Open Source as Revenue Driver: How Elasticsearch Uses Open Source for Revenue Growth, Tech Investing and Crypto, a Cautionary Tale of Bad Crypto Trading Psychology and Flash Crashes

On this episode Faizaan and I explore the Elastic IPO by doing a deep dive into its SEC Prospectus. We look at their business model and risks, and how open source software has been a giant revenue growth driver for them. We then get into an interesting discussion around tech investing and how it’s related to crypto. We look at some alerts that came through our platform: a cautionary tale of a trader who lost his life savings by investing at the peak and trying to recover losses with Bitcoin Cash and Ripple, and how to avoid doing that by looking at trading psychology. We also look at a flash crash in crypto and compare it with flash crashes in public markets. If you enjoy our podcast, please rate and review us on iTunes. It would help us out a lot. Thanks, and enjoy this one.

Here’s a transcript of QuantLayer Crypto Podcast #15.

Source: $ESTC’s SEC Prospectus Filling

The episode in its entirety can be listened to here:

QuantLayer is a software consultancy based in Brooklyn, New York. All opinions expressed by podcast guests are solely their own opinions and do not reflect the opinions of QuantLayer. The information presented should not be construed as investment advice. Guests may maintain positions and assets mentioned in the podcast.

Vikram: Hey, everyone you got Vikram here from QuantLayer, also joined by Faizaan known as the “wizard”. Hey Faizaan, how’s it going?

Faizaan: Good, good. How are you?

Vikram: Pretty good. Not too bad. So, might as well get right to it. The first topic we wanted to talk about today was the filing recently, the IPO filing recently by the company Elastic who ran the Elasticsearch product line. This isn’t crypto-specific but I think it’s pretty related, which we’ll come back to in a little bit.

Faizaan: Yeah, I think it’s relevant to tech and then definitely as you alluded to there will be some crypto tie-ins.

Vikram: Yeah. We’ll put the link to their prospectus in the show notes. This is what they call… an IPO prospectus is called an S1, and this is the first one they filed. They’ll file this and then the SEC has a chance to look it over and then comments and then they can respond to the comments and so forth. It can go on for a little while. But they file their S1 and probably they’ll end up going public in a few months. When you’re going through an S1, you know, again, the SEC is a treasure trove of great info and if you’re interested in any tech companies, interested in investing in tech companies, you know, it’s definitely the place to check out. You want to check out a company’s 10-K’s, their S-1’s, their 10-Q’s, there’s all kinds of filings you can look out to familiarize yourself with the space. But in the case of S-1’s, they’re usually a nice little graphic of the company and what they sell right upfront and with networking or hardware companies you might see things like routers or cables. With software companies, it’s a little harder because there’s nothing really to visualize. Their S-1, they just described themselves in a box and I’ll just read it here. They say “Speed, SCALE, Relevance: Elastic is a search company. We focus on value to users by producing fast results that operate at scale and are relevant. This is our DNA. We believe search is an experience. It is what defines us, binds us, and makes us unique.” I didn’t mean to like giggle at that but it just reminds me of like Obi Wan Kenobi’s description of the Force to Luke Skywalker. He said something like, you know, the Force is what gives a Jedi his power; it surrounds us and penetrates us, it binds the galaxy together.

Faizaan: That’s nice.

Vikram: Yeah. Anyway, this early part of this prospectus has a lot of great information you can review to get a basic idea what the company does, what their financials look like, what risks to their business are. Things like that. It’s called the Prospectus Summary and it’s a shorter part upfront in a prospectus. In the Elastic one, it’s about 18 pages. If you want to really get into the nitty gritty, you can ready further because it’s, you know, say 100 pages. There’s plenty more stuff to read about. But we’re going to stick to the Prospectus Summary here and so within that they have a, you know, after the little graphic describing themselves, alluding to the Star Wars, they have a description of themselves. I think it highlights, you know, what search is and why it’s important so I’ll just read it here. It also highlights like all the customers they have. This is the part that kind of surprised me. This is from their prospectus. “When you hail a ride home from work with Uber, Elastic helps power the systems that locate nearby drivers and riders. When you shop online at Walgreens, Elastic helps power finding the right products to add to your cart. When you look for a partner on Tinder, Elastic helps power the algorithms that guide you to a match. When you search across Adobe’s millions of assets, Elastic helps power finding the right photo, font, or color palette to complete your project. As Sprint operates its nationwide network of mobile subscribers, Elastic helps power the logging of billions of events per day and manage website performance issues and network outages.” And it goes on and on. It mentions SoftBank, Indiana University, and so forth. Then the final line, that little section is “All of this is search.” This is me again. They’re basically walking through a lot of their major customers and what their use case is around search art. Like I said, I was pretty surprised. I didn’t realize that they worked with Uber and Walgreens and Tinder. In particular, Uber and Tinder are interesting at least to me more so than Walgreens because of the type of solution that brings to them. If you think about UX and search, consumer applications like Uber and Tinder have pretty serious requirements, so they’re used regularly, they have geolocation requirements, complex matching algorithms in place and so on. Faizaan, what are some of the types of things that Uber and Tinder would care about that Elastic would offer them?

Faizaan: Yeah, that’s a good question because I think when you think of search, you just think traditionally like I type in words into a box and I get things that match and the most simple version of that is like if I search for the word ‘car’, give me things that have the word ‘car’ in it. Then it sort of gets more complicated from there where give me words that have those letters in it or similar letters and then you can start thinking about like the, you know, the actual meaning of what I’m searching. But what’s interesting with the two examples that you gave specifically is that it’s not language. The Uber example, you’re essentially doing some sort of geofencing where you know where everything is, you’re updating in approximately real time and then you’re able to search based on, you know, coordinates. Then the Tinder thing, again, there’s an element of probably geofencing but also an element of a recommendation engine that’s being added to the search. Both of those require more than just, you know, text and I think that’s a lot of what has made Elasticsearch so useful is that it’s gone beyond just like text or even just like geolocation.

