This post was last updated on July 11th, 2019 at 08:21 pm
Today’s episode is an exploration of crypto trading bots and algorithmic trading with Daniel Anderson. Daniel is a software engineer and a freelancer who builds tools for cryptocurrency traders. Before he got into the crypto space, Daniel was a software developer for IBM.
This conversation is broken up into 4 chapters:
- Chapter 1: What trading bots are, how they work, and why they’re useful
- Chapter 2: Approaches for setting up, testing, and executing algo trading strategies
- Chapter 3: Popular algo trading strategies
- Chapter 4: The future of algorithmic trading
Topics Discussed In This Episode
- How John Mcafee inspired Daniel to start building crypto trading bots
- What trading bots are and what they do
- How algorithmic trading works
- Factors that new algo traders must take into consideration when backtesting
- How to account for slippage
- Portfolio allocation strategies
- How trading bot platforms allocate funds
- How fees are accounted for in algo trading
- Why you won’t find trading bots on decentralized exchanges (yet)
- Popular algo trading strategies
- Why doing arbitrage with crypto is harder than it looks
- Algo trading strategies for accumulating BTC versus USD
- What the future of trading bots looks like
Links Relevant To This Episode
- Cryptoinvestor Weekly Newsletter
- Clay Collins
- Daniel Anderson
- Daniel on Twitter
- Coinbase Pro
- Bitwise Hold 10
- CoVenture Crypto
“You’re going to take that set of rules, and you’re going to translate them into code into an environment that’s monitoring markets 24/7. That’s what a trading bot is in its essence.”
“Algo trading is all about the process, so you can actually think of this process as similar to the scientific method.”
“The other piece of advice is exactly what you suggested. Take our bot that we have with the lowest risk profile in something that’s just really simple to get people started.”
Welcome to Flippening, the first and original podcast for full time, professional, and institutional crypto investors. I’m your host, Clay Collins. Each week we discuss the cryptocurrency economy, new investment strategies for maximizing returns, and stories from the frontlines of financial disruptions. Go to flippening.com to join our newsletter for cryptocurrency investors and find out just why this podcast is called Flippening.
Clay Collins is the CEO of Nomics. All opinions expressed by Clay and podcast guests are solely their own opinion and [00:00:30] do not reflect the opinion of Nomics or any other company. This podcast is for informational and entertainment purposes only and should not be relied upon as the basis for investment decisions.
Clay: Welcome to this exploration of crypto trading bots and algorithmic trading. Today’s guest is Daniel Anderson. Daniel is a personal friend, a software engineer, and a freelancer who focuses on crypto trading bots and algorithmic trading, incidentally. Before diving down the crypto rabbit hole, [00:01:00] Daniel was a software developer for IBM.
Today’s conversation is broken up into four chapters. In chapter one, we explore what trading bots are, how they work, and why they’re useful. In chapter two, we discuss various approaches for setting up, testing, and executing algo trading strategies. In chapter three, we consider some of the most popular algo trading strategies in existence today. And finally, in chapter four, we close our conversation by considering the future of [00:01:30] bot and algo trading.
We’ll get right to this episode in just a second, but before we get started, I’d like to pause for a moment to tell you that this episode is brought to you by the good folks at Nexo. Here’s a word from them.
Nexo is the world’s largest and most trusted crypto lender offering automated instant crypto credit lines, which allow you to use your crypto assets—Bitcoin, Etherium, and XRP—as collateral to get cash in over 45 fiat currencies and stablecoins without selling your [00:02:00] crypto assets.
Nexo also offers interest earning accounts yielding up to 6.5% per year for stablecoins and Euros. Interest is paid out daily and you can add or withdraw funds at any time.
Nexo is also a strategic partner of exchanges, OTC desks, traditional and crypto funds helping them earn interest on idle stablecoins and fiat. The company’s growing portfolio of structured institutional products includes fully collateralized continuously rebalanced swap agreement [00:02:30] allowing counterparties to effectively manage their balance sheets.
Thanks so much to Nexo for making this content possible and available to you for free. Nexo’s a great company, they make a great product and you can find them at nexo.io.
This episode is also brought to you by the Nomics API and CSV data export service. If you need an enterprise-grade crypto market data API for your fund, smart contract, or app, or if you need historical CSV [00:03:00] dumps of trading data from top exchanges or even obscure ones, then consider trying out the Nomics API or our historical data export service.
Our API enables programmatic access to clean, normalized, and gapless primary source trade data across a number of cryptocurrency exchanges. Instead of having to integrate with multiple exchange APIs of varying quality, you can get everything through one screaming fast fire hose.
If you found that you or your developer have to spend too much time cleaning up and maintaining [00:03:30] datasets, instead of identifying opportunities, or if you’re tired of interpolated data and want raw primary source trades delivered simply and consistently with top-notch support and SLAs, then check us out at nomicsapi.com. Or if you’d like to order historical cryptocurrency market data as CSV exports from top exchanges, email us at firstname.lastname@example.org.
Okay, back to our regularly scheduled program. Here’s my conversation with Daniel Anderson. Enjoy. [00:04:00]
Daniel, what’s the origin story around your involvement with crypto trading bots?
Daniel: I originally got interested in cryptocurrencies once I got some skin in the game in around 2016, making my first purchase with actually Ethereum. I’m one of the few people who [00:04:30] I guess got to Ethereum before Bitcoin, but I guess it just depends on when you entered the game.
From that point on, I was just constantly watching the markets, started learning more about trading, and then I transitioned to a position in October 2017 at IBM. I was still at this point in time involved in the markets, trading the markets, but I have been increasing my exposure to production-level software, and machine learning [00:05:00] algorithms, as well. I just kept thinking to myself, “Man, I really got to start automating some of these strategies and thoughts I have.”
So, me and a close friend of mine started building our backend trading engine at the end of December 2017, during the peak of that bull market there. We just kept building and building. At that point in time, I was testing a number of different things with our trading bots, [00:05:30] making this thing fool proof and generic, so that any strategy could really be plugged in. That was my ultimate goal so that we could test and iterate over as many possibilities as you can imagine.
That’s what really transitioned me into cryptocurrency trading was taking my knowledge set at IBM, some of that production software experience and machine learning experience, and saying, “Hey, I think I can apply this to the cryptocurrency markets and we could build something really cool, here.”
Clay: What were the outcomes of those initial bots [00:06:00] that you had set up? Did they seem to work?
Daniel: At the time we’re developing these bots and our engine was very early, so we were mostly setting up simulations at this point in time, but one of the first bots I actually ran was really kind of a joke, a John McAfee trading bot. You might remember there was a point in time during the bull market where everyone was jumping on the bandwagon and they were saying, “Okay, John McAfee is tweeting the coin of the day, I’m going to buy the tweet.”
I [00:06:30] set up in our back-end engine a Twitter monitoring tool that was going to monitor his Twitter. As soon as it pulled that tweet I was going to buy. I had whipped this code together, not thinking about my API limits to Twitter, and totally blew them out and didn’t even make the buy. After that, the whole strategy was pretty much blown because everyone was doing it and he was tricking all the people who were doing this by posting either multiple symbols or images.
