ARTIFICIAL INTELLIGENCE: Self-driving cars and self-speaking news anchors inch us closer to dystopia.

Financial services regulators have been so hard on crypto currencies, roboadvisors, digital lenders and payments companies. It's as if that money is a life or death situation! But getting a permit to drive a robot car on a public road without a human being holding the wheel -- not a problem in California. Waymo, which is the Google car spinout, has been given the green light to put 40 autonomous cars on the road. This is already happening in Arizona, with 400 users that can get into a robot car via an app around Phoenix.

3dc3f300-8696-4d51-ae39-a4581f08f17d[1].jpg

We don't want to be alarmist, of course. Statistically, these machines are likely much better than humans at driving -- they are just more likely to make mistakes that humans would think are preventable. The same process took place in regards to machine vision, with early prototypes making classification mistakes between cats and dogs; now, such algorithms can tell apart the difference between hundreds of breeds. So we hope to see similar progress as driving and visual data is incorporated into autonomous car systems. We'd be remiss not to mention our white paper on the topic, which models out how the insurance industry may lose its lunch when cars don't crash. On the other hand, we note that the DMV required a $5 million bond to put a self-driving car on the road, so the risk is still wildly unknown.

fd5430d6-f02d-4862-8d0e-321cf43a30af[1].png

In a more sinister move, China's state-owned news agency recently launched "composite anchors", which is a machine vision version of a news anchor that can be manipulated with text. Here's how it works. You shoot dozens of hours of video of a person speaking, and then spin up neural networks that can (1) manufacture sounds similar to the target's speech and (2) manufacture video resembling the human making that speech. Presto -- just type in whatever into a command box, and your generated anchor will say it, in any language you would like. Given the recent video editing experiments that the White House supported in relation to denouncing a journalist, we are acutely terrified of how this can impact the attention economy. Not to mention the implications for selling a human likeness for endless manipulation. 

ad628a3f-c9da-4f6b-8136-24ca748f5c44[1].png

Source: SF Chronical (Waymo), South China Morning Post (AI Anchor)

ROBO ADVISOR: $40 Billion Per Month Goes from Active to Passive, But Robo Performance Mixed.

One part of the digital investment management story is the shortening of the value chain in wealth and asset management. As active asset managers (fund manufacturers that pick investments to create alpha) face compression driven by asset flows into passive products -- indexes packaged in ETFs -- one answer form asset managers have been to build out their own distribution channel, where they control asset allocations. This is why roboadvisors have primarily gained traction with manufacturers (revenue sale) and not distributors (efficiency sale). So let's highlight a few relevant data points.

First, Autonomous asset management analyst Patrick Davitt just put together our October sector data, which is highlighted below. Looking at over 9,200 active funds and $9.3 trillion in assets, a full 63% under-performed their benchmark in October. Out-performance in a down-market is supposed to be the reason active management exists! As for 2017, there was a 50% chance of out-performance, a coin flip on whether it's better to hold an active fund or just the index. In terms of actual assets, regardless of market environment, about $20-40 billion is flowing out of active funds and into passive funds. Hard to find a more clear example of a secular shift. Part of this story of course isn't fair to fund managers. When bad things happen in an active fund, you can blame and fire the fund; but in a passive index, you blame the market and hope it recovers. This is a permanent, psychological disadvantage.

0a94d2ce-f031-4558-80ee-acb64abfb084[1].png
7aaadd83-2e04-4578-9c5c-ddfef4ee976c[1].png

The second part of the story is the fantastic Backend Benchmarking Robo Report (link below). The analysis follows the performance of 24 roboadvisors, with several over a 2 year horizon, which we partly highlight. Notably -- Merrill, TIAA, Zack's and Morgan Stanley are all listed as incumbent robos. Our estimate of $600 billion in the strategy feels increasingly correct. In the charts below you'll see 2 treatments of the data: (1) annualized returns vs standard deviation, sized by Sharpe ratio and colored by incumbent/startup status; and (2) an upside and downside capture ratio plot, which shows how good an allocation is at capturing alpha during market momentum. In the first analysis, incumbents like FidelityGo and Vanguard look stronger than the independents in terms of the unit of return per unit of volatility. In the capture category, TD Ameritrade, Personal Capital and Wealthfront stand out. Merrill Edge is the worst on capture, and FutureAdvisor has the worst 2-year performance. What's most telling perhaps is that 77% under-performed their benchmark (as set by this third party) in Q3, and 82% under-performed over a 2 year period. Hard to fire the whole market.

0560cd0b-956e-4272-a816-6df636595c25[1].png
de074906-261f-471c-b097-447ab1a79d2a[1].png

Source: The Robo Report (download here), Autonomous NEXT (robo AUM), Autonomous Research (flow data)

ONLINE BANK: Lloyds to cut 6,000 jobs to create 8,200 digital ones, Natwest launches SME neobank

Digital hurts! In a sort-of-confusing announcement, Lloyds is getting rid of 6,000 jobs; but it's also adding another 8,000 jobs, for a net gain of 2,000, as part of a £3 billion plan to invest in digital banking. Why have a call center in Kent, if you can have a chatbot in Facebook Messenger? Reportedly, many of the existing staff will be re-skilled for new roles. But the reality is economic dislocation as a paper industry moves online -- data scientists and engineers are not the same as branch operators and lending officers.

As another example, take RBS/Natwest and their latest launch of Mettle, an SME neobank. The tech was built mobile-first by fintech consulting outfit 11:FS and Capco, with the capabilities of opening a business current account in minutes, build invoices, and automate payment reminders. Business financial management and forecasting would sit on top -- trying to apply the personal financial management concepts of the retail market in an SME market that would get immediate, tangible value of a Quickbooks with a bolted-on bank account. Think about who built this thing -- a third party composed of entrepreneurs who launched a set of neobanks and roboadvisors in the UK (Monzo, Starling, Nutmeg). You can't get something new without trying something new.