Vikram: Yeah. That’s interesting because the user doesn’t know that they’re involved in the search, like their use case is not necessarily search. It’s help me get a car or help me find a date. It is search I guess in the broad scheme of conceptually what they’re looking for something, but it’s not anything like, you know, putting words into a box.

Faizaan: Yeah, and so it’s interesting because, yeah, like when you think of search, the words into a box that arguably actually doesn’t cover the majority of the use cases that they say they support. It’s really much broader. If you go to their site itself, you know, they say logging metrics, site search, security analysis, app search, analytics, and then there’s a few more items on that list. When you dive into that, what you realize is there’s a few pieces to that. The first is indexing the data correctly, like you can’t just store everything as text. Like they’re able to store locations as locations in a way that they can be created and compared quickly. They’re able to store like other data types so that they can again do aggregations on different sort of metrics. They provide logging as a service as well so you can again do aggregation on your logs very quickly. The way they store this stuff and then of course, there’s the search pieces that we’ve grabbed this data, we’ve indexed it the correct way with the correct types and now we can actually run the aggregations on it, and that’s where the real value for the product comes in but it’s all three layers that are pretty useful. What’s I think very powerful with Elasticsearch is like you can start small and scale up. If you want to build just a text search, like let’s say you have a very simple app, we’ll go, you know, rails and PostScripts, and you just want to do some text search on some fields, you can do within PostScripts. If you want to add some like GIS stuff for locations, you can add the extension and you can do that. But then as it starts getting more complicated, you’re going to start thinking about moving to another product, so now either due to the complexity of the type of search you’re doing or the, like, scale, you just have too much you don’t want to hammer your database, and what’s great about Elastic is that they’ve been able to essentially solve both of those problems. You can do relatively complicated stuff and you can also scale up while having started with a relatively simpler limitation.

Vikram: Yup. They get into that broader definition in their prospectus as well which I’ll read here. This is from them. “Today, what we search has grown to include a rapidly increasing amount of structured and unstructured data from a multitude of sources such as databases, websites, applications, and mobile and connected devices. While search experiences often begin with search boxes, they are not confined to them.” This is what we were talking about before. “Dragging your finger across a map on a smartphone screen is search. Zooming into a specific time frame in a histogram is search. Mining log files for errors is search. Forecasting storage capacity two weeks into the future is search. Using natural language processing to analyze user sentiment is search.” You know, they’re basing everything is search. What they’re saying here is they’re kind of expanding their definition of search to include things other than searching for text in websites. I guess, Faizaan, do you think this is fair like search and indexing obviously go hand in hand but you think what they’re saying here is fair or are they kind of just expanding their core market definition to justify like a higher valuation because they went public?

Faizaan: In general, I think it’s fair. If you look, you know, if you go through their website and a lot of what they’re saying, a lot of what they are selling is use cases and not like the underlying technology but I think that’s smart. But I think it’s entirely reasonable because essentially you’re using Elasticsearch for a lot of what is a bit too complicated… or not too complicated. Like, it’s not ideal to do within your relational database. But then if you again sit and like build a no sequel store and build your own aggregation layer and analysis and all on top of it, is going to be time-consuming, probably slower and less reliable. It goes back to the “don’t build what you can buy”.

Vikram: Yup.

Faizaan: They’ve probably broadened the definition a little bit but I think it’s reasonable.

Vikram: Yeah. Then they go into this other part of the Prospectus Summary which is another… this is a super interesting topic and it will tie into the crypto stuff. It’s about their open source business model. Here they go. “Our origins are rooted in open source, which facilitates rapid adoption of our software and enables efficient distribution of our technology. Developers can download our software directly from our website for use in development and production environments. Since January 1, 2013, our products have been downloaded over 350 million times. These downloads include both free and paid products. Open source also fosters our vibrant community of developers who help improve our products and build on top of them.” Then they provide some stats to that last point about the vibrant community. “As of July 31, 2018, our community included over 100,000 Meetup members across 194 Meetup groups in 46 countries.” If you ask me five years ago when I was trading full-time whether number of Meetup members was a useful metric for investing in a software company, I probably would have said ‘no, that’s stupid and you’re fired.’” It would have seemed like a pretty superfluous metric to me, much like the whole eyeballs thing during the dot com heyday, you know, it’s not a sales margin, pricing, earnings metric. It’s hard to use to model out a company’s performance with something like Meetup group members. But my view has changed now. I’ve done a 180 after h Iave been developing full-time. It’s definitely an important metric to consider when you’re investing in an open source company and it’s again not going to be, like the stock price is not going to be tied to the number of Meetup members but if you have a solid community and Meetups are one metric to look at that around, then it probably is a good indicator. Why is it so? They alluded to that in the following section. “Our business model is based on a combination of open source and proprietary software. Many features of our software can be used free of charge. Some are only available through paid subscriptions, which include access to specific proprietary features and also include support.” Then they continue. “Our sales and marketing efforts start with developers who have already adopted our software and then evolve to departmental decision-makers and senior executives who have broad purchasing power within their organizations.” Faizaan, what are your thoughts here? How do you think like things like Meetup members, growing Meetup groups around the world and things like that can affect an opensource project?

Faizaan: Yeah. I think they’re absolutely on the nose with that. I think there’s a subset of the industry particularly in like the space like rails and its offshoots, like node and elixir or whatnot, where that’s how a lot of developers choose the tech that they want to work in, find jobs, build professional circles, experiment with new technology like it’s really very meetup and community driven. It doesn’t have to be physical meetups. There’s a lot of activity on Slack, Discord, IRC Channels. But for opensource technologies, the community is absolutely essential because for anything to become relatively big, you need that involvement to grow the tech itself and also like building the community is the market that you’re eventually going to go sell into. You know, speaking for me personally, I can think of multiple examples where technology that maybe I didn’t end up using professionally but like I went and experimented with, was the result of going to a Meetup and seeing a talk on a given use case of a specific use of tech, or I can think of instances where there was actually a developer evangelist from the company that produces the technology there giving a demonstration that was compelling enough to get me to try their tech.