I refocused to doing more indicator-based trading. We have found some success [00:07:00] with even some of the simple indicators, SMAs or EMAs, and some other indicators we’ve come up with ourselves.
Clay: Let’s kick off chapter one, which is a primer on trading bots. Daniel, lets get really basic here. What is a trading bot at its core?
Daniel: In layman’s term, a trading bot is really just a piece of code that automates a strategy. You can think of that, you have an algorithm that you’re interested in, a set of rules [00:07:30] that define when you’re going to buy, when you’re going to sell, or when you’re going to go short, or go long. You’re going to take that set of rules, and you’re going to translate them into code into an environment that’s monitoring markets 24/7. That’s what a trading bot is in its essence.
Clay: I think there’s a lot of terms that are intermingled and often used together, people refer to trading bots, high frequency trading, there’s machine learning, there’s algo trading. How do you think about [00:08:00] this universe? How do you map out and create the landscape among these different but often interrelated approaches and tactics?
Daniel: There’s so many terms, so many different ways to think about these. What I would say is you’ve really got, at least from where we’re concerned, three different buckets here. I’m going to leave out the machine learning stuff for now and I think we could talk about that a little bit later. You’ve got high frequency trading, algorithmic trading, and automated [00:08:30] trading, and how these three concepts are really related.
High frequency trading is a class of automated trading that focuses on having the fastest connection and exploiting spot and futures prices on an exchange. Funny story with is, back when automated trading was starting, people were taking advantage of these spot and futures prices, basically minimizing the risk [00:09:00] as much as possible by doing this while maintaining a higher return.
I think something like Knight Capital had a glitch in their system and lost somewhere between $400 and $440 million, so you can imagine every second you’re not doing something, you’re losing money. That issue, which obviously your systems unit tested it’s good, typically this wouldn’t happen if you’re an HFT firm, but it did. That issue’s not going to exist at that scale on some of these other things.
Clay: You mentioned exploiting [00:09:30] spot and futures prices. How does a high-frequency trading strategy think about the spot price and the futures price? Is it about spotting a delta between those things, and then placing bets against that delta? How does that work, specifically?
Daniel: The HFT is all about the speed. Assuming everyone has the same connection to the exchange and the same numbers, the only advantage you’re really going to have an HFT that’s [00:10:00] going to make you the winner is a shorter wire or faster connection. You’re looking at these firms that are literally right next to New York Stock Exchange or something. These are the people who are going to be exploiting that and these are the people who are going to be winning on that.
In crypto, this really doesn’t—not on that scale—exist yet. I think Coinbase is working on an HFT-compatible system, probably more for the future, but a crypto is just too liquid for this sort of thing right [00:10:30] now to really be profitable and there’s just not enough capital flowing in. You’ve got your slippage issues and all that as well.
Clay: Maybe the markets aren’t deep enough to make this happen. Let’s move to algo trading.
Daniel: Algo trading is all about the process. You can actually think of this process as similar to the scientific method. Algo trading, you’re talking more about the theory behind it, and you’re talking about designing a strategy that [00:11:00] might be profitable. It’s split up in a few different steps. You’re going to have a hypothesis step where you’re going to identify maybe an inefficiency in the market or something that you could specialize in that would be profitable. You’re going to write down that strategy, or code it up, or however you want to do it.
Once you’ve designed it, you’re going to move onto back testing. After back testing, you’ll move on to forward testing, which is basically testing in a live environment without using the real funds, [00:11:30] but you’re testing in real time, on real data points. At that point, you can take your metrics and your results, and you either rewind back to the hypothesis stage or you deploy to a production environment. That’s where your automated trading comes into play is right at that deployment to the production environment. Or likely has already been in play at the forward testing stage, but for the most part that’s when you’re going to transition to a real automated system.
Clay: It sounds like step one is [00:12:00] generate a hypothesis. Step two is finding a way to actually operationalize that hypothesis, which often sounds like the hardest part. You can have some theory about why markets move, but actually creating something that a computer can trade against and not only coding that up, but supplying your algorithm with indicators that are reliable and actionable, sounds like a pretty difficult part.
Generate a hypothesis, operationalize the hypothesis, [00:12:30] and then you said back test, then forward test, and then either implement in you said live testing, and at any point along the way you could go back to your original hypothesis and revise. Then, the last step is executing in a production environment.
Clay: Can you give us some simple, very accessible examples of algo trading? [00:13:00] A really clear algorithm that might be simple but is easily understood by the average person.
Daniel: Imagine you want the ability to be able to lock-in a certain profit at a certain point in time, but you really don’t want to be monitoring that market 24/7. All you want to do is sell your asset at 1%, or 2%, or 3%, whatever it may be. Going through that process, you might look and say, “Okay. I’m going to have a buy signal.” [00:13:30] Buy signal can be generic, so maybe you’re just going to buy if the prices move down 5%. This is a really simple example. At that point in time you’re entered, you’re in a position. Then you would document, “Okay, my sell signal will be about 5% target.” In this setting your hypothesis is, “Okay, I think that anytime the price moves down 5%, that it will either revert or move back up and I can lock in a 5% profit.”
That’s a really simple example, [00:14:00] and that would be your hypothesis. At that point, you would back test that, you’ve clearly defined your buy and sell signals. If you’re getting more complicated, you might define, “This operates under a certain market environment, so maybe bull or bear.” We can obviously talk about some of those things later, but in that situation, that would be your hypothesis, you’ve then back tested it, forward testing, you’re putting it in a live environment, and then if it’s good, then you’ll lock it in and automate it.
Clay: I think it’s important to note that a lot of [00:14:30] these kinds of simpler strategies are often encoded in specific order types. If you’ve got stop loss orders or limit orders, it’s not exactly algorithmic trading. But there are things like this that are probably familiar to most traders. Would you say that’s true?
Daniel: Most traders. Definitely day traders are used to taking advantage of a number of different buy and sell tools, whether some exchanges offer more than others. [00:15:00] By default, those exchanges are equipped with a buy order, a sell order, and usually some form of a stop loss. This could be an advanced stop loss function like a trailing stop loss or floating stop loss, or it could just be a flat price where you’re only going to take a 5% loss. So, you’ll set a stop down there to help minimize your risk.
Clay: Technically, do algorithms prefer to sort of manually handle things like stop loss? [00:15:30] Or is it helpful as a developer of these types of algorithms when specialized order forms, or order types are offered by an exchange? Do you ever set parameters with specific order types that are helpful? And does that get you sometimes faster execution verses checking for the data yourself?
Daniel: Right now, what you’ll notice on a lot of exchanges is that Binance, for example, only offers standard stop loss. They don’t [00:16:00] offer the trailing stop loss, at least they don’t, yet. What does that mean for algo trading? That actually puts algo traders at a huge advantage, because they can now have a system that monitors the markets 24/7 and can create their own stop loss functionality.
