working on credium.io
Cryptocurrency trading and the use of bots has seen a surge in the past couple of years but along with this has been the rise of related scams and misinformation campaigns.
I really want to clear up what is possible with cryptocurrency trading bots and what isn’t in order to help create a single resource for truth.
I know what you’re thinking — what does this guy know about cryptos?
Did I invest my life savings into Bitcoin in 2009?
Am I a crypto billionaire?
Do I have an enviable track record trading volatile assets at multiple Tier-1 US banks?
My investing journey started on the trading floor of JP Morgan in 2009 but it wasn’t until 8 years later when I started building algorithms to trade Crypto’s so I feel I’m at least a little bit qualified to talk about this subject.
Besides, I am building a crypto trading bot platform so I better be :)
No Leverage Needed
Leverage does more harm than good. The idea that you find 3% excess returns and lever it up 10 times is lunacy.
No strategy, however consistent, is infallible.
With high rates of leverage, events that haven’t been modelled can wipe out your capital completely.
The daily range on Cryptocurrencies is pretty huge already. You don’t need leverage to generate decent returns.
As you can see below for Bitcoin there are many days when the peak to trough (high to low) range is as much as 15% with the average daily range sitting at 5.5% over the past 5 years.
The space has new alt coins appearing and new developments every week. Getting involved at this stage is like having front row seats to the newest rollercoaster at the theme park.
Cryptocurrency adoption is growing.
I could go on.
Tailwinds of adoption vs headwinds of regulation make an interesting trading environment.
Less institutional money than other asset classes
I know everyone in the crypto space is hyper-paranoid about what the “whales” are up to but the reality is their trading activity is generally additive to the space (whales have obviously been net-buyers over the past 10 years).
You should be a lot more worried when Hedge Funds actively trade the space. Short attacks from them have caused stock prices to move as much as 65% in some cases.
Unlike company valuations, there’s no unifying model for how to value a coin. Stocks can be valued by NPV or book value. There are many suggestions for how to value a coin here here and here (none of which would have likely granted you any profits last year given the entire market fell).
This creates opportunity.
Simply put, an asset class that’s been around for just under 10 years is going to present more mispricings than one thats been around for 100 years.
Potential for greater performance. The best performing trading algorithms have a performance that’s significantly greater than the best performing humans. It’s worth noting here though that the majority of crypto trading bots available to the public lose money.
Automation. Bots can automatically execute trades when a particular level has been reached at any time of the day. They react with predictability and logic and when it comes to probability theory and statistics, this approach tends to produce more favourable results over time.
Emotionless. Bots can analyse data without emotion or bias. They understand the implications for a wide range of events and typically wont get wed to their positions.
A.I. A.I. & Machine Learning is beating humans across the board, from Go to Poker. We’re right at the beginning of the A.I. age with the advent of self-driving cars and healthcare diagnoses. Credium’s crypto trading bots utilise A.I. extensively.
Scale. Bots are able to process significantly more data and in much shorter timeframes than humans. This means they won’t miss an important announcement on a portfolio company. It also means they can analyse more companies and thus diversify risk better.
Speed. Speed reduces slippage as reaction time isn’t needed and opens up arbitrage opportunities too.
Longevity. Bots can be up and running 24/7. In a product like cryptocurrencies this is especially useful. Humans need to eat, sleep, live. A human typically works 40 hours a week however a bot can work 168 hours a week therefore can capture 4.2x more opportunities.
Consistency. Bots by design are consistent. When they engage in a piece of information or an announcement they’ll react in the same way over and over again.
Strategy Diversity. Because bots are able to scale and run at speed, they can also work on different time frames and with different strategies that are uncorrelated which reduces their risk. This is because the P&L volatility is lower which would cause your risk adjusted returns to be better.
In-built backtesting. Every strategy can be backtested multiple times and in multiple environments automatically. This gives you more confidence that the particular strategy works and when you should stick with it or not. This also helps you manage risks better.