Vikram: Yeah, the whole developer evangelist thing, I didn’t buy at first to be honest.

Faizaan: Yeah, I thought it was a ridiculous thing and a ridiculous term and then you go see it and it’s not a, like fluff role.

Vikram: It’s not at all.

Faizaan: It’s usually a very, very competent technical person that understands their technology and its use cases like inside and out and can demonstrate it effectively. That’s been my experience.

Vikram: Yeah, and they can answer questions like okay, this seems interesting and I want to include it in my workflow so it does X, Y and Z and then they can just tell you. But I guess that’s their whole role, like they’re trying to facilitate the developer community, so it makes a ton of sense. They even point to this phenomena in their prospectus which is they have a section on growth strategies and it’s actually the first on the list. It’s “Increase product adoption by improving ease of use and growing our open source community.” I guess let’s drill down into this a little bit. Faizaan, when you see like an open source project and it becomes easier to use, like how might that translate into more revenue?

Faizaan: Okay, so first, there’s something like Oracle, IBM, whatever selling their enterprise licenses directly to generally larger companies. We’ll take those sorts of products out of the equation. Then we look at something that let’s say like we have a product we’re working on, we’re less likely to go to Oracle and pay $50,000 for a database license. We’re going to go look at what are open source solutions out there. The sorts of tools and technologies that these smaller dev teams, startups, and now a lot of bigger companies too are using are all really you’re seeing it’s a lot of open source that’s then backed by a for-profit company. I would say that like if I see something is open source, it has a lot of usage. It’s like the de facto choice for that specific technical problem or one of the options. That’s great. But if I’m going to make it a part of my core infrastructure like “oh, this is what I use for my database or for my core search or deployment,” it actually helps a lot to have a like for-profit company behind it because you know there’s a certain amount of momentum if, you know, if like Elastic for example is getting paid enough by a bunch of big companies to run their product, you know there’s a good amount of momentum behind the open source project as well.

Vikram: Yup.

Faizaan: It goes the other way as well where as they grow the community, some subset of that community is basically their like total addressable market. I can create an open source search tool and if no one is using it, there’s no one to sell cloud services or regular services too for that tool. But if you get a million people using it and some subset of those go on to become very successful startups like the examples you gave of your Uber or your Tinder, now you’ve just grown the market that you can sell product then too.

Vikram: Yup. That’s a pretty important point because like you said, if you end up with a future unicorn as a customer and you’re embedded into their architecture, that’s a pretty big win.

Faizaan: Yeah. You got a huge enterprise sale locked in that you just grew with when they were small.

Vikram: Is it an official definition of unicorn, I mean, you have a billion-dollar valuation? Is that…

Faizaan: Yeah, I believe so.

Vikram: Okay. But just because you have a unicorn as a customer doesn’t make you like a baby unicorn.

Faizaan: No. You’re just the… I don’t know if there’s a term for baby unicorn. I know there’s decacorn which is a $10 billion company. Apparently, unicorn is not cool enough anymore.

Vikram: Another interesting thing in their prospectus is this thing that they have called “Net Expansion Rate”. Like across the board of public companies, especially in tech, they all try to find these like unique metrics that are associated with them, so like, you know, margin isn’t necessarily like a unique metric but some companies will point and say like “Oh, our gross margin is higher than all our competitors and.” You know, just tweaking revenue a little bit means a ton of profitability and things like that. I think this is going to be a pretty important part of the Elastic story, this thing called Net Expansion Rate. They explain it here. “We believe that a useful indicator of our customers’ tendency to expand their usage of our products is our Net Expansion Rate, which measures expansion in existing customers’ annual subscriptions over a twelve-month period. Our Net Expansion Rate was 142% as of July 31, 2018 and over 130% at the end of each of our last seven fiscal quarters.” They define it more in-depth in their MD&A; that’s the Management’s Discussion and Analysis of Financial Condition & Result of Operations. If you want to do a deep dive in a company that’s just filed to go public, that’s the section you want to look at. It will just lay out like how the company’s business is, its risks, all the leavers that they have to make and lose money. Anyone that’s interested in investing in an IPO should read that part of their prospectus. It’s highly, highly educational. What do they say about this Net Expansion Rate in their MD&A? “The Net Expansion Rate includes the dollar-weighted value of our subscriptions that expand, renew, contract, or attrit. For instance, if each customer had a one-year subscription and renewed its subscription for the exact same amount, then the Net Expansion Rate would be 100%.” So, what is this? Net Expansion Rate, it’s basically the amount of customers increased their spend at Elastic. It’s an indicator that the Elastic is gaining share at a particular customer. Why is that important? The more and more entrenched they can be at a current customer, the more beneficial it is for them long-term. When we have use cases like the data, machine learning, artificial intelligence that are going to become a larger part of the equation for software companies, that will benefit them over time if they’re already embedded in like a lighter use case of their technology.

Yeah. They’re valuing their TAM (the total addressable market) as $45 billion and, you know, the last fiscal year ending, I think, yeah, this last fiscal year they’ve done 160 million. A natural play there is just to be able to sell more product into their current customer base for those verticals that they just aren’t part of yet.

Faizaan: Yeah. I think there’s an extra layer there. You said total addressable market, 45 billion, big data MLAI and then the 160 million that they’ve done. But I think, you know, going back to the community building, there’s the “Okay, we have $45 billion of potential spend,” but the first piece is just convincing people that Elasticsearch is the solution over some other similar competitors but you still have not monetized that group. The first is essentially turning them into users and then turning them into customers. With these like open source backed companies, it’s an interesting dynamic because you can get people using what you’ve put out and still not monetize them at all.