And the last data point is Zopa, which is the UK-based digital lender that hasn't gotten public (that would be Funding Circle). They've just raised £60 million to build out another next gen digital bank. The company already has the revenue side built out in place from p2p loans, having lent about £4 billion of personal credit since 2005. But without a banking license to take deposits, it doesn't have reliable capital for the bumpy economic cycle. Like every other personal finance Fintech out there, the company plans to offer savings accounts, credit cards, investments, and various other financing options. Everyone pivot together now!

b8e0a737-750f-4321-a2b2-60fdeadcf7e8[1].png
581c5940-5ff8-4c76-b0f0-d0b4b81f433e[1].png

Source: Finextra (Mettle), ComputerWorld (Mettle), Independent (Lloyds), Techcrunch (Zopa)

CRYPTO: The $1 Trillion market where Stablecoins will succeed and why most will fail

Last week we talked about the value of having a bag of cash in a down-market, using Circle as an example. This week, the news broke of Coinbase raising $300 million from Tiger Global (which had also invested in roboadvisor Wealthfront, among others). Circle and Coinbase earlier joined efforts to popularize their stablecoin USDC, their version of crypto cash pegged to the US dollar, which has reportedly had over $125 million in circulation since September. Meanwhile, USDT (Tether) associated with Bitfinex has been seeing outflows and general anxiety about whether the currency is a fraud -- with the total market cap falling by over $1 billion. Tether just released a statement from a Bahamas based bank that claims the firm has $1.8 billion in portfolio cash value; however, this statement was not signed by a named officer and disclaimed all liability. So at the very least, we can say that Coinbase+Circle seem to be forming a more credible stablecoin alternative than Bitfinex+Tether in the short term.

But what should we think about the usefulness of stablecoins in the first place? The core thesis is that BTC has not been used as a currency because of its volatility, and therefore merchants and individuals would not rely on it as a unit of account or medium of exchange. This premise is not entirely true -- volatility is only partially explanatory of why BTC is not being used by consumers. In our view, the main barrier is not volatility but ease of use and form factor. It's just too hard to figure out how to actually pay with BTC or any other digital currency for real (i.e., non digital) goods and services. Second, volatility in Bitcoin has actually subsided over the last 6 months -- that's not enough for long term company planning, but if it were the problem in commerce, then we would have seen a spike in economic activity correlating to this volatility damper. 

f2f6040d-f443-4794-8f03-eed76fd9fe6b[1].png
5bbd7a39-0850-4199-b07e-81ff1fd2f21e[1].png

Any floating currency needs to be collateralized, whether or not it is printing money algorithmically or has bots arbitraging itself against exchanges. Otherwise you cannot fund redemptions (and if you can't fund redemptions, then you are just printing specious moneys). Holding the peg to your desired currency basket, whether USD, yuan or Euro, requires being able to defend the currency with capital reserves. Any private capital reserve can be broken by a larger private capital reserve -- or even by a government actor. Consider Soros and the Bank of England. As a result, these coins are fragile and ripe honeypots for attack and manipulation. In the case where the reserve becomes so large as to be unbreakable, and where the currency is meaningfully used as a medium of exchange, it becomes a threat to the world's actual reserve currency, the USD. The US sovereign is unlikely to allow private parties to issue and own a digital dollar at scale -- though the Treasury may be catalyzed to mint digital dollars as a result.

Here's what we think will work -- private company networks that ride the blockchain rails with the equivalent of a Cash Sweep account or a Money Market Fund. Imagine opening up a Schwab brokerage account. Your free cash in a portfolio -- let's say 1.5% -- would get invested into a cash sweep vehicle, which could be a money market fund, or a trust company cash account, or something similar. For a crypto financial company, you are unlikely to want to hold a financial license for traditional banking or investment services. But you still need to manage the cash somehow. So efforts like UBS settlement coin, or any of the exchange-backed stablecoin projects, could fill in the gap of moving USD around within a limited size network in order to reduce friction between going in and out of fiat. If the network gets so big as to include the entire economy, then it again pops up on the Treasury's radar. That's not to say it's a dead end -- MMF assets were nearly $1 trillion for retail and $1.8 trillion for institutional investors. And banks print money by issuing credit all the time, levering up the economy many times over, they are just heavily regulated to do so.

610db5e5-08e1-42f3-ae60-d4af4faf8b27[1].png

Source: Cointelegraph (Coinbase), Bitcoinexchangeguide (USDCCirculation), Centre, Medium (Bitfinexed), Bloomberg (Tether), Coindesk (Tether Bank Statement), Investopedia (Sweep account), ICI (Money Market Funds), Bitcoin Volatility 

ROBO ADVISOR: Titan startup mimics hedge fund trades, repeats mistakes of the past

339e5e95-71c6-45e6-9d07-117635426b77[1].png

It's 2018 and startups like Titan are still launching B2C roboadvisors claiming to invent the "modern, mobile version of BlackRock". Did we forget that FutureAdvisor, a modern, mobile version of a money manager, was bought by BlackRock in 2015 for $150 million, and is now being deployed both B2C and across financial institutions? Or that SigFig (previously WikiInvest) has gone through the same pivot, and is now powering financial advisor platform CoPilot for Citizens Bank, backed by UBS. Or that HSBC just signed Marstone as its provider of similar software? Or that WisdomTree did the same with AdvisorEngine, or Invesco with Jemstep?

Titan scrapes hedge fund filings data in order to mirror their purchases into a basket of 20 stocks for the price of 100 bps per year, which is 2-4x more expensive than most roboadvisors. This was also done before. Remember AlphaClone, or Covestor (sold to Interactive Brokers), or Motif (now sells IPOs), or Kaching (now Wealthfront)? The idea that there is a "pro-sumer" audience that wants to delegate investing a little bit, but still retain control to pick directional themes, has been repeatedly proven wrong. Having raised $2.5 million and grown assets under management to $20 million does not change the underlying issue -- the market does not exist at scale.