That’s what we do. We basically have the rules in place, so that when we’re doing 24/7 market monitoring we can have any sort of stop loss we want. Whether it moves with the price, or it moves with the target, [00:16:30] or however you want to do it, it doesn’t matter if you can do it. That’s a huge advantage there as well. As far as execution type goes, one of the biggest issues here is the market liquidity. If you’re trading in a high volume market, it’s usually okay to do a market buy or sell with your trading bot, depending on how much capital you have.
That’s typically okay, and that might be your best option in some cases. The issue with using buy and sell orders with these [00:17:00] trading bots is that there’s no guarantee that you’ll get filled, and then your back test then becomes inaccurate. If that fill never goes through, the back test may not be applicable anymore, either. With those things in play, you have to do a little bit more bookkeeping to check that these things, the sell or buy order went through, and that the bot can continue trading. In some situations the market buys and sells actually do make sense.
Clay: It sounds like [00:17:30] one of the services that you guys offer is the ability to actually get advanced order types layered onto exchanges that might not have those order types It’s a way to execute trades in a way that maybe you’re familiar with but isn’t readily available on these platforms. Is that right?
Daniel: Yes and no. It will be available, but it’s not available, yet. I say that because it’s in our backend systems and we’re testing this stuff. You know it. [00:18:00] This stuff has to be unit-tested to oblivion. That’s really where we are now with that stuff, but that’s the idea and one of the huge advantages of going with an automated trading solution because you can really take the market under your own control. With the level of advanced exchange functions, you can really build on top of the exchanges offerings.
Clay: Let’s move to chapter two, which is a discussion of the various approaches for setting up, testing, and executing [00:18:30] algo trading strategies. What are some of the factors that algo traders maybe don’t take into consideration when back testing it?
For example, it strikes me that slippage is an issue on market orders, so you want to make sure that you have order book data. Just because the spot price was at a given place, it doesn’t mean that you can move the volume that you want to move without significant slippage. If [00:19:00] you assume that your huge fake virtual orders are moving the market not at all, then you might have a winning strategy. But if the thin markets for your shit coin, you could really be hosed. It seems that might be a factor.
Another factor might be exchange fees. There’s a whole host of things that I imagine are not looked at during back test, that could prove to be highly [00:19:30] significant when you’ve got real skin in the game. What are some of those factors?
Daniel: You rattled off a few, but there’s really a lot of these. I’ll start by responding to the first couple you mentioned. You mentioned slippage and exchange fees. You’re absolutely right in saying that these things are huge issues in a back test. I would say a lot of people getting started with algo and automated trading do not account for these things in their back tests. How can you account for these?
It’s improbable [00:20:00] that you’re going to have order book data going back to inception. That’s a lot of data to have, but you can estimate these things with certain formulas. That’s one tool you can use is you can estimate exchange fees.
You know Binance is charging a .07% fee or whatever it may be, if you’re using their B2B token. You can actually calculate how much you’re getting hit with fees throughout that entire period. That’s something you will want to be doing, and you’ll be surprised. People say that [00:20:30] if you’re making about 1% a day, that with all your capital and you’re always getting 1% a day, that’s something like a 3000% gain. It works the same way with exchange fees, with the way these accumulate. If you’re paying .5% in exchange fees every day, you can bet that that’s going to take a lot out of your strategies.
As far as slippage, having order book snapshots is great. That’s a way to on-the-fly while you’re forward testing, being able to account for slippage. You might even have some smart [00:21:00] parameters in your trading bot that says, “Hey, I’m estimating slippage is above a certain percent, I’m not going to do this.” Or maybe they’re going to sell half of the capital instead. You could do something like that in scale, based in real time, but as far as for newcomers I would say this, definitely take in exchange fees. You will be surprised how much that’s going to be hitting you.
Clay: We just identified problems with simulations or things that a newer person to [00:21:30] this space might not be thinking about. What are some other pitfalls during this process of coming up with a hypothesis, designing the strategy, historical testing, forward testing, live testing, production testing, what other issues come up?
Daniel: I think a lot of people miss some very, very valuable metrics. We use a scorecard system that measures—this is not full-proof by any means—risk and profitability. We have a risk score [00:22:00] and a profitability score that we calculate, and each of these scores comprises of a bucket of metrics describing that kind of scorecard. Then we can weight them appropriately based off, are we okay doing a more risky strategy at this point in time or are we not? We can weigh the risk score a little bit less and the profitability score a little bit more.
Now, lets dive in on some of those metrics that might be involved in these buckets. When you’re talking about risk, volatility is obviously one. [00:22:30] Slippage and fees is another. Then, you also want to be looking at something like max drawdown. Max drawdown is going to specifically tell you your percent difference between the peak and trough of your equity curve throughout a trading period.
That’s going to be a very important metric to consider during both back testing and forward testing. It’s actually interesting. We have an SMA strategy that we always laugh about because it performs so well over even [00:23:00] since inception, but it’s max drawdown is literally 85%. That is not a strategy you’re going to want to deploy.
Clay: Hey, this is Clay cutting in to define SMA. SMA stands for “simple moving average” and it’s calculated by adding an asset’s price over a certain period of time, and dividing the sum by the total number of periods. For example, if you wanted to calculate the 30-day SMA for a crypto asset, you would do this by [00:23:30] adding up the closing price for the asset for each of the last 30 days, taking that sum and dividing by 30.
Okay, back to Daniel.
Daniel: If your max drawdown is above 30%, you might want to look the other way. It depends. You could literally design a strategy that’s only focused on minimizing max drawdown, and then figure out the rest later, and then kind of layer on top of that. As far as the profitability stuff goes, you want to be looking at things like [00:24:00] your win-loss rate, how many times have you won versus how many times you’ve lost, or the difference between your biggest win and your biggest loss, or the percent difference, per se.
You could also be looking at profit factor, which is a huge one. Profit factor, is the sum of all your percent gains divided by the sum of all your percent losses. You want this value to be above one, if this value is above one, then that’s similar to having a more so [00:24:30] winning strategy. Profit factor is good as well.
Clay: The issue with max drawdown being high is that you could have a strategy that’s just crushing it day in and day out, but the reality is that you’re essentially gambling because any day you could wipe out months or even years worth of wins. Is that correct?
Daniel: That’s correct. In crypto, it’s even worse because you have flash crashes. I want to say [00:25:00] it was summer 2017 in June, Ethereum had a huge flash crash, and went down to one cent on GDAX before it was Coinbase Pro. At that point in time, Coinbase refunded everyone. They had margin trading available at that time, too, which they no longer have. Imagine having this killer SMA strategy that’s up 8000%, but you had a stop loss sitting before you bought into that run, and then it wiped out nearly everything because that stop loss got triggered.
That’s what [00:25:30] happened to thousands of people trading that day. Now, that’s an outlier case, because you’re not going to have a flash crash every other day, but if you’re trading in crypto you want to be thinking about these things.
Clay: I know algo strategies contain rules about trade execution, but are there also algo based strategies for allocation? You mentioned flash crash. I imagine if you have a flash crash bot, it’s probably ideal to have money on [00:26:00] hand and locked up around that particular strategy in case there is a flash crash. Does algo trading include portfolio allocation strategies or is that a separate thing all together?