Vikram: Yeah. That’s why when someone is successful like this, you know, doubling revenue almost year on year and having a mostly open source software base, it’s pretty interesting. This Next Expansion Rate, I think it’s going to be a metric that analysts will care about on all the conference calls once they’re public. Analysts always care about like as I mentioned before every company has like a handful of metrics that really drive their growth margins, pricing, unit sales for like retail companies. These guys, I think, Net Expansion Rate will be up there along with revenue of course. They highlight this as a risk factor in their prospectus as well. This is what they say. A risk factor is basically a list of all the things that you and your lawyers are worried about with respect to your business, so it will be in a 10-K, it will be in an S-1 prospectus. It’s basically for the investors. It’s, you know, cover your ass for the company so the investors know, you know, what are all the possible things that could go wrong at the company. This is the risk factor that they have for this particular thing. “The markets for some of our products are new, unproven and evolving, and our future success depends on the growth and expansion of these markets and our ability to adapt and respond effectively to evolving markets.” In general, as far as their risk factors go, a lot of them are pretty boiler plate like we may not be able to compete successfully against current and future competitors or “If we are unable to increase sales of our subscriptions to new customers, sell additional subscriptions to our existing customers, or expand the value of our existing customers’ subscriptions, our future revenue and results of operations will be harmed.” I mean, no s***. If you don’t sell more product, your business is going to do poorly. To be honest I was pretty disappointed in their risk factors. A lot of companies have more specific ones but most of the ones they listed were pretty boiler plate, apart from that one that we just talked about. There’s also one more that’s pretty relevant to like the whole open source thing and I’ll read it here. “Because of the rights accorded to third parties under open source software licenses, there are limited technological barriers to entry into the markets in which we compete and it may be relatively easy for competitors, some of whom may have greater resources than we have, to enter our markets and compete with us.” What do you think, Faizaan?

Faizaan: I think this is the main risk factor for them because we talked about phase one is getting people to use your product and the second is monetizing them. If what you’re providing is like, you know, whether you want to call it like infrastructure as a service, platform as a service what-have-you, elastic specifically is competing against the likes of AWS. Now, you know, you can download the Elastic search source and run your own node; in production you probably want to be running some sort of cluster of at least 2 or 3 nodes which then there’s a lot of overhead to doing that, so then you start looking at using some sort of a cloud hosting service and you can go to Elastic, but because the product itself is open source, there’s nothing stopping AWS from providing the same thing and the risk there is that AWS has a lot of customers that do all of their other cloud services from AWS. It’s a lot easier for them to just add one more item to the bill and go set up a separate thing with Elastic outside of whatever virtual private cloud they’ve set up or whatnot. This one is interesting because I’m curious to see how they’ll compete. Now that being said, we have seen a few tactics that these companies will use. They may open source the main product but like especially for larger could deployments, there may be a modified version that they run that’s more performant, better at handling very large clusters, things like that that’s not publicly available. For customers that are most profitable, the ones that are going to be running these large clusters, they can offer something a little bit better than they can get from someone else that, you know, isn’t the one essentially that has the most contributors to the open source. That’s still a risk because if AWS sees a lot of opportunity, again the project is open source so they can add maintainers to the project. We saw this happen with a node in Joyent, so node is basically a Javascript on the server. It was this company Joyent that provided, you know, services and a lot of the core team for node. Essentially a lot of the open source developers didn’t like the direction Joyent was taking it. They forked node and then there was a big kerfuffle and then, you know, it ended up being merged back together and then there’s like a foundation that runs it. My understanding is that because it’s open source and enough of the people were not associated with Joyent they have less control of it than they did initially. There is that risk but in general, the company that did create the open source technology can run a modified version because they’re going to have a better understanding of it.

Vikram: But with respect to like continued development by Elastic, that’s going to be tougher for AWS to maintain, no?

Faizaan: Yeah. There’s the piece of it where like, you know, the bit that’s open source, everyone is going to keep getting access to it. If Elastic throws a bunch of money at open source maintainers whether they’re like internal staff or full-time external people, everyone is going to get whatever gets pushed in the master, but they can maintain internal versions that are much more performant for like very large deployments, and we’ve seen that happen, I would need to double check but I’m pretty sure ReTiS Lab does the same thing, so like the original creator of ReTiS works there and the version they use for like their cloud support is slightly optimized for that sort of thing. The other thing that they can do is more products. You know, I can spin up my own thing, there’s a lot of reasons why I shouldn’t. I can go in AWS, spin up an elastic search cluster, they’ll handle a lot like the dev ops portion of it. But then you can start looking at, you know, the use cases that we had mentioned earlier like logging or analytics and you see if you go to an Elastic site, Elasticsearch is only one of a number of products they offer where they have Logstash, they have Kibana, they have ECE that helps solve some of those problems, you know. You can build dashboards. So, you know, Elasticsearch provides the indexing and the underlying creating language but you can very quickly have a UI that gives you dashboards into the metrics that you might care about. You can very easily spin up logging. Then that makes it much more compelling to use their suite of services because they’re solving more of your like platform or application level concerns.

Vikram: Yeah, I think that’s what they’re saying to that point in their S-1. They say “Our business model is based on some open source, something proprietary and then some features are free of charge and then some can only be through paid subscriptions.”

Faizaan: Yeah. Yeah, we have seen that too with open source where you only get some features in the… and that seems to have worked pretty well. I think in the Ruby world, Sidekick which was a tool we used to use, there’s a free version that is like ubiquitous and then for some projects where we needed some more advanced features, I think it was in the thousands of dollars a year to have the enterprise version that actually was the code was slightly different. That can work as well.

Vikram: Yeah, so like the open source, the free open source features are like the gateway drug and then all the other stuff that adds so much more on top of it are the rest.

Faizaan: Yeah.

Vikram: We talked really high level but like what they’re up to and stuff, and then also it’s worth looking at their numbers because they’re pretty phenomenal. In 2017, this is our fiscal year, so in 2017 they did 88 million; and 15 million in license, 65 in subscription and the remainder in services. Same number in 2018 where 160 million in total revenue; 26 million in license, 124 million in subscription and the remainder in services. One interesting thing about services is they don’t make a lot of money. Last year they made 2 million on 8 million. This year they lost 2 million on 10.5 million revenue. It’s definitely not going to be a big driver of earnings growth by itself like it will probably help sell product. That’s the whole point.