This is a chart of the daily volatility of Bitcoin since inception.
Whilst as you can see, Bitcoin volatility has been going down over time, it’s still significantly higher than, say, Gold. The cryptocurrency space can quite easily drop 30–40% in a single day.
Volatility coupled with human emotions is bad for returns.
Wealth is created from low volatility and consistent compounding.
This is a chart of peak drawdown per year for Bitcoin. As in, how much would you have lost from peak to trough if you just bought and held.
As you can see in both 2015 and 2018 you had a greater than 80% drawdown. Across the timeframe (last 6 years) you had a mean drawdowns of 50% per year.
I don’t know about you but I’ll never be in a position where I’m comfortable losing over 80% of my investment portfolio.
It’s all well and good saying just buy and HODL but the problem is: negative returns hurt.
No doubt you’ve seen this chart before.
This is the forecast chart of Bitcoin using seasonality and halving events.
I think this is the chart that guys like Mcafee are using for their very detailed scientific forward valuation models when he claims he’ll eat his d1ck if we don’t get there.
Point really is that if this trend were to continue, we are literally forecasting periods of pull-backs in future. This is a log chart so those pull-backs are 70–80% each.
Probably worth noting here that given Bitcoin is limited to 21 million coins, this chart is forecasting Bitcoin sometime in the year 2029 with be worth:
21,000,000 x $1,300,000 = $27,300,000,000,000
thats $27.3 trillion.
In 10 years.
When the total global money supply right now is about $5 trillion (even modelling 3% a year inflation for 10 years would take us to $6.7 trillion).
Regardless of the direction of Bitcoin and other cryptocurrencies, ideally we’d like to be in a position where we can capture most of the upside and avoid most of the pull-backs. In this scenario our risk-adjusted returns will naturally be higher. A well designed crypto trading bot will do just this.
Since 1970, Gold has growth at a CAGR of around 7.8%/year. ($37 in Feb 1970 to $1497 in Aug 2019).
Assume the same growth for BTC then we’ll hit $100,000 in about 29 years.
($11,376 x 1.078²⁹ = $100,446)
But you’re assuming:
BTC is still around.
Why would BTC not be around?
Simple. Regulation or replaced by something better.
We’re not in some sort of bubble already.
Why would Bitcoin be in a bubble?
You see, Bitcoin has actually grown by about 510% per year. Oct 12 2019, 1BTC = $0.0009 vs $11,376 today. If 510% CAGR for 10 years isn’t a bubble then I don’t know what is.
Supply doesn’t increase drastically.
How can supply increase drastically?
Not supply of Bitcoin per ce but more increased supply of newer, better Altcoins and then Bitcoin investors rotate. Unlike private bankers with IPO’s, there’s no one stopping unlimited ICO’s popping up.
0. Reduced Fees
Most exchanges knock your fees down by 60–100% if you’re trading north of $50m per month. This is huge in itself and drastically increases profitability of any strategy. That’s a volume most people won’t be able to achieve by themselves.
0. Multiple instances of an algorithm aren’t fighting against each other
Imagine a rudimentary market maker algorithm where it always improves it’s price to sit on the top of a queue. Having 10 instances of this on the exchange can be pretty dangerous if they’re all improving to get to the top of the queue.
This is a pretty basic example but usually a bot will be optimised to run as a single instance and the logic breaks down when you run multiple instances of the exact same bot.
0. Reduced infrastructure costs
Getting a decent server (fast and near to the exchange) is expensive. Having to only pay this once is a bit more efficient than every client paying for the exact same thing.
0. Protection from bad actors
If anyone knows your trades they can reverse engineer your algorithm. And if they reverse engineer your algorithm they can either front-run all your trades or they can bully the price to force your algo to puke. Either way, your algorithms end up not being profitable.
0. Protection of IP
Once again, if you have actually have found some Alpha with your algorithms, if someone knows all of your trades they can reverse engineer your algorithm, and just run more size. Before long the Alpha your bot found will disappear.