Daniel: It can definitely fall into that bucket as well. A strategy can be very generic and you could have an arbitrage strategy that manages five strategies and moves between them based off the market state. That’s one example.
You’ve also [00:26:30] got your portfolio management theory, which will probably hit some of this stuff as well, but a strategy in its essence is very generic. It could, in fact, take on some of those traits. It might be the case that you always want to have 1% of capital set aside on the bottom of an order book across the top three markets just in case there’s a flash crash and just have a bot dedicated to doing that.
That system, from an automation perspective, can be a little bit more complex [00:27:00] because you’re monitoring in real time different agents that are trading across markets and doing different things. But that does not mean, obviously, that it cannot be done or that people aren’t doing it. That’s really where some of these key insights, especially in this space, and a lot of profitability lies is setting a little bit of capital aside to take advantage of some of those market inefficiencies.
Clay: How do trading bot platforms, in general, address [00:27:30] allocation of funds? Can you lock up funds specifically for given bots? Are they all pulling capital from the same pool? How do you account for a situation where maybe one bot says that you should throw a bunch of money to buy into a flash crash, but yet another bot is saying that there’s something else you should buy over here, and there’s limited funds? How does that work?
Daniel: This is something that [00:28:00] we are also actively working on as well, to make this thing as efficient as possible, because there are a lot of complexity in the bookkeeping side, really. One thing to take note of is right now, if someone comes on our system, builds a trading bot, and deploys it, it’s going to hook into an exchange. All the funds they have are sitting on the exchange, and at no point in time are we custodian peoples funds, not yet.
How does that work? User has five bitcoins [00:28:30] sitting on their account, and then they can allocate to that trading bot how much percent of that portfolio the trading bot is allowed to trade. So, they set constraints on it. These constraints can be generic and down the road they will be to where a bot can only trade 5% of the available capital at any point in time, or a bot can only trade maybe a flat amount.
Clay: Hey, I wanted to pause for a second to let you know that this episode of the Flippening podcast is brought to you by Nexo, who make [00:29:00] a great product. Here’s a word from them.
Nexo is the world’s largest and most trusted crypto lender, offering automated instant crypto credit lines, which allow you to use your crypto, for example Bitcoin, Ether, or XRP, as collateral to get cash in over 55 fiat currencies and stablecoins without selling your crypto assets.
Nexo also offers interest earning accounts yielding up to 6.5% per year for stablecoins and Euros. Interest is paid out daily and you can add [00:29:30] or withdraw funds at any time. Nexo is also a strategic partner of exchanges, OTC desks, traditional and crypto funds, helping them earn interest on idle stablecoins and fiat. The company’s growing portfolio of structured institutional products includes fully collateralized continuously rebalanced swap agreements allowing counterparties to effectively manage their balance sheet.
Thanks to Nexo for making this content possible and making it freely available to you. [00:30:00] Check them out at nexo.io. Again, you can check out what they’re doing and learn more about their lending and borrowing products at nexo.io
This episode is also brought to you by the Nomics API and CSV data export service. The Nomics Market Data API offers squeaky clean and normalized primary source trade data offered through fast and modern endpoints. If you’re a professional investor or developer, you need accurate prices, and I can tell you that our competitor’s [00:30:30] prices are simply not accurate in most cases. We’re actually going to come out with a report about this soon.
I should also note that the NomicsAPI is used by many of the largest and most influential companies in the space, including top exchanges, $100+ million AUM funds and prop shops. Hit us up if you’d like more details or go to nomicsapi.com. Furthermore, if you’d like to order historical cryptocurrency market data as CSV exports from top exchanges, email us at [00:31:00] email@example.com.
Okay. Back to the show.
Daniel: If you set up these constraints correctly, you can create that kind of environment where different bots can be assigned to different things. Imagine a situation where you’ve got two bots responsible for 90% of the portfolio. What’s going to happen—we’ve seen this in testing as well—a little while back where you’ve got one bot grabbing, performing a buy order, 90% of your [00:31:30] portfolio, and then what’s remaining is that remaining 10%. The next bot takes 90% of that 10%. Basically, what you’re doing now is you’re dollar cost averaging across all your funds. When I was testing that some months ago, it actually turned out to be weirdly profitable for that period. I’m not sure if it’s a coincidence or not.
Clay: Was that during the bull run at the end of 2017?
Daniel: That was not, actually. This was maybe March, April, [00:32:00] but I thought it was pretty funny. I spend a lot of time coding on our systems, unit testing and stuff, so I didn’t go too deep into that. I just looked at the code and I said, “Hmm, I don’t know about this,” and so I worked on it. That was interesting.
Clay: Kind of thinking about unaccounted for costs or factors that might not be included in most algorithms. What about decentralized exchanges and fees, moving [00:32:30] money into and out of the exchange? For the most part, transaction fees, at least right now, are pretty low with decentralized exchanges, at least the ones that do a lot of stuff on chain. There’s gas costs and things like that. Have you found that these kinds of fees move the needle or are they fairly insignificant right now?
Daniel: The challenge is the fees tend to scale with the market cycle. They’re very low right now, and it’s not a huge issue. Right now [00:33:00] we are focused on bots that operate on one exchange, and that’s not always going to be the case. You’ll eventually want to have bots that will be able to withdraw and transfer over to other exchanges, maybe because you want to exploit arbitrage opportunities or do something like that.
Back testing something like that and measuring that does become really important. I think the challenge there is being able to know what might the fee be based off how much demand [00:33:30] is there. You could get into analyzing the on chain data and that does become important. Right now, it’s something that we’re not focused on, but if we were focused on that, we would be analyzing the on chain data and we would be, in real time, grabbing these average fees or whatever they may be, because I’m sure you’re aware of things like gas wars and stuff like that. That’s probably the next generation of high frequency trading across Dexus is having just crazy gas wars with people to get on the Dexus. [00:34:00] That’s probably where it’s all going to go.
Clay: It seems like bot activity in the crypto space is almost exclusively focused on centralized exchanges versus decentralized exchanges. Do you think it’s fair to say that it’s not practical right now to be executing bots on decentralized exchanges?
Daniel: I guess you could say it’s not practical in the sense the liquidity is just not there yet, and that’s the challenge. At least with Binance and [00:34:30] Coinbase Pro, if you’re a high net worth player trading with a ton of money, you’re probably not even going to be on Binance or Coinbase Pro. You know this. People are going to be OTC.
At least for now we operate on the centralized exchanges and that’s because they have liquidity for the top three, top five markets. Our live trading, we will have constraints in place so that low liquidity markets are going to be off limits for a little while, until we see some better. We want to see [00:35:00] better stuff there, even on some of these centralized exchanges. Some of these “shit coin markets” is not going to be a good look if someone back tests something like that, they don’t look at the correct metrics, and then they’re really upset because it performed 50% less than they were expecting or they lost a lot of money.