Faizaan: Yeah. That’s probably another differentiator going back to like why would I choose Elastic over AWS. There may be some bias towards, you know, if I’m buying a service contract, I may be more trusting of the company that essentially created the technology over like a third party provider.

Vikram: Right. Total revenue growth, 81 to 82 percent. They’re still in the red. Operating expenses grew from 116 million to 167 million, so that’s about 44 percent growth, so there are still top lines growing faster than they’re burning capital which is good.

Faizaan: Yeah, and I think that’s pretty standard for other companies that have been in this position like Mongo and whatnot, if I’m not mistaken.

Vikram: Yeah, and you know, they’re probably going public to help with that burn but if they’re going to double year on year, you know, next year, I don’t know what like the model numbers look like but I wouldn’t be surprised if they say like next year’s estimate is going to be like 50 percent or 70 percent revenue growth. Actually, I don’t know, like I haven’t been in the cell site in a really long time but I remember like back when I was banking and went through a couple of road shows, there was a point where bankers were no longer allowed to give forward estimates for the companies that they were taking on the road show. Either the company couldn’t give forward estimates or the bankers couldn’t, like in their road show presentation. Like you can’t say specifically “my estimate for the company is going to be up 50 percent” because it’s not like public knowledge. It’s only knowledge for like the people who are involved in the road show. I don’t know if it’s insider info but it’s not fair to the rest of the market participants. People would always dance around it like I remember investors will just ask all kinds of questions to kind of get like an answer, “Oh is it going to grow as much as it did last year?” If it doesn’t grow… no, I’m serious. If it doesn’t grow as much as last year, how much could it grow by? It would just go… there’ll be a long dance like 30-minute long dance in a 45-minute long meeting to figure out what next year —

Faizaan: Warmer, colder.

Vikram: Yeah, exactly. But everyone’s part of the game. Everyone knows what game is being played. Yeah, with all of those numbers out of the way, we should probably talk about, you know, we’ve used Elasticsearch in a number of projects.

Faizaan: Yeah, it’s pretty much part of our like go-to stack at this point, I would say. You know, obviously we’re big fans of Elixir for our backend code unless there’s a specific reason to use something else, that’s what we like to start with because it just gives you a lot of stuff out-of-the box that lets you work fast and is fast and just gives you a lot. PostScript is our go-to database. Then for frontend, it’s you know, Ember or React. Then usually Elasticsearch is in the mix as well. How I like to structure stuff is I’m a big believer in like keeping your data like pure, so I don’t like to de-normalize stuff in my database to speed up queries, if avoidable. I really like to keep my database as like a source of truth that’s very easy to maintain and keep correct. Then any like optimized queries that need to happen or stuff like that, then I’ll use Elasticsearch as a place to store de-normalized data, index data in a way that can be queried quickly, do complex querying. It’s almost like the database is the source of truth and then Elasticsearch is there to facilitate any querying or denormalization of data, and that’s worked pretty well and it’s pretty quick to get up and running once you’ve done it a few times.

Vikram: At what point do you say, you know, PostScripts, you know, you’ve been great so far but we should really move this thing over to Elasticsearch? Like what are some scenarios where that might happen?

Faizaan: It’s pretty straightforward. If I see that okay, the way this app is structured, this table is going to get very big and we’re running it a lot, that’s usually one sign. If the query itself, like if there’s queries that require a lot of like joins or essentially if I’m starting to look at like, “Oh, this would be way faster if I de-normalize the data” rather than de-normalizing the data, at this point, again, I really like to keep the database in like a normalized state with lots of validations and just like pretty strict rules about what can get into the database and how it’s stored. The moment I’m thinking about de-normalization or I see myself making some very complex queries sort of compose multiple tables into one, that’s usually a sign that hey, maybe it’s time to bring in Elasticsearch or if the table is just getting very large. Because again, Elasticsearch, you can scale horizontally, PostScripts not so much without charting and a bunch of stuff that usually you don’t want to get into.

Vikram: Right.

Faizaan: You know, we had talked about hey, Elasticsearch is actually really useful for crypto and we looked at all the use cases like they talk about security metrics, analytics, stuff like that, essentially I think for us it’s going to be a great tool for doing analysis of on-chain data. Just what it gives you is going to be a great way to store and analyze and visualize on-chain data and events. Then the other thing that I think is just interesting about this whole thing is, you know, you can read like white papers and you see a lot of market activity but I think reading like TechX once and seeing how traditional tech companies have performed does give you some insight into how you should be thinking about a lot of like ICO markets and things like that if you’re not as familiar with like traditional markets.

Vikram: Yeah. I mean it’s basically, you know, IPO company, it’s funny to call it mature like when you call a company ‘mature’, we’re thinking like Microsoft, you know, GE and stuff. But they’re mature as far as, you know, they’ve been around for like 2012 I think, so about six years, so it’s still pretty young. But they’re doing 100 million plus in revenue, so it’s mature in the sense that it’s not like a startup of, you know, 1 person —

Faizaan: Yeah, if you look at revenue and their user base and just even the amount of like, you know, just the fact that what goes into making an S-1 versus like what most ICO’d companies are doing in terms of like regulatory, like disclosure.

Vikram: S-1 versus white paper, what do you trust?

Faizaan: Yeah. I think there is just a lot there to learn.

Vikram: I think that brings us to like a related topic around just broadly professional investors and technology companies and what brought it to mind was, you know, I was looking for like what other people were saying about Elasticsearch. Seeking Alpha, it’s this site, I can’t believe how much it’s grown. It’s grown so much. They cover all kinds of things. They cover stocks, they cover cryptos and commodities now too. There is a post that an author had written about Elasticsearch, like a first look at the IPO. I thought, “Okay, I’ll check this guy out.” This guy is pretty… it will be on the show notes, but is a pretty well-received person, has a bunch of, you know, 7000 followers on Seeking Alpha which is a pretty good number, so I’ll just see what he has to say. I’ll read the first part of his analysis. “Elastic mostly works on the back end, powering application interfaces without users ever realizing how their searches are being executed. As such, this is a search company that does not compete against Google’s consumer end, at least, not in the same way that Microsoft’s Bing (MSFT) does (though Google offers similar products on the enterprise side that do compete against Elastic). Unlike Google, Elastic doesn’t earn revenues from ad placements within its search engine. It’s not an internet company, but an enterprise software company, one that collects subscription revenues from its customers by licensing the use of its Elastic Stack to power applications.” Okay, this is me again. I find it to be like a really naïve explanation of Elasticsearch.