If you think we're being too critical, here's what appealed to token Millennial Matt Low from our team: "We are all human and succumb to peculiar logical blindness when the words “hedge fund”, “algorithmic trading” and “Mobile BlackRock” are placed together in the same article. This was particularly the case when reading up on Titan, which appeals to my mindset of supporting anything but the glass tower financial monoliths of Wall street and Canary Wharf". Fair point, Matt. But there's only so many of you to go around. To make Titan actually work, you'd need to funnel in $100 million of growth capital to acquire customers, cross-market banking, payments and insurance products, and then sell the whole mess to BlackRock.

e7a74c10-acb6-404c-8c91-21973670d646[1].png

Source: TechCrunch (Titan), Wealth Management (SigFig), Company Websites

ARTIFICIAL INTELLIGENCE: From BMO's Chatbot to a full virtual avatar

Let's paint the progression. Finn.ai, a chatbot company that integrates banking services into Facebook Messenger and other chat channels, launches Bolt for the Bank of Montreal. The project took 10 months to private label and deploy to BMO's 12 million customers, covering 250 questions at launch. As the app gathers more information about what customers ask, its usefulness grows and it becomes an increasingly relevant channel for customers to ask their financial institution informational questions.

bd55987e-762a-40c2-97ba-066f2fd58f5f[1].png

By the end of 2017, 40 million smart speakers were installed worldwide, with 2018 projected to land at a 100 million install base. People are getting more than one instantiated smart assistant -- littering their home with several Echo dots. And, reportedly, Alexa lives in 3,000 others types of smart-home devices -- giving this bot army 45,000 skills, from Spotify to financial conversations. Google and Apple are working to catch up to these numbers, releasing eerily realistic robo-conversationalists like Google Duplex that can answer telemarketing calls and book hair salon appointments. And maybe cancel your financial subscriptions. What's perhaps most important is that an answer to a voice query is 40-80% fulfilled by the "featured snippet" at the top of a search engine's list, according to Martech. That means no more long tail of any kind, full stop.

c12bc0cc-7c56-4e78-93a0-656a2d02b04d[1].png

Eventually, it is no longer enough for our avatars to be disembodied functions powered by a retail product recommendation engine. From virtual worlds and into augmented reality, agents will take on hyper-realistic rendered physical bodies. A recent Intel whitepaper describes how machine learning is now being combined with a modeling of animal and human bodies under a physics simulator to quickly and realistically build CGI for games and movies. Instead of an artist using intuition to draw the perfect frame, machines build skeletons with physical properties, connect bones with digital ligaments, fire up virtual muscles, and package the final version in the species of choice. Machine learning algorithms add realistic, generalized movement to the equation. They can also add speech, appearance and function. And maybe even help the UBS chief economist look a little bit more lifelike!

f0b0778c-0de0-4ee2-8466-25dc54c01e18[1].png

Source: Finn.ai BMO case study, Fast Company (Amazon AI), The Atlantic (Smart Speakers), Martech Today (40% answers from featured snippet), Intel (Rendering CGI), Financial Review (UBS CIO)

VENTURE CAPITAL: What Goldman-Backed Circle, VaynerMedia and Social Capital have in common

There's a feeling brewing out there, up in the atmosphere. Symptoms are starting to coalesce. And the feeling is: cash is king, venture capital is mis-configured. Maybe something about endless growth in the economy, or maybe a 10 year bull run in the financial markets makes people pause about the current environment. Here are 3 examples: (1) an interview with former Facebook exec and Social Capital billionaire Chamath Palihapitiya, (2) a video by media entrepreneur and centa-millionaire Gary Vaynerchuk, and (3) Goldman's crypto investment Circle. Let's start with the personalities.

Chamath talks about his increasing disillusionment with the VC world, pointing to the current model: encouraging founders to take highly niche bets, then overlevering them with equity capital, forcing unprofitable growth in order to create the narrative of growth, and leading to wild paper gains that subsidize the success of the money manager. Looking at the fintech IPO market, the fall from grace of the American digital lenders comes to mind. Or the private valuations of Robinhood or Acorns. His advice instead is to grow slow but consistently, instead of trying to grow fast.

Gary is not as down on the venture industry, but he does discuss why he is building a media agency rather than a tech company. VaynerMedia runs at several hundred million of revenue, which would be valued much higher at a tech multiple. But his plan is to capture market share in the coming downturn in the cashflow business, and then use that cash to go on an acquisition spree for assets that are now private and over-priced, but will desperately need cash and exits in a drought. A tech garage sale if you will.

Which brings us finally to Circle. We think that historically Fintech has been pretty disadvantaged: cash-cow incumbents were incented not to innovate, disruptors had no cashflow and were highly targeted bets at remaking particular products, mostly pivoting into partnerships. Non-incumbents could not afford to go full stack and remake the industry. Crypto has changed that by making some businesses -- like OTC trading, derivatives, exchanges, media -- incredible cash-cows with billion dollar revenue lines. These revenues won't stick, because they rely either on inefficiently high prices, or unusually high demand spikes. But the risk assets that smart operators -- like Circle and Coinbase -- buy with that cash, really matter. Circle has just bought SeedInvest, an equity crowdfunding platform that it will mash up with its other acquisition, crypto exchange Poloniex, to beef up its STO prospects. Cash is king! Maybe that's why Circle is also pushing that stablecoin with Coinbase. 

b03d7232-cec9-4ea5-97b6-a52b8818a5b8[1].png
1bccf62a-1da9-48b2-ae86-02e129111147[1].png

Source: Youtube (Chamath PalihapitiyaGary Vaynerchuk), Bloomberg (Circle), Websites (Circle, SeedInvest)

REGULATION: It's opposite day in the United States: protect the banks, not the innovators

Here's an odd one. FDIC Chairman Jelena McWilliams attended the American Banker's Association conference and focused on how to simplify regulatory supervision in order to help banks compete with Fintechs. In a similar vein, US State regulators continue the legal fight against the OCC, a federal agency trying to allow tech companies -- mostly lenders and payments companies like Square and SoFi -- to have a special Fintech charter. Part of this grind is the alphabet soup of American regulators and inevitable conflict over jurisdiction. As an example, the SEC just launched a new hub for innovation and financial technology, much in response to the rise in digital assets. Still, a meaningful portion of the American regulatory apparatus is functioning to protect its banking coral reef from competition.

