece of the carried interest).

Revealed preferences would suggest that the data just isn't very good. Much of the time Google's ads simply show me the last thing I bought on Amazon, after all.

I think the question can be interpreted in a broader sense. Google has access to a huge amount of data and has tremendous investments in the best AI and ML researchers and engineers. Is the best end goal of all of this really "efficient ad placement and sales"?

I'm just saying that the fact that the empirical answer to your question appears to be "yes" suggests that the data isn't really that useful.

It's surprising to me that there is so much skepticism about the investment value of Google's data. Google sees what people are searching for before anyone else. Is it so hard to imagine that, of the thousands of geniuses who work there, they couldn't find a handful who would know what to do when they see a spike in searches for "mesothelioma lawyer"?

On a related note, why are these things called hedge funds? Is actual hedging a significant part of their strategies?

I am a bit creeped out by Google, but probably liked a lot of people, I am taking the convenience over the loss of privacy.

I mean, currently I get an email reminder from the library that a book is due, I delete that email, and then on the day of renewal my calendar (with no help from me) says "that book is due!" Or an airline itinerary forwarded to me appears in my Google Now page (I think that's what it's called).

AIs reading my mail to "help" me, and somewhat succeeding.

So yeah, I think we are at some kind of precipice, and as a center-right independent I favor strong national regulation creating data sandboxes for this kind of thing.

This part was good:

So I worry that the sum of people who have special interests, which are governments, foreign governments, special interests of one kind or another, lobbyists, and so forth, will be able to create an environment where people are not able to distinguish between legitimate individual speech and marketing speech, in the sense of stuff that’s being sold to them or peddled to them by some foreign power.

By the way, Ivanka's emails probably were Gmail, but we aren't going to bother you about it. We know you have no interest in "locking her up," because you were never serious about national security anyway.

The farce is already fully exposed.

'Ivanka's emails probably were Gmail'

Not according to the Post - they used Microsoft products.

And leaving aside a number of non-trivial details, not a single piece of classified information seems to have touched Ivanka's in box (or out box, for that matter).

That's not really the source of the farce, the idea is that you could go through the whole "lock her up" cycle that learning email retention rules.

Suddenly the Trumps have no idea.

The farce is overarching, but there is an actual difference between what the former of Secretary of State did handling her e-mails, and what presidential daughter Ivanka did with hers.

However, Colin Powell also broke the rules, and actually advised Clinton to do the same - 'Colin Powell told Hillary Clinton, his successor as secretary of state, that he used a personal computer to email foreign leaders “without going through State Department servers”, a seven-year-old email exchange reveals.

Powell dismissed some of the official security restrictions on him as “nonsense” and questioned why his personal digital assistant (PDA) was any more vulnerable to spies than a TV remote control or “something embedded in my shoe heel”.'

Or in his own words - 'What I did do was have a personal computer that was hooked up to a private phone line (sounds ancient.) So I could communicate with a wide range of friends directly without it going through the State Department servers. I even used it to do business with some foreign leaders and some of the senior folks in the Department on their personal email accounts. I did the same thing on the road in hotels.'


'“However, there is a real danger. If it is public that you have a BlackBerry and it it [sic] government and you are using it, government or not, to do business, it may become an official record and subject to the law … Be very careful. I got around it all by not saying much and not using systems that captured the data.”' https://www.theguardian.com/us-news/2016/sep/08/colin-powell-hillary-clinton-email-state-department

Tyler---why did you omit Schmidt's answer to this question? Because you think it will make you look better?

He annihilated your question. Search relies on users to voluntarily provide their data---it is unlikely that they would do so if their information was being harvested for trading purposes.

It is disingenuous to omit his answer.

Without defending Tyler or Google, consider how many lesser investment pages have "quick links" to quotes for stocks mentioned in an article.

It is certainly easy and low tech to make the JavaScript on those pages update an internal database for that investment website and to allow a control panel where hosts can see the pulse of reader interest.

Does anyone think about that?

"If you owe the bank $100, that's your problem. If you owe the bank $100 million, that's the bank's problem." - J. Paul Getty

There's a corollary in tech. If a solvent company has your personally identifiable information (PII), that's their problem. If an insolvent company has your PII, that's your problem.

When I worked at Google X, this came up as a business model for new bets. It was always shot down immediately because it might have severe regulatory consequence for the main company.

And there are tech companies trying to do a version of this bet. The ones in satellite data are really on this train. See Planet Labs or Descartes Labs.

Former hedge fund quant in a big tech company here. I asked the same when I first arrived. The best answer I can give is that most tech companies are built upon the trust of a very large mob that we call our customers. And as with all mobs, the moment they get even a whiff of anti-social behavior from you, they will attack mercilessly. For now revenue and growth in tech is beyond the wildest dreams of the likes of Two Sigma, Winton, Element, etc., so there is no reason to take that risk. For now. The second best reason I can provide is, public markets are for chumps. Sure they are liquid, but you are fighting for scraps. With all that data, it makes a lot more sense to make early, big investements in new business or fixed assets. Your average models for data center demand or the value of a new product are a lot more reliable than any quant model of the markets. A straight forward application or Markowitz's portfolio theory says that investements with greater reward and lower risk such as the kinds tech companies make dominate strategies with lower expected returns and higher risk, hence zero weight on those. What is AQR's Sharpe ratio again?

Aren't hedge fund quants using data to identify patterns and thereby predict markets? I may question whether quants are better soothsayers than analysts, but hedge funds are replacing analysts with quants and those who manage hedge funds aren't stupid. My question is where do quants fit in at hedge funds? Can they become partners and become rich? Or are they treated as freaks who will work for high pay but never become partners or rich?

Isn’t AQR’s sharpe ratio around 0.5? Not a barn burner, but better than nothing.

Things may be changing in big tech, there are more and more teams dabbling in this area.

Thanks DF. This makes more sense than my own disconnected musings.

Nailed it.

Leverage and illiquid assets.

I am gonna be a bit of an a$$hole and paraphrase Richard Dawkins: "Just because you can formulate a question, doesn't mean it's a good question".

They are, except they specialize in investing in startups. Companies like google get a lot of insight on which sites are "hot" from their internal data. That's the most direct application of their info (who clicks on what, and what sites to surface to the top) to investing that you can imagine. Plus it doesn't run into problems of insider trading since it's not public companies.

hedge funds are getting access to search engine like data but in different ways.

Most hedge funds have VC arms that make investments in private companies not just on the basis of an investment thesis and target return. They make investments to get access to unique data. E.g. I may invest in a failing business because that business has unique data about consumer behavior that provides signal as I invest in public equities.

Re #2: It's already here -- Viant bought Myspace profiles in 2016.


Comments for this post are closed