Faizaan: I think naïve is a very nice word to use for this.

Vikram: To say that it mostly works on the backend doesn’t compete with Bing the way Bing competes with Google and all that stuff, and then it doesn’t earn revenue from ad placements just is a woeful understatement and bad representation of what Elasticsearch does. Now, all these things are true but they miss the mark. It’s like saying that a marijuana company doesn’t compete with Anheuser-Busch the way MillerCoors does. I mean, you’re missing the whole point of how people get high.

Faizaan: Yeah. I would go even further. You know, I would use the example, let’s say we want to use transportation as an analogy. Elasticsearch essentially provides infrastructure or like some underlying stuff for the company that’s ultimately the end user company such as like a Tinder. I would say that it’s more like comparing like a tier 1 automotive supplier, like someone that makes the subassemblies for Ford versus Uber, so you know, in the comparison like it doesn’t compete against Google. It’s really like saying the company that makes car parts doesn’t compete with like Uber giving rides. That’s absolutely through like the whole revenue model is —

Vikram: That’s totally true.

Faizaan: Like what you’re ultimately providing, one is like you’re essentially a vendor for someone that provides some service to the end user and the other is a different end user service. But both happen to be search in the way that both of these happen to be transportation but it’s a s*** analogy.

Vikram: Yeah. In general I think there is like, I don’t know if it’s a growing disconnect but I do think there’s like a general disconnect among professional investors and what tech companies are doing. Do you think that could be the kind of thing that would get worse over time as tech becomes more and more specialized?

Faizaan: It’s hard for me to know if it’s getting worse or better. I mean, you know, you’ve been a tech investor longer than I have been interested in tech. Do you think that people’s understanding of tech now is better or worse than it was, say, when we were talking about like not web but stuff in the like…

Vikram: I would say our average understanding of tech is better, like on average if someone says it’s like the Uber of blah, like you can very quickly know what they’re talking about. The thing I don’t know yet is how specialized. What about the specialized knowledge? Like internet company versus enterprise software company, okay that’s fine. But then you can get even more specialized with like not just search but the stuff we were talking about earlier with like logging and the whole stack and all that. The thing I don’t know how it’s progressing and that’s why I’m just worried about is the more specialized tech gets, like will it just be less and less people who understand it? Like the same guy has written a ton of other great articles. It’s just his understanding of this particular space is not good. I just wonder if that is going to be the new norm.

Faizaan: Yeah.

Vikram: Because what Elastic does is pretty specific, right?

Faizaan: Yeah. No, absolutely. Like it’s really hard to understand what they do if you’re a non-developer. Honestly even if you’re a developer and you’ve never used it, a lot of the value is in the nuance. My view towards this and I think I used to feel much more strongly that like if you’re not technical, you’re going to have a dangerous misunderstanding of like how to value a tech company. I don’t feel as strongly that way anymore on the, like more mature end of the spectrum, like when we’re talking about public companies. What I’d say is if you’re nontechnical and we’re talking about when Elastic had let’s say 5 million in revenue and you’re comparing it to some other companies, if you’re nontechnical it’s really, really, really going to be hard for you to judge like which one of these is going to be a winner. I think like you have to understand that nuance. I don’t know if that’s as important at 200 million in revenue because at that point, they’ve proven their thesis in terms of like we think that this implementation of this technology solves this problem best in a way that like we can capture our market share. They’ve proven that. Then maybe their valuation really has more to do with things like how good they are at selling or doing other things that they need to grow as a technology company. They’re not as directly tied to like the nuance of the underlying technology, if that makes sense.

Vikram: Yeah.

Faizaan: Because there’s a lot of bigger tech companies that sell shitty products and seem to do pretty well.

Vikram: Yeah. I mean, they are pretty small. Like 160 million is still pretty small, right? Versus like Microsoft, Google of course, and the rest of the crew. Maybe they’re a little more leveraged towards tech, I’m not sure.

Faizaan: Yeah. But I guess my point being like when they were very small, you really would just assess like is this a useful thing. Now, the question becomes more around like can they compete against AWS offering the same similar thing and that has less to do with their technology than about like how they’re actually going into a market now. I guess that was my point. I think probably tech investors are reasonably qualified to answer those sorts of questions.

Vikram: Yup. I guess do you think that’s analogous to crypto or like how do you think about that?

Faizaan: Crypto is a whole another animal I think just because we’ve added this whole dynamic of the way… like just the amount of funds and how they’re being raised. Companies are essentially going public at this stage where I talked about like even before the stage of the 5 million revenue. It’s not even a technical person assessing is this useful or good. It’s an even stage before that where people are just implementing what’s in their white paper. That’s a whole different animal because I can’t think of somewhere where we’ve seen that much funding go to companies that are like essentially operating as a public company would but before like anything has been proven out.

Vikram: Right. On the alerting side, a bunch of interesting alerts come in recently where the first one is a cautionary tale that came through the platform. It was a CNN article titled “Bitcoin Crash: This Man Lost His Savings When Cryptocurrencies Plunged”. The story is about a UK real estate developer named Sean Russell and this is how they described him. “Russell rarely played the stock market and had little investing experience when he put around 120,000 into bitcoin in November 2017. He was stunned when that turned into $500,000 in just one month.” I remember this period like really clearly. Everything was going up like every day it would be like 20, 30, 40 percent. Like 2x and things would just go up all the time. Bitcoin was fun too.

Faizaan: Yeah, I think anyone listening to this is probably pretty tuned into what happened in November and December.