When you look at the spirit of regulation in Europe, much of its mission is actually to increase competition with banks, helping Fintechs and other innovative players take market share from incumbent national champions. PSD2, the major directive in this regard, is colloquially referred to as "Open Banking" -- quite the different mindset. The desired strategic outcome is that many incumbents will be low cost capital-providing utilities, and some players will be platforms or aggregators of tech, capital and user attention. Tech companies can back into these positions as well. If regulators instead protect the capital providers and hurt competition as a result, you will end up with disconnected tech infrastructure and a 20th century financial product-push economy. 

c2b03180-00c2-4f26-97a7-525e43d33f58[1].png

Source: MarketWatch (Cards), CNBC (Amazon Lending), Autonomous NEXT (Travelers), Statista, Amazon

PAYMENTS: American Express and Amazon help Amazon sell more products.

Ok, yes, we just talked about how the Amazon / Travelers partnership is primarily a way for Amazon to sell more of its IoT device and play kingmaker. But listen -- this is another great symptom that highlights why Amazon's entry into financial services isn't a threat to financial companies. It's a threat to e-commerce, the actual target of the platform. So in this example, American Express has partnered with Amazon to provide a credit card targeted at small businesses. It's a clever product which allows the small business to either (1) get cash back on purchases or (2) defer the interest on their card on that purchase. It's up to the small business, which may need the extra credit for a late-paying customer in one case, and the rewards the next.

Here's the magic. The cash back is 3% on all Amazon purchases (1-2% elsewhere), which means more shopping on the platform. But wait, there's more! If the card holder is an Amazon Prime customer, which is not a hard feat, they get 5% back. Similarly, the interest-free period is 60 days for regular holders, and 90 days for Prime holders. What this card does is make Amazon Prime shopping irresistible for a small business -- while driving Amazon's key metrics of Prime subscribers and retail volume. Sure, it's nice for AmEx. But all they get to do is sell a financial product that would apply in some retail channel anyway. Amazon gets to shift the flow of retail into its walled garden, and then monetize a sticky business customer over and over again! The cross-sell is bigger than the sell.

This is the monopoly moat of a platform, like Apple negotiating the record labels out of existence with the iTunes store by holding all the customers. Not only does Apple get the share of the music revenues, but it gets to sell all the iPods. Further, Amazon has done a remarkable job of handing out a financial feature to each big bank. JP Morgan has checking accounts, Bank of America has merchant lending, and so on. This distribution of seats at the table to the top financial incumbents is predictable -- both by power laws from the bank point of view, and by the stability of the capital base from the platform's view. At the same time, the net effect is that all these financial firms should want Amazon's share of commerce to keep increasing. 

473e8ce4-152f-4339-8d64-b48e6009f857[1].png
5999251c-212c-4e19-968c-e663ceeade97[1].png

Source: MarketWatch (Cards), CNBC (Amazon Lending), Autonomous NEXT (Travelers), Statista, Amazon

WEALTH MANAGEMENT: Fidelity launches Crypto custody, has won this game before

Everyone knows: Fidelity has made its move into crypto custody. The firm has been toying with an offering into the space since starting to mine Bitcoin since 2015. The product itself is exactly what institutional investors, i.e., fund manufacturers, have been complaining about over the last year: (1) a custody platform, akin to Coinbase / Xapo / Bitgo / Kingdom Trust, and (2) an order routing system that creates best execution across exchanges, independent versions of which also exist (e.g., XTRD). But to package it and make it accessible for the traditional financial services industry is a massive leap for the asset class.

Here's what many people don't know. Fidelity is one of the top 4 investment advisor custodians in the United States -- including BNY Mellon Pershing, Schwab, and TD Ameritrade. Together, these firms control about $2 trillion in advisor assets, with another $1 trillion in independent RIA assets sitting on smaller players or self-clearing firms like LPL. These custodians know (1) how to service a long tail of small independent money managers, (2) throw annual conferences attended by thousands of people to look at investment products, (3) enable hundreds of wealth tech companies to sit on top of their core services, and (4) deliver performance reporting and other tools helping regular people access their assets. That is not something any of the crypto players come close to doing or understanding.

To moderate our excitement, we highlight that serving a manufacturer (i.e, a crypto fund) is not the same as serving a distributor (i.e., a financial advisor). Still, we believe the software is transferable to some extent, and the entire world of digital wealth management awaits open APIs here. Second, we think best execution will be a real boon to the space, unbundling what an exchange should do from what a broker should do. If regulators like the NY Attorney General continue to find bundling and conflicts of interest offensive, some US firms will have to be broken up into component parts and spun off. Not Fidelity -- which will benefit from being impartial. And further, with enough volume and a good routing system, perhaps arbitrage bots and crypto market manipulation may start to fade out of the ecosystem. Fidelity's entry -- though long time in the making -- is a clear win for crypto.

c884afc6-567c-4b4c-b6cd-d0ec7ff0c165[1].png
d3764b93-5f74-4ff2-88d3-62d764644721[1].png

Source: BloombergCNBC, Fidelity (RIAsCrypto)

CRYPTO: Surprises from the Dallas Digital Assets Strategies Conference

We chaired a unique event in Dallas this week, and a few key takeaways are worth mentioning. First, something that really stuck out was the audience itself. We informally surveyed about 150 attendees, of whom 50% were financial advisors allocating assets for retail and HNW investors. Further, 70% of the audience owned Bitcoin, 50% owned ETH, and about 15% participated in an ICO directly. Two people, not including Lex, had the unfortunate pleasure of buying Crypto Kitties. The largest financial advisor in the United States, Ric Edelman, who runs $200 billion across 85,000 clients, stayed with us for the full agenda. Just a year ago, the overlap between the wealth management and crypto communities would be a null set. 

