The Depths it Takes To Make A Thriving Cryptocurrency Exchange

Please welcome Oleg Belousov to Hacker Noon! Oleg is the CEO of n.exchange, the easiest way to exchange cryptocurrency, for individuals and companies alike. n.exchange is a past weekly sponsor of Hacker Noon, and I think his experience through the crypto boom is very valuable to our readers.

David: Can you tell me a little bit about n.exchange — its scale and its progress?

Oleg: I founded the company two years ago, and we had quite a few pivots along the way. As an employee of Rocket Internet previously, where I was a senior software engineer, I decided to implement what I had learned there. Initially, I tried to implement an existing business model, and we’re talking late 2015, early 2016 here, to a basically new and unexplored market. My idea was to implement a Kraken-like, but more instant, bitcoin exchange. Essentially a consumer-focused bitcoin exchange interaction CIS. The reason for that was, at the time I was trading it privately via localbitcoins, and I saw that there were larger windows of opportunities in Russia. It was 5%, even higher sometimes on weekends. Whenever there is an arbitrage opportunity, it means that it’s just easy enough to buy cryptocurrency there, right?

Between that and the currency being high yield, I decided to start my own exchange.

What I tried to do initially was to open a physical location in Moscow for a physical exchange. Then I met some banks, to try to open a corporate account. I formed a company there.

David: What did the physical location look like?

I actually have an article about that. It was just an office with an iMac, and people would come and trade. As you can understand, it wasn’t readily available, because all I had was one or two representatives, and we were limited by the physical factors. Opening hours, and basically the amount we could process, and the amount operators can process and handle simultaneously. It was more of a PR move to basically show our faces to people, to encourage people to trust us. We tried to do this Russian operation, with Russian processing systems, but there were two factors.

First of all, Russia is a difficult place to do business, and secondly, the market wasn’t ready yet.

The banks, they were kind of agnostic about it. They said, “Oh, we don’t know what it is. We cannot be doing that.” Because they don’t recognize it as a means of monetary remittance. And so eventually we had to get out of there, and we moved to credit card processing. And again the market was still in its diapers. We tried to partner up with Simplex, but they were too busy, and it didn’t work out. We ended up working with the Chinese aggregator result, without an own MID (merchant ID).

It was via an aggregator, via Construction Bank of China, and after a small pilot for F&F(friends and family), they just didn’t send us a settlement payment.

Then, we had to pivot once again. We said, “Okay, the market is not ready for us yet, a small startup to offer Fiat trading solutions, so let’s try to do only crypto-to-crypto, you could say kind of like ShapeShift.” This was when we pivoted to a service-based architecture, where we offer our API separately, and we offer several products leveraging our API. This includes Crypster, which is a chatbot using our API, and sending the protocant post which we used to launch the proto as it currently is.

Then we decided to open towards the front end. So we decided to position ourselves while working with a skeleton team. We basically created a gross model, which is divided into two or three different flows. Essentially, any amount of cryptocurrency could be sent directly to their wallets, knowing the rate in advance and having an option to reserve the trade for 15 minutes from the opening of the trade until the deposit, up to 15 minutes. And this-

David: Why is the time just 15 minutes?

The time is 15 minutes because the cryptocurrency market is pretty volatile at any time, today as much as a year ago. As a financial option, we need to keep the deposit window to a reasonable length so we don’t have to take a position against our customers.

The longer the window is, the bigger the risk that the customer gets one rate, and by the time he deposits, the rate in the market is different. And so to compensate for that, we charge a markup of 0.5%.

Because we feel that if the market changes 0.5% in that time, we buy the risk and make no money in the worst case scenario. Essentially this is the longest interval we can pull out for a low enough markup, to still be competitive with the traditional solutions such as Coinbase, Bitfinex, or Kraken.

Our leap of faith, our assumption, as a lean start-up is that there will be a considerable amount of people who would pay the extra 0.5% to be allowed to make the trade automatically and non-custodially.

When we let them deposit and withdraw immediately, we save them six operations:

1. Depositing, waiting until the deposit comes through.

2. Opening a limit trade.

3. Risking the chance that the market will change until they open the new trade.

4. Waiting for the limit trade to fill (or not).

5. Adding a new withdrawal address and confirm it by email.

6. Finally, they need to make a withdrawal request and also confirm it by mail.

So we’re talking about six operations here over a timeframe of several hours depending on how busy the blockchain is. The bitcoin blockchain at some points, unless you pay a priority fee, will take several hours to confirm the transaction.

So this assumption was actually validated, and we achieved pretty hefty volumes. The biggest trade that was made on our platform was 165 bitcoins, and the way we implemented it, to accommodate the volume, was by using a third party, namely Uphold. We had a business account with them. Everything was fine. We even discussed a lower fee with their sales-level executives in the call, and everything was done transparently.But, business is business, there are thoughts and discussions all the while there is a million dollars in the bank. And at the time those 165 bitcoins went through, it was well over a million dollars. And this turned on some red lights in Uphold, which is an American company, and as such, they have to be extra careful. And so they froze our funds and our customers’ aggregated funds.