0. Aligned interests
Bot developers can give you a basic version of their algo then just make a better version and run it themselves. The newer version will execute trades faster than your version so you’ll always help their profitability (whether or not the algorithm was profitable in the first place!).
This is basically the equivalent of UK/AUS retail brokers that trade against their clients. The end consequence is that you wont consistently make money (about 70–75% of retail broker clients lose money).
0. Fair playing field
Obviously whoever executes an algorithm first is going to have asymmetrically higher returns. With centralised trading, someone with deeper pockets won’t be able to buy the same bot you’re running then pay to get nearer the exchange (therefore beating your shared server version to trades).
0. Less market impact
Trading more than $500k of Bitcoin in one go has a notable price impact. Running multiple instances of the same bot means that you can’t drip an order through the market so as you start to scale, every trade will have a market impact. Ideally your trading activity shouldn’t have a market impact so you can scale your strategy if you wanted to.
Make sure it doesn’t use technical analysis as a signal
Technical analysis (TA) doesn’t work. Period.
You can overfit on TA indicators to make it look like a combination of indicators works but it won’t hold in a test set.
Make sure no one can get hold of the bot’s trades
As we discussed above, trades can be reverse-engineered by bad actors. This literally happens every day in regulated capital markets so don’t hope that it won’t happen in an unregulated market.
Make sure you and the bot developer have aligned interests
If the bot developer charges you a fixed fee and has no responsibility or accountability over your profitability, then just like retail brokers in UK/AUS, they can trade against you to make even more money.
Make sure they’ve invested institutional capital in the past
Unless they have some experience in financial markets it’s unlikely they’ll ever have a deep understanding as to what impacts the price of a particular asset. If thats the case they will likely be spending your cash on their “education” until they get it right.
Make sure they don’t claim their bots produce passive income
It’s best to steer clear of anyone claiming “passive income” from investing. There are no guarantees in financial markets, especially not Cryptocurrencies.
Furthermore, withdrawing monthly from an investment fund usually kills your portfolio, withdrawing partial profits does not.
There are loads of examples of this if you look at retirement portfolio’s but the lesson is that if you have the possibility of negative returns (which you do with all investing) then you definitely don’t want to take regular withdrawals. I talk about this here too around half-way down.
Withdrawing partial profits is fine because you’re not affecting your original pot and you’re still compounding some of the returns.
Credium is a crypto trading bot platform that was built on top of proprietary trading algorithms developed in-house with the sole goal of achieving client profitability.
Our trading bots were designed to be “set and forget”, so no tweaking or fiddling needed to get up and running. This is ideal for beginners or experts who want a completely hands-off approach.
Asymmetric wealth creation. I believe that retail clients should have a method by which they can have high returns as the best Hedge Funds are typically only open to institutions and HNWIs only.
Sub-standard platforms. Giving sub-standard tools to retail and charging them for it isn’t something I agree with.
Crowdsourced models don’t work. The “crowdsourced hedge fund” experiment doesn’t seem to be going as planned with overfitting prevalent amongst strategies. Besides, the exploitation of data scientists through Kaggle competitions is not something we believe in either.
We develop in-house proprietary trading models whilst giving investors full control over their investment through our platform.
If this is something that’s interesting to you, feel free to join us.
Disclaimer. Financial markets are inherently risky. Unlike stocks which have intrinsic value and are correlated with the global economy, cryptocurrencies unfortunately aren’t productive assets and they have no intrinsic value. Therefore you should be very conservative if you plan on getting involved with cryptocurrencies; the risks are pretty huge. The entire space could in theory be made illegal by governments which would cause cryptocurrencies to lose a lot of their value. Using cryptocurrency trading bots compounds the risks even further. This post isn’t intended as investment advice so please do your own research before investing in anything.
Create your free account to unlock your custom reading experience.