On the fund manager side, we had Tuur Demeester from Adamant, Sean Keegan of Digital Asset Strategies, Kyle Samani of Multicoin, Mat Hougan of Bitwise, and Bart Stephens of Blockchain Capital. We were impressed by the very variety of investment strategies on display. For example, Tuur primarily runs a Bitcoin investment strategy, using leverage on/off BTC to amplify alpha.  Similarly, the custodian Xapo only custodies BTC, backed by a reserve of coins -- $10 million worth bought in 2014. Others run index funds -- with Bitwise creating passive indexes and Digital Asset Strategies trying to deliver smart beta on the same baskets. And of course, Multicoin and Blockchain Capital both take fundamental venture-style bets on direct projects. We were reminded again of the BCAP token offering, a security token that Blockchain Capital launched as a unit in its fund. 

We can't do justice to all the conversations (i.e., custody, regulation, markets), but another one that stuck out for us was an asset allocator panel. Paul Pagnato of PagnatoKarp, a wealth advisor to large family offices, sees crypto living inside the venture capital allocation slice. James McDonald of Vishnu Wealth Management talked about building a 10-15 coin basket with the largest liquidity, while protecting for downside exposure. And we'll end on the perspective of Tyrone Ross Jr., who never wanted to put his clients into crypto assets. Instead, his clients just started disclosing to him their over-exposure to digital holdings -- so he had to design hedges and diversification strategies that would balance out the idiosyncratic risk. He also had to start reading white papers and websites to figure out what his clients were talking about. Advisors with such an appetite will retain their clients relationships, while those like Noriel Roubini will be drowned out by the winds of time.

  Source: Tyrone's     Video    , DASS     Twitter    , Lou Kerner on     Nouriel Roubini    .

Source: Tyrone's Video, DASS Twitter, Lou Kerner on Nouriel Roubini.

AUGMENTED REALITY: More productive enterprise use AR for field work

Interesting study for an enterprise usecase of augmented reality -- productivity in the field service management industry. Think about telecom technicians installing cable, mobile doctors or nurses in healthcare, engineers deployed to sites in industry, or other service providers that travel to location. The study found that companies deploying AR solutions -- like (1) remote video and analysis of the site by experts, with work carried out by lower skilled labor, or (2) AR-based on site training modules -- saw better customer retention, satisfaction and operating metrics. Imagine coming to a complex piece of industrial equipment, and seeing three dimensional notes and explanations on how things should work.

There is a potential parallel to financial services front offices. Much of the framework-setting in finance falls to a centralized function, whether that's a Chief Investment Officer creating portfolios or a Chief Data Scientist creating a loan underwriting model. Yet physical presence and emotional resonance still matter. Perhaps the human relationship is managed in person, but relevant financial stats and metrics about the relationship are physically tied or annotated to particular geo locations, or to the emotions of the customer. Clients are less and less likely to come to a branch or advisor office, spending time instead in their phones. But can the front office come to them, with a CIO in their AR glasses? Or, could machine vision be used to assess client net worth? If Truata can scan a windshield to figure out insurance damages, could a machine estimate car or house value in real-time net worth?

While certainly not an exact match, we can see the future in some analogous applications. Below you'll see concepts of some of the apps being developed for headsets like the Magic Leap. For example, AT&T is bringing DirecTV to AR, with up to 4 different channels streaming at once, which is how communication with a colleague can happen. Or, it could be done through a render, like we see in Avatar Chat -- remember that VRChat has millions of users. As a last point, Adobe has entered the game with a professional creative suite in augmented reality, bringing what is currently a novelty toy one step closer to being a platform.

7483f991-7ba9-40f3-a1b8-ffc99f295b8a[2].jpeg
bd129061-ce53-4f0d-8e8d-daa065a32dc0[1].png
e91c540f-dc31-4bba-931a-da5d27f2f47e[1].jpg
aee5cd87-fb4a-47b6-ab7c-9fe4568caae3[1].jpg

Source: Aberdeen Study; Techcrunch (Adobe), Upload VR (Avatar Chat), Variety (AT&T)

CRYPTO: Did ICOs raise $300 million or $2 billion in September? Depends who you ask.

Last week we published some grim ICO figures, which made the rounds in the media, suggesting that token offerings are 90% down on a monthly basis relative to the peak. We were challenged on our figures based on two sources: Elementus and Coinschedule (ICO Rating is another great reference). While our number floated down to $300 million, some of the others saw September as $1B+. As an aside, we want to show the largest number possible to frame the best story for a delicate and growing space. This is why we began adding venture capital equity investment into crypto companies. When looking at that particular chart, our trend is at over $1 billion in August. So let's explore the delta.

First, there are some chunky and problematic figures which we chose to treat differently. For example, CoinSchedule lists Rubi-X as a $1.2B ICO entry for the month of September, which we have not been able to verify otherwise and chose to exclude. The Venezuelan Petro ($700MM+) we also ignore, as it is at best a government-backed monetary unit, and at worst an experiment in sovereign fraud. Second, there are various timing differences. Take the $134MM into tZero, which we had already accounted for at announcement in July. Lastly, taking a closer look, many of the ICOs we chose not to include are self-reporting a "completed" ICO, and then a data spider is taking their softcap as the amount raised. We generally exclude data where the confirmation of a meaningful raise ($10MM+) is hard to pin down.