Vikram: Everyone is waiting for like when is that going to happen again. To be honest, when is alt season coming back. That’s like what they’re waiting for. They’re waiting for when all the alts go crazy again. We’ll see. Then that continues. “The dream didn’t last for Russell, who works as a property developer in the United Kingdom, buying homes and fixing them up. The price of Bitcoin surpassed $20,000 in December before collapsing. It

now trades at $6,300.” I don’t know if that $20,000 number is right maybe. There’s not a lot of nuance. It’s just it was around there but just so we don’t get called out by people saying “Hey, bitcoin was never over 20,000.” This is CNN. This is not QuantLayer. “Russell attempted to mitigate his losses by shifting money from bitcoin (XBT) to an offshoot called Bitcoin Cash and other cryptocurrencies including Ethereum and Ripple. But that didn’t work, and Russell says the paper losses on his initial investment have reached 96%.”

Faizaan: I mean when you’re losing money, the best thing to do is move it to stuff with increased volatility if you’re trying to like retreat, right?

Vikram: Right. You know, we obviously feel bad for this guy but like, there’s just all around, there’s this bad decision-making. One tenet of investing is don’t invest more than you can lose. People say it all the time. It’s kind of like this petty thing you say now but it’s so right, like don’t invest your life savings in something that is as volatile as bitcoin or the crypto market. It’s just silly. It doesn’t make sense. People also make bad decisions when they’re losing. Like I’ve seen this with traders a lot and that’s why PMs (portfolio managers) typically have stops for their traders, so a lot of like really short-term oriented trading shops. These public markets, I don’t think crypto markets are like this will have like pretty short stops. Like if you’re down 5% on the day, on your portfolio they’ll cut you off. They’ll just liquidate all your positions and say, you know, there’s a term, there’s like “go to the movies.” Basically the idea is like when you’re down a lot, just leave, go away and then go relax, like literally go at 2:00 watch a movie and then forget about your day, like take a vacation for a couple of days. Because psychology, it does matter quite a bit.

Faizaan: Yeah.

Vikram: People make bad decisions when they’re losing and people make really bad decisions when they’re losing a lot like so your portfolio manager of whatever doesn’t have any kind of risk management in place and you’re down like 10 percent, 15 percent, 20 percent, traders will start piling on. Like, “Oh, I’m just waiting for it to come back,” and “Oh, it’s down 20 percent, that means it’s cheaper and I can buy more.” “Oh, it’s down 30 percent? It’s even cheaper, I’ll buy more.” And they’ll go like all the way down they’ll just do that. I’ve seen that happen, it’s terrible.

Faizaan: Yeah, I can see that. I used to play poker and that’s like called going on tilt when you’re just… yeah, you’ve lost it basically at that point and you’re just doing crazy stuff.

Vikram: Yeah, psychology really matters. You want to be careful with this stuff. Investments in what are mostly experimental assets should be viewed through the lens of possible failure, like what if it goes to zero. That is an entirely possible scenario, so you want to imagine for that. When someone says don’t invest more than you can lose, they’re saying like what if this goes to zero, like how will that affect your life. It’s tough. Like I do recognize that it’s tough. Like we all have biases when things are going well, particularly in trading, because like you can see the number on that second by second basis. Like when things are going really well, it’s hard to see where they can go wrong. When people are like don’t be emotional about investing and that’s harped on so often, that’s what they mean. It’s like if you’re feeling really good about your investments, then just like pause. Like take a pause and reflect on like what you’re doing.

Faizaan: It goes both ways.

Vikram: Yeah, exactly. If things are doing really hard, like if your portfolio is getting crushed, like that’s what they say, go to the movies. It’s like you got to get your mind into a different place because the mind that you have right now that’s trading is not the right mind. If you have losing sense of mind, like you can’t turn that into a winning mindset just immediately. It’s not fluff. It’s like it can sound a bit fluffy but it’s really not. Anyway, so that was, you know, all in all t’s just a sad story, a cautionary tale of how not to trade.

Faizaan: Yeah, it’s unfortunate like 96 percent is pretty brutal.

Vikram: Yeah.

Faizaan: That really doesn’t happen unlike… That’s analogous to like gambling losses more so than trading, it feels like. Another alert we had come through just to touch on that, in the last podcast we talked a bunch about UX particularly around like passkey, passphrase, password. Same chat. Well, I’m just going to quote. “Hello! Who is admin? I forgot my log-in and password for web wallet account with 5 millions of BCN. Help me please!” Just to touch on this wallet UX is a very big deal and we see it consistently coming through on the Telegram chats.

Vikram: Yeah.

Faizaan: Another interesting Github alert that came through. Some of these, there’s not necessarily a whole investigation to follow up. It’s just the titles are very interesting and they allude to stuff that you definitely want to be keeping an eye on. This one was attempt at fixing wallet syncing crash. So, take that for what it’s worth.

Vikram: Attempt.

Faizaan: Yeah.

Vikram: It’s not a fix.

Faizaan: If I had a bunch of money in a wallet sitting on that coin, that’s probably something where I do a lot more follow-up into their code and what’s crashing and what are the implications.

Vikram: Yeah. A good version of that would be “fixed his wallet syncing crash.”

Faizaan: Attempt at fixing is like what’s the status and what’s the consequences currently. You know what I mean? You know, I’m making first questions. Right?

Vikram: Did you fix it? No, when people do this and I understand why people do this, like maybe you can’t test it locally or something. They need to put it out and then just try it in the real world. But yeah.

Faizaan: That’s actually his follow-up question is “Okay, I see the code is on Master but is this on the TestNet? Where is this being dried to? I also want to know that.”

Vikram: Right.

Faizaan: Yeah, it’s just the title of a lot of these commits can raise a lot of follow-up questions if you are interested in that specific point. Then another one that came through recently, it was the September 5 flash crash. Basically what happened was… and the causes, there’s a bunch of different theories people have on the actual cost but I’ll just describe what actually happened. Ethereum went from trading just under 500 ether per minute to over 10,000. That’s over a 2,000 percent increase. That’s a lot.

Vikram: Yeah.

