The Bancor’s Sinks Or How To Lose 4600 Dollars in a Single TransactionHi all,
First of all, I have been a strong supporter of the Bancor project since its beginning in June 2017. The Bancor team has indeed conducted a very good job by developing their partnerships and launching new features every month. However, I wanted to share my recent and unfortunate experience using the new Bancor Webapp.
On January 15th, I decided to jump on the opportunity of seeing the Binance Coin (BNB) joining the Bancor Network through a new Token Relay that connect BNB with BNT (Bancor’s token). I had previously bought 3 206.27 BNT (28 400 dollars based on the January 15th average exchange rate). That first transaction went very well through the Bancor Webapp. I indeed converted my ethers (ETH) into BNT in a simple click for only 8 dollars of gas price. At that time, my 23 ETH I had converted into BNT were still worthing 23 ETH once converted into 3206,27 BNT a few seconds after. Back on January 15th, I used the same process to have my total BNT balance converted into BNB. I was confident that the webapp would run perfectly well and the exchange rate announced on the Bancor webapp confirmation screen was matching the market based rate of the BNB/ETH on the Binance exchange website. I thought that it would be more efficient to use my BNT to buy BNB rather than going through an exchange. Indeed, it is the fundamental purpose of the Bancor Network to allow any smart Token to be converted into any other Token connected through Token Relays and the BNT connector. I proceeded the transaction after checking the advanced settings with a BNB default price limit of 5% which I confirmed.
Once the transaction confirmed on etherscan.io, I checked my new balance of 1 215.37 BNB in exchange for my 3 206.27 BNT. At that moment, I was really amazed by how fast and how easy it was to convert my BNT tokens into another Smart Token thanks to the ‘magic’ of the Bancor Protocol. BUT I checked how much my 1215.37 BNB worthed based on the current market price… only 23 830 dollars ! It took me like 30 seconds to accept that idea that my 28 400 dollars of BNT became 23 830 dollars once converted into BNB. I just lost 4 600 dollars in a single transaction (16,2% of the initial amount !) …
I can tell you that was of course extremely painful. I am a student that invested a good part of his savings and I just lost 4600 dollars not on a market dump, but due to the Bancor Protocol.
Here is my transaction concerned on etherscan.io:
https://etherscan.io/tx/0x81e18be561a076f392ec10bbd309ac6647f54df6a0a7636c3e4e180568a2e829I spent hours understanding how I just ended up with that loss. My first thought was to check how it would cost me to have my 1 215.37 BNB converted back into BNT through the same Bancor webapp. I checked again the exchange rate proposed by the webapp. It seemed again OK to me compared to the Binance market price. But when I typed the amount of BNB I wanted to transfer, I then saw that the exchange rate changed a bit. However, that small price change was enough to lose another 3500 dollars in the transaction. If I had proceeded the second transaction, I would have lost a total amount of roughly 7 100 dollars (based on the January 15th market prices)… The worst transaction cost you could ever imagine.
It took me a while to investigate on my own and understand how the Bancor Procotol made me lose 4 600 dollars and how I escaped from suffering another 3 500 dollars loss.
Actually, the Bancor team is releasing every week new partnerships and Token Relays to develop the Bancor network. That is indeed a good thing to sustain the project on the long run. However, when a brand new Token Relay is put online, the Bancor Protocol price discovery is not reacting in the same way that we can experience it when we trade BNT with the same Bancor Procotol. In the situation of a new Token Relay smart contract, the “connectors”, which means the Tokens balance backing the Token Relay is at a far lower level that the amount of ETH backing the BNT. When you proceed a transaction, the more token you send to the new Token Relay smart contract and the more the exchange rate is rising up sharply in one single transaction. The price volatility in one single transaction is barely seen when people are sending for example ETH to get BNT because the amount of ETH backing the BNT is huge (currently 45 000 ETH ). It would require a huge amount of BNT or ETH to have the same price volatility in one transaction.
Maybe all that mechanism is very familiar when you read the Bancor whitepaper and you thoroughly understood the deep implications of the Bancor Protocol. But even for an enlightened user who got the Big idea from the Bancor project, you can easily be fooled by the Bancor webapp.