David: That must have been a drag.

That was not a nice day, but we managed to mitigate it. It took about a month using diplomacy until we managed to get the funds unlocked, and everybody paid. So it was a challenge, and we couldn’t trade during November and December 2017, which was a very hot time. It was a big loss of revenue for us as well as for Uphold, because on our good days they would make tens of thousands of dollars from us, through fees. I do understand why they did it. They are looking at the long-term and they were afraid of possible regulatory consequences, but, yeah, for us it was obviously a blow.

W went from a huge success and launching a product that was in the fifth place of product hunt organically, having around ten white labels working with us, some of them using our paid hosted services, using the support of our development team, and launching some complimentary products such as crypstor.com, leveraging our API, all the way back to square one.

We had to cease using external liquidity providers and change to a model where we hold the reserves. Now when the customer comes through and makes a purchase, they buy it out of our reserves, and we get into a position which our trading algorithms have to rebalance. So, as you understand, from volumes of hundreds of bitcoins, we shrank to volumes of several bitcoins and on a really good day several dozen bitcoins a day. And it was very sad for us.

Simultaneously we were exhibiting at WebSummit 2017. Actually, this situation happened one day after a pitch at WebSummit (not sure if there is a connection between the two, but there might be).It was another challenge, and we had to pivot and change our business strategy again, but also one of the connections we made in WebSummit was SafeCharge, and we were able to facilitate Fiat trading, and the beautiful part is that they allow us to sell any token that we find worthy, which has let us sell directly by credit card.

And so, we lost the ability to allow clients to trade 10 or 20 or 30 bitcoins in a single trade but we did gain the ability to allow Fiat trading. Several months after this situation, we figured out that we would need an order book, and since there is no liquidity provider that we could use that would suit our needs, we would probably need to develop our own solution. That is how we got the idea to make a DEX, like an off-chain HybridDEX. A decentralized exchange with an off-chain matching engine.

For an implementation proposition, as well as more details please take a look at our white-paper.

David: Once you allow credit cards in the system, it opens up a lot of possibilities to a lot more customers, but it has a much higher transaction fee. How have customers been responding to using credit cards?

The responses are mixed. Right now I think that in any market and especially in this market, a fee of 5% is quite high. However, it is still the best among our direct competitors like ShapeShift, Changelly, and ChangeNOW. And you can see it, via all the aggregators we’re listed on (Bestchange, Cryptocompare, CoinSwitch etc).

But still, I do think it’s very high and we are actively working on reducing it.

I think that in a market with high daily variance such as crypto (tens of percents in some cases), people want that ability to purchase the coins exactly at the time you want to get into the market. Otherwise, if you don’t have the money in your Coinbase account, you need to wait three days for the money to be credited and then you can trade. So I think that in some case this fee is justifiable for the customer.

Nevertheless, you see all these new solutions. In Europe, you see Sofort transfer and soon open banking and SEPA2 (or SEPA Instant) which allows immediate settlements until 12,000 euros per transaction. And in the U.S. you have automatic clearing houses (ACHs) which are sometimes immediate. I think that the banking world is moving and what we would need to do soon is to shrink our fees to keep being competitive.One of the things that we’re currently working on is integrating Amex, which is very popular in the States. And one of the main upsides to supporting credit card processing is that we can serve clients which are first-time crypto buyers or just clients who are maybe unbanked or cannot, for any reason, send the money via bank, because they live in a developing country. In any case, whether it’s urgency or a lack of the monetary tools necessary to remit money internationally, we have made it our primary goal to make easy-accessible and fast crypto purchasing available for the vast majority of the planet’s population.

But in order to provide this service, we have to take on the burden of risk and it’s quite substantial. Fraud costs us money, and more importantly, it gets our brand name stained by the individuals who fall victims to this fraud.And so, one of the main topics on our daily agenda is to fight fraud and to do it successfully and effectively.

We’ve developed a set of tools to mitigate the risk, starting from detailed KYC procedures to technological tools to help us identify suspicious transactions.

So as you can see, there are many challenges. Right now, in a micro-view, we are working to expand our credit card support.

We still had relatively zero chargebacks last year, which is phenomenal. And close to zero chargebacks this year, very few, most of which were mitigated or refunded.

As you understand, if you analyze the DMU of the user (decision-making unit — how he chooses which exchange to trade with), I believe that first decision they make is, do they trust the exchange? And so it’s important for me to emphasize that on Trustpilot, we have the highest trust rating among our direct competitors. If you look at ShapeShift, it’s pretty dim. If you look at Changelly it’s a bit better, like 5 out of 10, versus say ShapeShift’s 3 out of 10, and we are on 8 out of 10. So the customers like what we’re doing.