That really is a challenge and Dexus are faced with that right now where there’s almost a liquidity crisis. I believe that Dexus will get [00:35:30] there. This might be a little bit out there to say, but I think they’ll probably be the future of OTC trading once the fees and the liquidity stuff can get figured out.
Clay: That makes sense with solutions like air swap and other similar solutions.
Clay: Let’s transition to chapter 3, which is an overview of some of the most popular algo trading strategies and how they work.
We were talking about these steps that go into building [00:36:00] out and testing, and then deploying an algo trading strategy. Let’s walk through a hypothesis, and how you might go about converting that hypothesis into a series of roles. We talked about the up 2% down 2% strategy. Let’s talk about something a little bit more conceptual, maybe a roll around news or a sentiment, something that’s not [00:36:30] as straightforward as our previous example.
Daniel: Let’s talk about trend. I think the trend is a great example to grasp this entire process. Imagine you’re an algo trader, and you’re interested in devising strategies that take advantage of a trend change. I’m going to give a very simple example here and by no means do I think people should go out, take this example, and start trading [00:37:00] it right away. In this specific example, it does have a terrible max drawdown, so I will say that.
Imagine you want to open a position when the trend changes to a more bullish trend. So, you’re trying to kind of predict the trend. You would maybe use, in this most simple case, a simple moving average or an SMA is the acronym. The 200 SMA is the average of the last 200 prices and it works in a rolling window. Every time a new daily candle [00:37:30] opens, it’s going to knock off the last one, add a new one, and then you’ve got a new rolling average. You take that value and that plots a line on your chart. That gives you a good indicator of what trend you’re in. If you’re trading below that line, you’re typically in some sort of a downtrend. If you’re trading above, you’re in an uptrend. Generally, that’s what people say.
I could then take that knowledge about how that SMA works and I could say, “Okay, I’m going to devise a strategy that will open a position once [00:38:00] we’ve changed trends.” How would I do that? My hypothesis would be if the close price moves above the 200 SMA, then I’m going to assert that we’re moving into an uptrend and I will open a position at that point in time. That would be my hypothesis. My hypothesis would be I want to open a position when the trend changes and that would be the rule set that I follow. I would move forward to back testing that and I would see what is the profit factor, max drawdown, [00:38:30] win-loss ratio, expectancy—I didn’t mention earlier, but expectancy is your expected returns per trade, which is also a very important metric—return on investment, average return on investment.
I might combine these things. Then back test looks good, so I’d deploy it in a lightly automated environment, forward test it for maybe anywhere between one to six months, depending on what you’re trying to do. If it’s a shorter term strategy, maybe [00:39:00] a few weeks if you’re trading on five minute candles or something. You’re trying to take advantage of the current market, but we’ll stick to the daily for this. That would be the process. I would highlight, this is what I’m interested in, this is what I know based off this environment, and this is what I want to test. If I’m happy at the end of that forward testing period, then I can deploy it to a real time trading bot.
Clay: What are some other general categories of algo strategies?
Daniel: You can go as far and deep with this if you [00:39:30] want, but for simplicity’s sake, there are some that will resonate with most people that are definitely worth discussing. We talked about trend following, which is typically denoted by using a simple moving average or maybe an exponential moving average. There are also mean reversion-based strategies, which you’re basically betting on, “Okay, this strategy moved so far away from its mean, that it’s going to return.” [00:40:00] A mean reversion strategy will be making a bet that the price will return to the mean. There’s a number of different indicators you can use for those.
Another category would be sentiment trading. Sentiment trading is really interesting, and it’s by nature a machine learning task. What that means is that it requires a lot of data in real time to classify and tell the trader or the algorithm what the current state of the market might be, [00:40:30] how bullish is it, how bearish is it. Instead of using the price data you can use news sources, or tweets, or you could even use Reddit activity. You might even throw GitHub activity in there, as well.
Clay: Interesting. What’s another category or strategy?
Daniel: You could also be doing strategies that are super niche or specific. One example might be front-running other trading bot. Good luck, but you could try. [00:41:00] That could very well be a strategy. You could also front-run if you know certain firms are auto rebalancing their portfolios in certain points in time. You could create strategies that front-run those.
Those are very specific strategies that rely on a certain set of knowledge. They’re also flawed because what happens if that fund goes down, or the fund changes when they execute or changes their rebalancing? You’re going to have no knowledge of that. Where people might think [00:41:30] these are risk free, it’s really you get into a lot of game theory here.
Clay: Thinking about some of the larger, or more well known crypto index funds, there’s the Bitwise Hold 10, CoVenture Crypto has an index fund. There’s a lot of crypto index funds, but if you knew that one of these funds had a lot of assets under management—most of them, I think all of them, publish what their indexing, and [00:42:00] rebalancing criteria is—it does seem like you could buy coins or sell coins in advance of those funds doing those things.
There’s probably always the anomaly where they go before their investment committee and their investment committee says, “This shit coin is too much in the gray area, we’re not going to buy it.” Or where someone makes an executive decision. This is probably not a 1000% reliable, but certainly for the top coins when it comes to [00:42:30] just changing the dial and the allocation versus buying and selling all together, there’s some potential opportunity there, but the markets aren’t deep enough and these index funds probably aren’t large enough in general for this to work, yet, but there may come a day.
Daniel: I would agree. I think that in traditional markets this is big. A lot of people like to do this in traditional markets because the volume is really there, the liquidity is really there. Right now, I think there’s just too [00:43:00] much going over OTC that you could trade OTC to try to beat these, but that price action is probably not even going to be shown on an exchange that you’ve got a bot running on, so I don’t think it’s very applicable right now.
As the space really matures, these could be some really interesting strategies to keep in the back of your head, as there are more and more kinds of funds, there’s more liquidity on something like a Coinbase Pro, and people start actually implementing self-balancing strategies on these centralized [00:43:30] exchanges. Then, there’s some real opportunity there.
Clay: In a previous episode, we interviewed Kevin from GAWA Capital. I think one of the most interesting aspects of that interview was he was walking us through what the real time order placement and trading environment is. Often it happens behind chat interfaces and there’s this ritual around [00:44:00] executing trades. Sort of the opposite of a safe word, there’s a phrase you say when it means, “This is actually done, and there’s no going back on this. If you tell us that you want to order, we’re going to order.”
I know it’s done outside of crypto, but I wonder if in crypto there are trading bots at some of these larger firms that actually operate behind these chat interfaces. I wonder how many of these OTC desks are sometimes receiving trade orders from bots. [00:44:30] Do you have insight on that or any intel?
Daniel: I can’t say I know specifically. You might be better suited to ask someone whose more involved in OTC trading themselves. But that’s an extremely compelling thought and that’s something I didn’t really even consider that.
Clay: We talked about front-running index funds during rebalancing periods. We talked about trends and SMA’s. What about timing strategies?