Further, Elementus tracks monthly flows as they happen. This means if an ongoing ICO is 40% through its time period, they will have counted accrued fund flow. We track data at period end, meaning that only closed offerings are counted. Such an approach will not give credit for capital in flight, and perhaps there are good months ahead if indeed flows are strong. But this methodology difference should generally wash out, unless large chunky and unusual things are happening (e.g., Telegram and EOS). 

Our final point on this is to revisit our data sources. We focus on analysis, and leverage other primary sources for much of the underlying gathering, which we then scrub. In looking at the process, we counted over 30 ICO trackers used in our aggregation process. The kicker though is the short half life of the sources -- charted below. We find ourselves swinging between various sites and their increasing and decreasing data quality! So if you ever think there's something we should pay attention to, do let us know. And without further ado, here are the adjusted figures, telling the same story as before.

84d1a718-d78b-4bd8-a684-831b0c1b77fa[1].png
918280df-100d-4f27-a0fb-1c91d562fda4[1].png
19feeb78-6581-40bb-a43e-956c09af9885[1].png
2dcc0052-c813-4cc9-81d5-9ccca37bd7d2[1].png
47529b13-b0c7-41f3-9d18-80e49ddf8af5[1].png

Source: Cryptoglobe (ICO data reporting), ElementusCoinSchedule

ARTIFICIAL INTELLIGENCE: Explaining Black Box Algorithms to Avoid Discrimination

Speaking of Amazon, news broke that the company had built out an AI-based recruiting tool that was supposed to help it rank candidates at scale. They certainly are not the only ones -- the tech startup space is littered with applicant management and analysis software, especially given that employees have many more jobs on average than in prior decades. What this AI did, however, was systematically discriminate against women, down-weighting resumes that included the phrase "women's" in descriptions or candidates that came from all female colleges. This result came unintentionally from the underlying data. If you correlate the language in thousands of employee resumes, you will get the status quo, which is that on average the Amazon employee, or any tech employee, is more likely to be male. Another artifact that mattered is the way candidates used language itself, which can be gendered in output.

Other examples of unethical AI are plenty. For example, image recognition algorithms make an error of 3% for white male faces but 30% on black females faces. Or, when used in automating sentencing criminals in the US, algorithms punish minorities more harshly. Or, when underwriting credit, AI disfavors historically disadvantaged protected classes using Zipcode. But the math isn't wrong -- it is in fact painfully correct. These outcomes are a mirror to how things are, not a solution for how we want things to be. Yet AI will be used regardless. Just this week, Lloyds adopted speech to text passwords for telephone banking, replacing pins with the sound of a customer's voice. Will this service work better for majorities and not minorities? Further, such security can be gamed using pre-recordings, or generated voices. Similarly, image recognition can be gamed with photos or by twins.

This is why we are excited to see two initiatives make the news. The first comes from the MIT Lincoln Lab, focused on machine vision. The software builds a visualization based on how a neural network sees an object, highlighting which parts and features of the object drive a particular decision. The picture below shows how the computer detects "large metal cylinders", first looking for size, then for materials and finally for shape -- each  highlighted by importance-ranking heatmaps. The second comes from IBM, called the Trust and Transparency service. In an example around insurance claims automation, the company shows an explanatory overlay on the AI that points out the probabilistic weightings for different drivers of an approval/rejection decision. A human analyst can then understand why the machine made its judgment. We think such tools will be required for any serious AI company.

b2f8d655-c7ea-442c-90df-5bfc871ee4a7[1].png
ff92aa6a-9ef6-401b-9be8-0c109374cb03[1].png
f3a0f7f4-6446-4a01-93b4-5747f9d21af8[1].png

Source: Reuters (Amazon), IBMMIT Media Lab, Business Insider (Amazon), FS Tech (Lloyds)

INSURANCE: Monopoly gains to Amazon's platform from Travelers Insurance channel

Travelers, the home insurer, has partnered with Amazon to sell smart home and security devices. The company is getting its own digital storefront (amazon.com/Travelers) on the Amazon site, where channel customers can get SmartThings water sensors and motion detectors, Wyze cameras, as well as Amazon's Echo Dot. For Amazon, this is a proprietary hardware and marketplace sale. For Travelers, it is a home insurance sale, bundled with the telematics. Additionally, Travelers has integrated two skills into Amazon Alexa, rationalizing to some extent why you need all this technology to interact with your insurance policy.

This is a powerful symptom. On its face, it may look merely like a new marketing channel for a web-first demographic with a few gimmicks thrown in. Couldn't Walmart, Overstock, and the rest launch some product pages and cross-sell financial products? Here's the distinction: Amazon is a marketplace platform, whose value increases if it can grow two sides of its network: (1) manufacturers of stuff, and (2) retail customers. The manufacturers could make financial or physical objects, which don't matter. In order to win the platform game over traditional retailers, Amazon can throw in bleeding edge tech for free (or at cost). Walmart makes no phones, tablets or Artificial Intelligence-based assistants. Amazon does, and it has Big Tech leverage over all the aspiring startups in the space that want its consumer pipe.

Relative to other Internet companies, Amazon has the luxury of being post search intent. The Web is not a free-market endless bazaar, but a few walled gardens with monopoly-like attention ecosystems. Google sits in the pre-intent part of the funnel. People search "home insurance" into the box and get third party websites formatted according to their own logic. These results are driven by two markets: (1) bidding against keywords and (2) optimizing search engine results against a global, non-discriminating algorithm. Amazon is fundamentally different -- a king-maker that can select who wins business within its platform, and which has no need for an open web for Prime customers. This means insurance companies should race to claim their own custom channels on Amazon's version of the web (i.e., Amazon On Line?), which incidentally ends up selling Amazon hardware. This leads to a dynamic similar to that which Apple had on the music labels with iTunes and the iPhone. No competitors in sight.

a074aea4-742b-4d8c-9e80-e0646ed25d7c[1].png
2951abc4-b832-490b-a1b1-0136a762d037[1].png

Source: Company Websites (AmazonTravelers), Media (DigInWSJ)

CRYPTO: September ICOs 90% Down from January, but Venture Funding is Ray of Hope.