My unfortunate experience made me remember a blog post from the Ethereum founder, Vitalik Buterin, tackling the token price discovery issue (
https://vitalik.ca/general/2017/10/17/moe.html). According to him, if a token does not have any sinks, meaning a way to hold a part of the Token total balance in order to make it disappear, the Token value may be extremely unstable. The Bancor Protocol is using the connector (previously called Reserve Token) as a sink. However, the Bancor type of sink is much more pernicious than any other type of sinks. The Binance team is for example cash burning every quarter a part of the total BNB supply to keep the token value rising up. In the case of the Bancor Protocol, we can find the equivalent of the cash burn in its price discovery mechanism.
If you send for example 3000 BNT to get BNB tokens, the Bancor Protocol will divide the 3000 BNT balance in an infinite quantity of transactions. It means that each of those transaction will be traded at a slightly different price. On the buyer side, it means that you will get in each transaction less BNB for 1 BNT. And the less the Token reserve balance of the smartcontract is important, the less you will get BNB for each BNT you are selling in my example.
The Bancor sinks are the cost of a constant token liquidity. There is no magic solution. It is a just a way to deal with the Token valuation issue and the double coincidence of wants.
My main feedback on the Bancor Protocol and webapp :
- The Bancor Protocol clearly disincentivizes people from trading big amounts of tokens relatively to the Token Reserve/connector balance. It also disincentivizes Bancor users from exchanging different tokens on a daily basis to limit their total loss due to the Bancor sinks taxing each of their transactions. It is maybe another explanation of the poor BNT performance since its last ICO in last June ?
- The Bancor webapp should CLEARLY warn the user when the exchange rate volatility is very high to avoid any huge loss compared to the current market price (specifically if the token is traded on a public exchange).
- The Bancor webapp user should be warned that the default settings for the price limit in the Bancor webapp is not the real one. The default price limit is 5% in the webapp. But the price considered for these settings is the final price, AFTER the Bancor Protocol impacted it. Indeed, the price computed by the Bancor Protocol is based on the amount of Token you want to trade, the Reserve Token balance or "connector" and the Token Relay balance. You cannot put a limit price on it by definition. The webapp only allow you to fix a price limit for the price already processed by the Bancor Protocol. It is indeed another way to be fooled by the webapp...And it made me lose 4600 dollars. Because if it was a real 5% limit price like on an exchange, I would not have lost all that money (16% of the initial amount !)
- The webapp is not transparent about all the different factors that can impact the exchange rate of your transaction. If a common user has to fight to get all the information needed (Reserve Token balance, Token Relay Balance etc.), have the ability to correctly understand it and not directly rely on the 5% default price limit in the webapp settings, all that is not in favor of the Bancor network democratization that the Bancor team is claiming to look for.
- The Bancor team should create a real typology of Tokens connected to the Bancor network. It is indeed important to differentiate the tokens that are already traded on crypto exchanges from the one which are only available using the Bancor Protocol. The Webapp users would see easily if they can accept to lose money from the Bancor sinks (the cost of the liquidity) because the tokens are not or rarely traded anywhere else. It would be like an arbitrage facilitator feature in the webapp.
- The Bancor project is aiming at emerging the long tail of the local currencies/tokens. In the white paper, they compare it with Youtube that enabled millions of people to share their video production to the rest of the world. The comparison is interesting but they seemed to forget that the Bancor sinks are actually another hidden cost that can be very painful for poorly funded smartcontract’s Tokens reserve. And the thousands of local tokens in the cryptoworld are by definition poorly funded, which implies a huge price volatility with inflated transactions costs if they are traded with the Bancor Protocol. The Bancor Protocol mastered the liquidity and the double coincidence of wants issues but not the transaction cost like Youtube did for its users. The entry barriers are still there.
- I contacted the Bancor team right after my unfortunate experience on January 15th. They answered me back the following day writing “First of all, thank you for your support! Let us look into this and we will get back to you as soon as possible.” It seems that they did not find it interesting because they never answered me back, even after I emailed them again multiple times on that issue… I lost several months of intern compensation, I just desired at least to have their own opinion on these Bancor sinks to help them on their project. Very sad.
Sorry for the long post, I had a lot to share. I hope it will help the Bancor Project to really take off and scale.
Why would you have to apologize for your long post? We should be thankful to your long post since it has outlined a problem that nobody is probably aware of, so your experience can be very helpful to many others. I think it would be in the best interest of the Bancor project if the Bancor team would become aware of such problems existing in connection with their protocol. If the "discovery of the price" mechanism is discovering a price which is so much different from the market price, I'd say that there could be a little problem.