Daniel: Timing strategy [00:45:00] is definitely a more generic term. You can be using specific indicators that execute at a certain time, or you might just have timing strategies that quite literally use the time stamp and they buy it at a specific time. This isn’t something that we’ve researched heavily myself, but as far as timing strategies go, we have a lot of technical analysis indicators that can be combined and mixed and matched. [00:45:30]
The timing there really happens when certain phenomena exists or certain conditions happen. Your timing might be when SMAs cross, or EMAs cross, or something goes above it, relative strength index, RSI value. Our timing strategies might be more focused on the current state of these indicators opposed to maybe buying at a specific point in time, which, if there are certain patterns that could be interesting as well. [00:46:00]
Clay: Let’s transition a little bit from talking about timing strategies to sentiment strategies. You mentioned that you can pull indicators from Reddit posts, from Twitter, from news articles, maybe GitHub profiles. What are some other useful sentiment indicators? How do you think of sentiment as a broader category? Has it historically been useful? Has it worked?
Daniel: You could categorize [00:46:30] this however you want. It’s mostly good or mostly bad, or maybe it’s mostly bullish or mostly bearish. How do you measure this kind of spectrum in the world of markets or crypto markets to be exact? I’ve worked with machine learning algorithms that do something called sentiment analysis. What sentiment analysis is, is it takes in streams of data in real time, and it trains these machine learning models [00:47:00] that tells you the general sentiment of maybe a body of text, or a basket of pictures, or a bunch of social media examples.
What does that mean and how does that apply to market? First off, I’ll talk just briefly about machine learning in its role in sentiment analysis. A lot of people hear machine learning and AI, and they think the two things are exactly the same. They’re similar, but they’re definitely two different things. You’ve got AI, [00:47:30] you’re talking about programming a robot to already have knowledge, or to react like a human would in a specific scenario. It’s more of an explicit programming task to create the illusion that it’s operating like a human.
Machine learning is focused on creating a machine or a computer that can learn and gain knowledge over time. This is a lot more interesting because it’s flexible. So, machine learning is [00:48:00] used a lot in market and sentiment analysis, really as a tool to measure current market states.
How would a sentiment strategy work? Basically, what you might do for a sentiment strategy is you would have some machine learning model that’s taking in maybe Twitter, Reddit, and other social media outlets, taking in data in realtime, then using that data—you would have to annotate this data, so this is where it becomes very laborsome—you would have to take the data in [00:48:30] your trading set and say, “Okay, this example’s bullish, this one’s bearish.”
You create basically flashcards for your system to trade on, and then you deploy it in real time. It has this preset knowledge, its learned, its trained itself, and it’s going to tell you the markets bullish, or the markets bearish. We measure this as a sentiment index. This is a value between zero and one, where zero is bad, and one is good.
Sentiment can really be anything, you could also fear versus greed, bullish versus bearish, [00:49:00] or good versus bad. You can use this to give you indications about market environment, or to quite literally trade off. If the index goes above a 0.5 you might open a long position, or if it goes below 0.5 you might open a short position.
Clay: You mentioned arbitrage for a second. I think a lot of people have spotted these wide gaps in the market across different [00:49:30] exchanges. Often it seems like it’s not actually that practical in crypto, yet, to deploy. Why is that?
Daniel: A lot of people who I talk to outside of crypto say, “Why are you spending all your time building these trading bots for other people and not just doing arbitrage?” I hear it all the time, because there’s so many, what seems to be arbitrage opportunities. There are these opportunities, don’t get me wrong, but it’s [00:50:00] a little bit more complex.
Right now, for example, if you go on Coinbase Pro, they’ve got a ETH/USDC, and ETH/USD market. The interesting thing about that is that anytime there’s a $1 spread between the two, that’s free money. However, the kind of catch there is that the USDC market is maybe 5% of the volume of the USD market.
The arbitrage, depending on how much capital you have, could only really work in one direction. [00:50:30] If you’re dealing with really small amounts of capital, it might be practical for some folks. Now, if you’re looking at cross exchange arbitrage you’re going to have huge liquidity problems. Sending out ETH from one exchange to another that has maybe 1% of liquidity, then after you get hit with fees and slippage, you’re not actually making any money. That’s a huge issue as well.
The last issue I’ll mention—I think you and I have actually talked about this before and this is part of the problem you guys solve at Nomics—is conflicting currency symbols. [00:51:00] I have actually witnessed certain symbols being reused for different tokens. Then you’ve got these smaller exchanges that have conflicting symbols that may look like a very good arbitrage opportunity, but they’re just completely different assets.
Clay: Yeah. Or they change the symbol one day and all a sudden your bots are spotting these huge arb spreads that aren’t actually real, and you just lost money.
Daniel: Oh, yeah.
Clay: What about arb loops [00:51:30] within an exchange? Let’s say you buy bitcoin from the BTC/USD pair, then you buy some ether with that bitcoin, then you maybe buy some ZRX with that, and then you go back into another token, back in ether, back to bitcoin. When you’ve completed this loop very quickly, you’ve made some money. It sounds like a fantasy. Do those things ever exist or is it complicated? [00:52:00]
Daniel: I’m sure they do, but I think you mentioned there were four swaps there. I’m just imagining how much of fee nightmare that could be. You’re doing this, then all it takes is adding a couple of more markets to that, and you’re back where you started, you rewound a couple of percents. I imagine those opportunities do exist, but it might be more interesting to maybe look at the correlation of two assets and doing some arbitrage between [00:52:30] the spread of two assets or when they drop below a certain level of correlation.
You could be market-agnostic in a degree to where you’re only looking to accumulate bitcoin, so you’re looking at the correlation between ETH and bitcoin against the US dollar. When the spread hits a certain amount, you might buy some ETH and increase your bitcoin holdings. There’re opportunities like that, that I think are probably a lot more profitable than trying to run this endless loop of arbitrage, because eventually even [00:53:00] if that works, it’s going to end. You’re going to solve the efficiency yourself and then you’re not going to be able to do it.
Clay: Kind of going back to the general category of statistical arbitrage, we spoke for a little bit about mean reversion, and you just mentioned correlations. Have you found there to be highly predictable and reliable correlations between assets, in the sense that it’s possible to spot lagging correlations? Maybe [00:53:30] if Monero or Zcash go up, then—I hesitate to even say it—perhaps that category, other assets like that like Verge or Dash might also go up. Have you found those to work?
Daniel: We haven’t assessed that example exactly, but we look at correlation and this is something we’re very interested in looking into more and making more accessible. I would say when it comes to these correlation coefficients, [00:54:00] the one big issue you’re going to see right now in crypto is that everything is really tied to bitcoin, and that creates a challenge there. You’re almost like, why not at that point just bet on bitcoin, but that’s not to say that there aren’t groups that move together.
I do believe these groups exist, and I’ve seen actually some folks on Twitter—there’s a huge trading community on Twitter—and I’ve seen people seemingly post that they found these groups of lower capped altcoins [00:54:30] that do move together, and it makes more sense for those type of coins because it could just be quite literally like John Dow and his buddies just rotating these tokens, then someone spotting it, and then it’s gone.
I think it’s going to be an extremely important thing as the market matures as well. Correlation coefficients can be a little bit more advanced to the concept, but it really can give you insights into how these coins move and how they move in relation [00:55:00] to one another.