4f63d28d-1dc7-425d-8a95-bf9b457ea6a6[3].png
  Source: Autonomous NEXT, Pitchbook Data, China     Microlenders

Source: Autonomous NEXT, Pitchbook Data, China Microlenders

We're really trying to make this look good! But it's not working. We've scrubbed token offering data from September, and the trend continues generally to be down. Last month saw about $300 million in ICO funds raised, with the month before that revised to a bit over $400 million, a far cry from the $2.4 billion in January of this year. If we include EOS and other chunky private token raises, the highs go to over $3 billion, suggesting that monthly ICO activity is down 90%, which of course looks a lot like Ether's price performance, but with a 3-month lag.

There are three narratives at play, which are worth exploring. First, perhaps investors have devalued the idea of buying a utility token (does nothing yet, legally non-binding), and instead want to buy equity in the same companies. To test this, we looked at Pitchbook's data on blockchain and Bitcoin venture capital raises, which you can see in the second chart below in the magenta color. There is indeed a lagged effect in venture as well, with increasing drips of capital, reaching over $1 billion in August 2018. Why is that? Two reasons: (1) fintech companies like Robinhood and Revolut pivoting into crypto and (2) Bitmain trying to vacuum up capital before the public offering. This gives us a slightly more balanced view of funding in the space -- with recent months seeing a decline in public crowdfunding, but an increase in private checks. Anecdotally, projects are selling equity and giving matching tokens for "free" to investors in the capital structure.

The second narrative is Security Token Offerings (STOs). We know many different platforms working on this space -- from Templum, to Tokeny, to Sharespost, to Indiegogo, to tZero. And while we'd love to plot STOs on this chart as well to offset the decline, truth is that STOs won't hit the market in earnest for another half-year at least due to regulatory indigestion. We tried to find that extra monthly billion in STO land, but it's not there yet. And last, we're testing a narrative about the collapse/crisis in Chinese P2P lending since 2015, and whether that risk-seeking capital wound up in ICOs. If you've got any hints on that last one about Asia, let us know!

ARTIFICIAL INTELLIGENCE: Paying by Smile with Alibaba or by Blinking with Ping An

c8cfca7d-6522-4213-90ee-01c0c27fcb3d[1].png

Chinese commerce is very digital already, far outpacing the US in both nominal and percentage terms. Since almost no mobile payments in 2011, China now sees almost 100 trillion yuan, or $14 trillion USD, in mobile payment transaction volume. This compares to less than $100 billion in the United States -- a 10x difference in adoption of using phones, rather than cards or cash, to pay for things. Further, unlike in the West, the vector of payments intersects much more closely with social identity and networking, which is the platform globally for developing artificial intelligence. Just check your Facebook Newsfeed.

So we give to you implementations of AI for payments in the East. The first is from Alibaba. If the customer has Alipay's app and has enabled facial recognition, a smart vending machine is able to scan your face and associate it with the payment account. We would guess that there is a geolocation element involved as well for two factor authentication, or perhaps just a phone or pin verification. The second example is the newly launched Ping An partnership with Danyang Rural Commercial Bank. The plan is to use facial recognition combined with "blink detection" to authorize a payment. The Bank claims to target 1,000 merchants for the initial pilot of the program. Reminder -- Ping An has built out machine vision capabilities to cut down on time processing insurance claims, and here it is trying to rent it out as a cloud service to other providers.

We end with a few questions. First, if Ping An was able to stand up real machine vision capabilities within a couple of years, what's stopping Visa or Mastercard or JP Morgan from building the same? Why have American finance firms failed to own the AI technology layer and its associated cloud? We think the answer has to do with the role of enterprise tech firms and implementation consultants in the US, which make the default option to out-source rather than in-source such capability. Why build, when you get this from Google for free as part of a cloud deployment? And second, we observe that massive data processing and hosting infrastructure is needed to accurately process image recognition on video for millions of people in real time. Likely, you also need high definition images to pick up blinks and smiles. So let's refresh that 5G network!

cd0fc4bd-4bb9-4a36-a86b-106b613549e3[1].png
5d254a84-b240-4a28-a8ef-dce0451d336a[1].jpg

Source: Walk the Chat (Charts), Fung Global Retail & Tech (Chart), TechCrunch (Alibaba), MPayPass via CrowdFundInsider (Ping An)

AUGMENTED REALITY: Government and Military Use Will Drive Magic Leap, HoloLens Adoption.

Last week, we spent a bunch of time talking about how consumer VR as a standalone platform is not turning out to be as good as iTunes, the iPhone, YouTube or the Web. One problem was the form factor, another problem was the lack of pirated content -- though games and adult content will slowly address this. This week, we want to point to IoT (Internet of Things) and Augmented Reality (AR). Do these themes have a reason for being and are they an opportunity for a major retooling of our interaction with technology? Here, we think the answer is a stronger Yes. But this is due to a surprising reason -- government and military use.

The Web was popularized through consumer use and now powers our digital selves. But it was brought to life and initial use as ARPANET in the 1960s through funding by the US Department of Defense. Imagine unlimited funding with life and death use cases by a nationally-embedded client base. This is also what the Chinese government is doing in relation to AI, blockchain and quantum computing, and get to the meat. First, Bloomberg reported that AR companies Magic Leap and Microsoft's HoloLens are bidding on a $500 million augmented reality Army project. The order is for 100,000 headsets which would run the Integrated Visual Augmentation System, overlaying intelligence on the physical world. These would be used for both training as well as in live combat. The manufacture of these types of devices would create an economic base on which consumer versions could be created, as well as condition a whole generation that using AR headsets is normal.