What we’re doing right now is working on this price issue to increase our user base. The lower your price, the more volume you will have. So if you were to sell bitcoin now at 2% or 1% under the market price, you would probably have the highest volume off all the exchanges, because people would just buy from you and sell it for 1% higher to your competitors. The thing is, with a small markup, it’s not as radical, so the break-even is somewhere near zero (plus processing fees).

If you have zero fees, people will come and take advantage of that. Unfortunately, we cannot afford zero fees in the long term, since credit card purchases are quite a costly process, and we would end up with heavy losses.

Unlike blockchain transactions, credit card transactions can be refunded and can be charged back. There could be fallback. The means that we use to ensure that there are no fallbacks are both technological and procedural.

We use the most strict 3D Authentication rules. This hurts our business a little bit, because it means that we cannot serve some users (especially in the United States_ which are not subscribed to 3D Authentication, unless there are several exceptions. The other thing which we do is, as I said, we are very thorough about our KYC.

We require a utility bill and ID from the very first dollar.For more serious amounts, starting at $1,000, we ask the user to provide a selfie with his ID and credit card with the sensitive data covered, as well as like a note that says ‘n.exchange’ and the current date.

If the user purchases more than $2,000 in less than a month, we also ask him to print out his withdrawal address and state on a piece of paper “My withdrawal address is the following.” And then you will only be able to withdraw funds which exceed the limitation to this ‘’whitelisted’’ address.

It’s a little bit of a clumsy process and when you combine that with the relatively high fees, many users are reluctant to go through the process. And so the main thing we are working on right now is that KYC process information, so users, especially small users or first-time users, could go through the process automatically without any human interaction within seconds. Just like that, flawlessly.

The second thing is the fees, so we are negotiating with our payment provider to reduce our rolling reserves and thus allow us to provide highly competitive rates such as only 3% or 4% above spot for EUR and other European currencies, and a little bit more than that for dollars, because dollars have a markup as our processor is European.

David: And what do you think the most important stats are for cryptocurrency exchanges, and whenever you look at building the KPIs out around your company? I mean, obviously, you know, how much, the total money in is very important. What other stats do you look at to help shape a healthy exchange?

I think that while turnovers are important, they’re not always real.

If you take FANG (Facebook, Amazon, Netflix, Google) you have more or less empirically measured capitalization, which is calculated by the actual bids in the NYSE order book, and some profit multiplier of the company. When you take bitcoin, you take the last bid multiplied by supply. The last bid is highly inflated. the quantitative trader comes and dumps a million dollars on any prominent exchange, he causes an immediate decline of 10%. This decline is called ‘’slippage’’.

Even though for a second, it still creates the psychological effect of a declining market.

This is called ‘’manipulation.’’

As for the supply, it cannot be entirely accounted for: some of it is locked, some people lost their keys.

The same applies to exchanges: some of them implement trading bots which trade with themselves. Some of them don’t even go that far, and they just publish fake volumes. And there was a big scandal about that with PBOC (People’s Bank Of China) and OKCoin, Huobi, and other prominent Chinese exchanges, back in 2017 when the Chinese exchanges crackdown happened.

I firmly believe that this measurement can only be taken internally, asking yourself the following question: are you cashflow positive and how much of your revenue stays on your balance sheets as profit? As is the case with any other business, it’s called ‘’the bottom line’’ for a reason.

I would like to add that we make our best efforts to not cross any moral lines and keep the business fair and straightforward.

To actually assure our users that we don’t do any manipulations, our API’s completely open, and you can see every deposit and withdrawal transaction for cryptocurrency trades.

While we don’t have a way to prove credit card deposits without compromising the user’s privacy, we still provide the withdrawal transaction details through our API for fiat purchases. So you can actually see that money came from our wallet to an external wallet of the user and this is all open for auditing by our customers and also for legal reasons. If for any reason any law authority wants to check which wallets we are sending coins to, they could see it by our API.

This is one of the things we hold most dearly: Transparency.

In short, I believe that right now we see lots of trends in cryptocurrency which are value-destroying, like tax evasion, scams, Ponzis.

I believe that for us to prevail and to create a real A-book broker and a real strong prominent blue-chip firm, we need to shift the paradigm and be value-creating. What does it mean, ‘’value-creating’’? Being transparent, auditable, preventing manipulations, paying taxes and doing all of this on a daily basis.

Be on the lookout for parts 2 and 3 of this series: How to Solve the Technical Changes in Making a Faster Cryptocurrency Exchange & What to Learn From Warren Buffett, What Not to Learn From Wells Fargo, and the Color of Japanese Candles. This is a part of the Lost Tech Interviews. Sometimes you don’t know important tech when you see it and sometimes life and tech get in the way of writing.

Connect with Oleg:

The Depths it Takes To Make A Thriving Cryptocurrency Exchange was originally published in Hacker Noon on Medium, where people are continuing the conversation by highlighting and responding to this story.