Clay: It seems like the strategy you pick, like most things in life, depends on what your goals are. If you’re a bitcoin maximalist and you really believe that bitcoin is the future and it’s ultimately going to win, your strategy might be not to maximize for USD gains, but to maximize your BTC gains and to accumulate bitcoin.
What are some categories of strategies that one should consider [00:55:30] when thinking about accumulation in BTC versus in USD?
Daniel: Really, it’s everything that we’ve discussed this whole time. You’re going to take a bucket of these, back test, forward test, that whole process we talked about, and you’re going to find one that accumulates really well, depending on how much capital you’re dealing with. Some retail investors can be very successful at this without automation because they can identify lower cap coins [00:56:00] that might outperform bitcoin on a given day, and a 10% gain on their stack, a 100% gain on their stack, but that’s not going to be a long-term strategy for someone with a lot of capital.
I think that’s actually a very compelling thing to mention is the accumulation of bitcoin, or the accumulation of ether. This space has such a rich bucket of people who have gone so deep down the rabbit hole and people who really believe [00:56:30] in this stuff, myself included. If you believe so much in bitcoin or you believe that as I think Tim Draper’s called for a quarter million in a couple years, if you believe that much then why are you not accumulating more bitcoin?
We have these trading bots and they run on all these simulations we’re running and testing. They’re running on all the Binance markets. If you’re looking at Ethereum markets, BNB markets, maybe [00:57:00] you really believe that Binance just want to accumulate BNB or bitcoin market. We’ve found that you can. You can find these accumulation strategies that are successful, but then the case of bitcoin goes to zero, God forbid, then you would be out of luck. But I don’t think anyone running those strategies is making that bet.
Clay: If you’re an institutional investor and this is your strategy, then that’s one thing. You’re a professional, you raised a [00:57:30] fund based on the strategy, investors gave you money, the credit investors gave you money explicitly, to execute against the strategy, and perhaps they allocated a portion of their investable balance sheet into putting money behind an algo strategy, or a bot strategy, or high frequency strategy, or whatever their strategy is.
If you’re a consumer investor, how [00:58:00] do you think about this? How do you recommend the average consumer investor think about the role of algorithmic trading? I think the reason why this is an important question is because prior to crypto and the development of this cryptocurrency investing ecosystem, it really wasn’t possible for the average consumer investor to deploy a trading bot. There’s the engineering side of this. [00:58:30] There’s the math behind this.
Let’s say one of your parents or a relative came to you and said, “Hey, your platform looks cool. I believe in you, I want to put some money behind this.” What advice do you give to them? How do you think about the role of consumers in this space? What is advisable and what isn’t advisable to them?
Daniel: You’re right, it’s a very important one. [00:59:00] There’s two different buckets. There’s the consumer that knows nothing about crypto, then the consumer who’s been involved with it. I guess there’s a third bucket, as well, of the very technically versed. I’ll start with the first one. They know nothing about crypto, they really like the platform, and they want to get involved. I would say, recently my grandmother actually reached out to invest in crypto currencies and I told her, “Just put in a $100 right now. Don’t put in anymore than that.” [00:59:30] She’s old, she doesn’t need to have a heart attack right now, so it’s a lot of volatilities.
That’s one piece of advice, and I think you’ve heard this from everyone. First off, treat it like gambling money. If you can lose a $1000 that might be ceiling. If you can lose a $100 and it doesn’t make a difference in your life, you’re not struggling to pay rent, and you’re not having to downgrade to ramen, [01:00:00] then that’s your ceiling and that’s what you put down. The other piece of advice is exactly what you suggested. Take our bot that we have with the lowest risk profile in something that’s just really simple to get people started.
Now, with that second bucket of people who maybe they just started trading crypto and they’re interested in this stuff, but they don’t know where to begin. I always tell people, “Learn about the technical analysis.” The technical analysis is the foundation [01:00:30] for these market movements and how they work. Understand some basic technical analysis indicators. That includes for trend you’re looking at, simple moving average, exponential moving average, MACD—moving average conversance, divergence—then look at some mean reversion. You might look at Bollinger Bands, you could look at RSI—relative strength index—that could be used for mean reversion, or trend following. [01:01:00]
Learn why these indicators mean what they mean and why there are certain values at certain points of time. Then you can really get behind this idea of why some of these things may or may not work, depending on what you believe. When people really get involved, I say just look at the TA and learn some TA first. I think that’s really important to get across.
Clay: Most bot platforms, yours included, include some means for testing strategies. In fact, I think it’s [01:01:30] probably unethical if they don’t. Some means for testing strategies, back testing strategies, and maybe if you’re not just the average consumer but maybe you fall into that prosumer category, or you’re really hungry for education and to know the market better, it sounds like it might be worthwhile to develop a different relationship with what’s happening in these markets outside of what you’re reading on Twitter, seeing in the newspapers, and hearing [01:02:00] from others. It might behoove you to see this space through a different lens, and one way to accomplish that is by learning about these indicators and looking backwards, which most people don’t do a lot of in general. It sounds like some of these strategies in the operationalization of these strategies are a good way to do that.
Going back to advice for the average consumer, I mentioned a [01:02:30] flash crash bot. If you really believe in bitcoin and there’s a price at which you think it’s just stupid crazy not to buy it at, it might be interesting to look into setting up a flash crash bot. It sounds like you advised SMAs, Bollinger Bands, and a few of these others. Is there anything else that comes to mind as being fairly straightforward, but also highly useful in terms of getting to know this trading strategy, and approach a little bit better?
Daniel: We did talk about SMAs [01:03:00] and Bollinger Bands earlier. I’ll prefix this by saying obviously as you mentioned before, not financial advice for anyone, but we tweet these out every day on Twitter. We tweet out our best performing bot, our favorite bot for the day, and we give out their strategies. Why do we this? People be like, “Why do you give out your strategies?” Honestly, because we have so many of these things running, we want people to see what we’re doing and we want people to see that there’s real data [01:03:30] and there’s real stuff happening here.
We’ve been tweeting these bots out for, I want to say, about three weeks now, and we’ve noticed one strategy. You can go on to Twitter and you can look yourself, so it’s not like I’m revealing any secret sauce here. It’s a local minimum maxima strategy. I’ll explain what it is and why I think it’s done reasonably well, mainly in a sideways market environment, in between these [01:04:00] market dumps. It tends to get a lot of good trades.
You’ve got a trading window. Maybe you’re looking at the last 20 days and you’re tracking the local minima over that period of time—the lowest price that’s ever been hit—and the local maxima—the highest price that’s ever been hit—in that moving window. So, say that’s 21 days or something. What can you do with that information? You can buy when it goes [01:04:30] below the local minima, which is a mean reversion, or you can sell when it hits the local maxima.
This is almost if you’ve heard of trend lines or support and resistance lines, this is very similar, but you can zoom in more. A lot of traders really like the support and resistance lines, and have been very successful with them. I think that’s why this strategy has done decently well. It looks to be mainly in some of the bitcoin [01:05:00] markets, but we found that one to be interesting, so I think it’s fun to share that with folks, and be able to talk about that one.