Another data point supporting this idea is the investment by local government entities (e.g., UK councils) in digital twins of their neighborhoods for urban planning. In particular, Liverpool is running a £3.5 million IoT program that combines the rollout of a 5G network with innovative health and social care services for residents. Of the 11 proofs of concept in place, examples include video connection between vulnerable people at home and their pharmacy, AR maps that bridge physical distance and combat social isolation, and sensors that monitor whether older adults are dehydrated. Similarly, earlier this year, Bournemouth was mapped into 3D, incorporating 30 different data sets, also as part of planning the 5G network. These live 3D maps, which could then be projected into the real world via AR devices, are a social good and should be part of centralized infrastructure. This in turn can further move the needle in consumer adoption and market maturity.

0e00e9e2-6a1b-45c4-a23f-ac332da0d5dc[1].png
c3690017-d12b-4b95-b3e9-57a42b60a336[1].jpg

Source: Magic Leap (BloombergNext RealityDaily Mail), UK Authority (LiverpoolBournemouth), Wikipedia (ARPANET

ONLINE BANK: Santander, DBS Getting Mobile First Right

You can split the last decade of Fintech into (1) unbundling of banking and investments into niche financial apps, and (2) the re-bundling of these apps into cross-selling machines once some amount of scale has been reached. See N26 (a bank with investments), SoFi (a digital lender with wealth and insurance), Acorns (a micro-investing app with a debt card). But direct to consumer startups are not the only ones getting it right. Today, Goldman with Marcus, JPMorgan with Finn and YouInvest, DBS, BBVA, Santander and the Nordic banks all have smart digital strategies that copy (or buy) the Fintech playbook.

The issue is that digitization is, to some extent, discrete. If you bring financial products into the 21st century, the remaining field for competition is audience gathering. So we note that Santander's Openbank has launched the following in addition to its neobank -- (1) micro-investing that saves a small amount per time period, (2) goal based planning and investing, and (3) a roboadvisor powered by a BlackRock investment team priced at 55 bps (unclear if this is FutureAdvisor, or just a BlackRock asset allocation model). On top of this, the app will have a password manager (surely people trust banks with their passwords, said no one ever), and a data aggregation service like Mint.com. For what it's worth, the app has a 3 star rating on the App Store, so perhaps people don't love it like they love Revolut or Monzo. But the Spanish bank claims to already have 1.35 million accounts, and is coming to the UK next.

Another forward thinking offering is Singapore's DBS. Like Openbank it has all the latest Fintech features, as well as a Facebook Messenger integrated chatbot. As a comparison, independent chatbot Cleo now supports 500,000 users. But of course, it is going to be much harder for Cleo to monetize, well anything, without manufacturing some sort of financial product as people don't pay for information alone. In the case of DBS, the tech-first approach has allowed them to double the revenue generation off a digital vs. a traditional customer (S$1,300 vs. 600), and decrease meaningfully the servicing cost (S$468 vs. 348), leading to a far better lifetime value. Can an independent fintech build that scale and product set as well?

4dc90b58-7595-4369-b640-e0bb0ae4fb78[1].png
d90c7f90-8aa6-4216-8504-62e1ff288933[1].png
c9bd331a-a22d-42ca-acce-2f2fb83b8b2b[1].gif
650625ad-a8e1-40fb-bd5f-a744f3bb474f[1].png

Source: FSTech (Santander), Coverager (CLEO), Bloomberg (DBS)

VIRTUAL REALITY: VR Failing as Platform as Headset Shipments Stall.

Virtual Reality isn't working out as a platform. Below, you'll see the declining popularity scores of the major headsets being sold on Amazon, as scraped by Thinknum. Samsung's Gear headset slipped from the top 100 after November 2016, Facebook's Oculus by July 2018, HTC Vive in June of 2018 and Sony Playstation in April of 2018. IDC also just updated their headset tracker and shipments are down across device types. Headsets that insert smartphones (e.g., Samsung, or Google Cardboard) have declined from 1 million in 2Q 2017 to less than 500,000 a year later. Tethered headsets (those that connect to PCs and devices) fell 37% year over year.

The only category that did grow is standalone headsets, which is a completely independent device for which you need no other peripherals. Oculus Go and Xiaomi Mi VR are examples. And while there might be several dozen million devices installed out there, the technology is not showing the hardware growth of the iPhone, or the software growth of a Whatsapp. So what can we take away from this stalling? First, VR is meant to be a platform war for tech companies, including a dedicated app store, video and gaming content, and the other benefits of owning the customer. However, to be a platform, you need to capture an audience. And to capture an audience, you need to have spectacular differentiated design and proprietary content which forces adoption. The design part is still in wild flux, looking for a form factor people like. 

On content -- today's platforms bootstrapped themselves off existing media. Youtube started out as a place to park bootleg movies. iTunes grew on being able to play pirated music and sync it to your iPod. Facebook gave a home to college kids' digital photos. VR does not have this luxury, because the making of VR content remains far outside the capabilities of the average user. Nor is there a large library of interesting things to digitize. So the bootstrapping is much more difficult, because not only do you have to create all the content from scratch, but you also have to teach your clients about a new experience, which suggests the audience is not built in. That's not to say that the payoff for winning VR is small, but in the short term it is primarily a novelty gaming system.

eebdd2ef-04de-4cb5-a41f-aae20a9f8590[1].png
4350d09d-4f0c-422e-b2a4-714fe994d822[1].png
2408f862-220f-4c78-80b1-4e533c5c17c3[1].png
84555252-92dc-45b5-9d66-273764d242d0[1].png

Source: Thinknum (VR ranks), IDC (Shipments)