Faizaan: I think it was Bitfinex which normally does about 20 percent volume, went to over 50% of the total volume of ETH. Then that was, you know, proceeded by a relatively big sell-off, hence the crash. Then bitcoin also you saw an uptick but it was not as much and it was after, so that leads people to believe it was more of a reaction to what was happening with ETH. One of the theories that was floated was Digix which was this large ICO was planning to liquidate $20 million worth of ETH, so that would potentially have a big move on prices but that’s not really been confirmed. If you do a search in our platform of just flash crash, you actually get a number of articles that come up that just have… you can just skim through the summaries of the different analyses and viewpoints that people have on what caused this. It’s interesting to read through some of those to see what people are thinking. Vikram, I’m sure you’ve observed some flash crashes in traditional markets.

Vikram: Yeah. They would happen for different reasons. There’s a few that I remember like there’s this term ‘fat finger’ and I guess it has its roots in like, basically the idea is like someone instead of pressing like a thousand pressed a million, they have such a large finger and added three more zeros to it. If you end up selling a thousand times what you meant to sell, you’re going to crash an asset, unless like at the professional level. Like if you were trying to sell $10 of something, like $10 of Microsoft stock and you end up selling $10,000 of Microsoft stock, that’s not a big deal. But if you’re trying to sell a million dollars of Microsoft stock and you try to sell a billion, that’s a problem. That happens with futures contracts for like the NASDAQ and the SMP. We’ve heard this term before, fat finger where someone ends up selling a lot more than they meant to and all the algos are watching. Now, the algos kind of do like there’s thousands of different algorithms that algorithmic traders use but some of the common ones are used throughout the market and if the market starts crashing, there’ll be additional selling. It’s not panic selling, let’s say, but it’s like algorithmic selling.

Faizaan: If the event is something that was way outside of some reasonable parameters, you can probably get some weird unintended behaviors where if like no one has ever did like a $5 million sell order and all of a sudden someone puts in a $5 billion one, you can probably screw what’s on these algorithms.

Vikram: Yeah. With some of these flash crashes, in the crypto market there’s been a few which have been like ether went into the pennies and some people what they do, and it’s smart, you should just do it, you know. If there’s a few exchanges that are like light on volume, you can throw some bids out there. Of course the risk is like you have to leave assets on the exchange which you don’t want to do for something that might happen but if you have a pretty small amount like you could just leave it and if it gets filled, like imagine just getting, you know, Ether is like $200, a little under 200 now, imagine getting filled at like 50 cents, right? It would happen on the stocks. I do remember it and I think there are a couple times where like people could profit from it and I think there is another time where the NASDAQ ended up canceling a lot of those trades.

Faizaan: Oh wow.

Vikram: Yeah.

Faizaan: One that I remember hearing of was the Knight Capital one which I don’t think it was fat fingered. I think it was a bug in the actual algorithm. Right?

Vikram: Yeah, right.

Faizaan: Where they lost like $400 million in 30 minutes or something crazy.

Vikram: That’s right. It was a few years ago.

Faizaan: Yeah. One more alert that I wanted to talk about today was basically opposed by this company called Menlo Core and so their whole thing is on proof of reputation and single sources of truth and Content Nodes. Basically to give you a gist of what they’re doing is if I’m a publisher, I can put some information on a blockchain and then independent parties can run a Content Node like basically making some sort of claim about what I said that’s verifiable. The idea is if I’m running a Content Node, it’s useful to developers because rather than having a single point of failure, like each of these Content Nodes essentially becomes an API that’s serving up the information that that publisher published. If I use their Content Node, I know I can trust the information that they’re serving me has not been tampered with through some verification process tied to that original blockchain that the publisher published on. But the Content Node is incentivized to run their node because they can monetize my cost of their API. The benefit of this is the publisher can put something out, know that it’s going to be served up in a way that’s truthful and not tampered with. The Content Node basically can serve up this information and get compensated for providing an API and as a developer I have this API that I can connect to for each Content Node and the more nodes there are, the more I can… it’s (A) verify that what they’re sending me has not been tampered with but also if like let’s say one node goes down, I can just use a different node and essentially like I have a backup API. They go into more detail about the exact implementation on their blog post and we can put that in the show notes. I just want to add a disclaimer that I haven’t studied their specific implementation so I’m not necessarily recommending Menlo Core specifically, but I just found the proof of reputation and the Content Node concept interesting because one thing that we run into when we’re doing on-chain analysis and looking at sources is that you can have certain sources that are a single source of truth that can like if there’s an issue, it’s market moving. The most notable example being coin market cap. I would suspect that there’s a lot of things that are connect to coin market cap that are trading algorithmically and if there’s major issues in that CMC data, it’s going to cause problems.

Vikram: Yeah. It happened even recently I think and there was some exchange, data that was messed up and it, you know, it affected all the algorithmic traders.

Faizaan: Like in my mind, a theoretical like blue sky scenario of all this would be that the publishers really just the actual original like let’s say the bitcoin node and then I can run a Content Node that provides like information about transactions, information about, you know, essentially I build an API on top of a bitcoin full node that’s easy for developers to interact with. You can run one, someone else can run one and so now we have this scenario where we can all verify, like we have an API that’s somewhat decentralized and verifiably truthful on what’s happening with bitcoin without having to run a full node or without having to even develop like our own analysis and aggregations on bitcoin data because I can even in theory publish how I do my aggregations and that’s verifiable against the node using the same preferred reputation system.

Topics:

  • Exploring the Elastic IPO
  • Prospectus summary
  • The “search” and the concept of “search”
  • What “Net Expansion Rate” is
  • Tech investing and how it relates to crypto
  • Uber versus Magna
  • Alerts:
  • Bitcoin crash
  • Attempt at fixing wallet syncing crash
  • Crypto flash crash

Links:

Hey everyone, this is Vikram again. Thanks for listening to us. If you’re
an exchange, a trader or working on a crypto project get in touch with us.
You can reach us on twitter at
https://twitter.com/quantlayer or email us
at
podcast@quantlayer.com.