Clay: How do you think about strategies that work in bull markets, markets that are going sideways, and in bear markets?
Daniel: First off, I do not have an algorithm with any sort of crystal ball. If anyone ever says they do, do not trust them. That’s the first thing I’ll say. Following that, I would say, [01:05:30] obviously, in a bull market, anyone can really buy something and hold it. It seems easier to make money, and I think that’s because also with the lack of additional instruments. You can’t short some of these assets. What do you do? You go on.
In these different market environments, there’s certainly a lot of different things you can try, but it helps to think about what are people doing in bull and bear markets, and how can you [01:06:00] leverage that? I’ll start off with kind of the bull market, my thought process, and then we can talk about the bear market a little bit.
In a bull market, people are just buying this stuff up. In crypto, if you’re looking at the wicks of these candles or you’re even buying as the stuff returns to the mean, instead of mean reversion, you’re buying as it goes back to the mean and then taking some profit on the way up. This could be trend following, for example. Trend following could [01:06:30] be something that’s very powerful in a bull market.
Now, in a bear market, people are going to be profitable, primarily by having some sort of short strategy or short exposure. Right now, we’ve seen Ethereum, USD shorts are hitting all time highs. Eventually, this will revert. Maybe this is the indicator for that. Who knows? But right now, it could be interesting to watch the levels at which people begin [01:07:00] shorting, and you could try opening shorts at some of these different levels or when it returns to the mean. This could also be thought of as some sort of trend following or some strategy.
I would say, watch the trend. The trend is extremely important when you’re trading any strategies. If you’re trending down, you want to be looking at some short exposure, some short-based strategy. If you’re trending up, honestly, try a lot of stuff. Trend following is always good. [01:07:30] As we saw in the last bull market, there didn’t seem to be a lot of rhyme or reason into why people were making a lot of money. A lot of that had to do with the ICO craze as well, which at this point in time is really wound down. So, who knows what the next market cycle might look like? The entire microstructure could change when new money comes in as well. It’s very accurate. This could all be extremely irrelevant.
Clay: What about when markets are moving neither up nor down? They’re just [01:08:00] moving along in pretty languid pace. What do you do then?
Daniel: Sideways markets, some people say they’re not as exciting, but these are great environments to accumulate. If you are that bitcoin maximalist, you might look at some support resistance strategy or a min-max strategy, so that you can accumulate as much bitcoin as possible before the next move. If you’re betting on a move up, or a move down, sideways markets are a fantastic accumulation [01:08:30] opportunities.
I think we talked about Bollinger Bands earlier. That’ll basically tell you how far off the mean something has moved. I think you’ve got the upper and lower Bollinger Band where it’ll tell you, “Are you two standard deviations above the moving average or two below?” That might be something you consider using in a ranging market. We also like that min-max indicator that I just mentioned, as well. It’s a matter of kind of trying [01:09:00] these different strategies and indicators, and seeing in these sideways markets how can you accumulate as much as possible?
Clay: Let’s cap off our conversation with chapter four, which is about the future of algo and bot trading. Daniel, when you think about the development of algorithmic trading, bots, and the ecosystem that enables this, what does the future look like?
Daniel: I’ll mention some things I’m excited about in the future, [01:09:30] some thoughts I have, and we can go off that. One thing I’m really excited about are security tokens. I think that the ICO craze is over now and I think a lot of people are excited about security tokens.
Another thing I’m really excited about is actually more decentralized platforms, and true ownership of video game assets. I think that this is really an untapped and high potential market where you’ve got people who will pay fortunes for unique skins [01:10:00] on video games or assets on video games. I think in the future as blockchain technology becomes more adopted, we’re really going to see a new marketplace open up that is probably going to be even crazier in volatility than crypto was. That is basically the trading of assets across video games.
Imagine trading your Crypto Kitty for a character in World of Warcraft. That’s a crazy example, but that paints [01:10:30] the picture. I think that blockchain really enables that sort of capability, so that’s something I’m excited for. Then, as far as the markets themselves, I think the price movements in the markets is going to change a lot once bigger institutions enter.
Right now with retail getting in first, which wasn’t the case with previous traditional markets, it caused a full market cycle to go through at hyper speed. I don’t think that that [01:11:00] will necessarily be the case moving down the road because I think the market microstructures is going to change. There’s a lot of interesting things to consider there. Do back tests still accurate for 2012 through 2018, if the market microstructures so different?
Clay: What about at the infrastructure level, with regards to enabling what you do? I imagine [01:11:30] better liquidity certainly helps everyone in this space, more advanced order types, it seems like it’d be helpful having an exchange that’s not BitMEX that allows for longs and shorts, and is available to US customers, or at least people in more jurisdictions. I think having more access to historical order book data. What are some of these other foundational things that are on your wish list for the future, in terms of enabling what you do in your platform? [01:12:00]
Daniel: Liquidity, obviously. Everyone wants liquidity. Coinbase is already offering new assets and I think that’s fantastic. I don’t want them to add 300 coins or whatever, but if they can add valuable assets and create more trading opportunities there, that’s fantastic. I would say that’s one thing on my wish list is to see more assets on Coinbase and more trading opportunities there, especially being based out of the US and that being [01:12:30] one of the exchanges we use a lot. That would be great.
Then, I really want to see a futures exchange that is friendly to the US and has a good API that we can hook into and take advantage of that. Right now, BitMEX is there and BitMEX is professional UI, it’s a great trading platform, they’ve got a good API. I think we just need more players in that area and unfortunately I think it’s going to take some time. I would say [01:13:00] definitely more futures exchanges, and some more availability to assets on these US exchanges is good.
In general, I’m excited for more liquidity on Dexus. I think more liquidity on Dexus and even Binance has got their DXS coming out. I’m really excited for that and I think that’s going to open up some interesting trading opportunities. [01:13:30]
Clay: Well, that wraps up this conversation with Daniel Anderson. I hope you enjoyed it. Before you go, I want to mention that since we’ve started producing episodes at a much higher rate, we now have room for a few more sponsors. If you like the work we do, you’ve gained value from this, and would like to support this show, then a sponsorship might be a good fit for you.
I can say from our own experience that Flippening sponsorships work. [01:14:00] Each and every time we put out an episode of this podcast, we mention our own API. And to date, every single one of those advertisements has resulted in at least one customer. In fact, we would do these shows even if nobody else sponsored because of the business it brings to us. And over 80% of paying customers mention that they heard of us through our podcast.
Alright, that wraps up things for this week. Stay tuned for next week’s episode. Until then, take care. [01:14:30]
That’s it for this week. To sign up for our crypto investing newsletter, listen to other episodes, or get the show notes from this episode, please visit flippening.com I also invite you to check out the startup that funds this podcast, Nomics, at nomics.com. Finally, if you get value from the show, the biggest thing that you can do to help us out is to leave a five-star review with some comments and feedback on iTunes, Stitcher, or wherever you listen to podcast. Thanks for listening and see you next week. [01:15:00]