This is a nicely designed site and some good data feeds. There is a big difference between spread, and arbitrage. Critically, any attempt to track arbitage needs to watch the orderbook rather than the last-trade-price. Ironically, the bid/ask spread on the same exchange isnt part of arbitrage. The difference between the total available bids on exchange A and the total available asks on exchange B (or vice-versa), is the opportunity.
I wrote and ran an arbitrage tracking site for a few years and eventually stopped because there did not seem to be viable way to take advantage of cross-exchange spreads. Sell-Move-Buy didn't work because moving coins took days. Having a pile of each kind of coin on each change is fast but defeats the point of arbitrage because the pile itself it subject to the price changes that arbitrage is made to avoid.
The chart below shows the result of looking at the orderbook of different BTC exchanges, seeing how much was actually offered at a winning price on other changes, computing the total, and taking into account exchange fees (more or less)
I always thought the code could be generalized enough to look at any number of exchanges and coinpairs. I'll keep an eye on tokenspread, it looks like a good start!
Agreed. I had run a small BTC arb program for a bit a number of years back. The exchanges were shoddy, there was a persistent bias in the arbitrage such that you ended up accumulating huge positions of BTC on one side, and lots of short cash on the other. Couple that with the difficulties of withdrawing cash from the system, fees, etc., it was not worth the risk. But this is not new.
What's pretty cool about this -- though it appears this site is not focused on it -- is the opportunity for full triangle arbitrages among the various crypto's, such that you may do, e.g., a BTC->ETH->LTC->BTC trade and wind up with more BTC than you started with.
Another point to consider; the big players have moved into this space, if you want to get into it you will need to be sure you're not at an informational disadvantage.
I created a script years ago that looked for arbitrage opportunities on BTC-E, up to 4 trades deep, and it used the order books properly. I left it running, and was surprised that it actually found some arbitrage opportunities, around once every 3-4 minutes. Unfortunately the profits were always lower than 3-4 transaction fees of 0.2% each, so my dreams of getting rich quick got crushed.
I'm was actually surprised BTC-E didn't do something like that themselves, since the they don't have to pay any transaction fees, and they also had zero lag. It's like free money.
Crypto arbitrage always seems so tempting, but there's so many gotchas. This site seems to do a good job of weeding out some of the gotchas by focusing on coins with high market caps.
Usually the "opportunities" in arbitrage that sites like this highlight ignore the fact that the targetted coin on one of the exchanges has abysmally low volume, so you'll be unlikely to be able to buy and sell it in time to take advantage.
Other gotchas include: time to transfer the coin between exchanges, fees involved (both in transfer and on the exchanges through buy/sell fees), and exchanges where deposits and withdrawals aren't fully automated (some Chinese exchanges were like this).
I think you can mitigate a lot of the time and volatility risk by trading on margin, i.e. buy coin on one side and sell on the other. You can still get called on either side but the risk should be smaller.
> Usually the "opportunities" in arbitrage that sites like this highlight ignore the fact that the targetted coin on one of the exchanges has abysmally low volume, so you'll be unlikely to be able to buy and sell it in time to take advantage.
I don't understand. If there's an order in the order book on the sell-side of one exchange that matches one on the buy-side of another exchange, what does volume on either exchange have to do with the ability to take advantage of this situation?
The more liquidity there is, the higher the chance that someone will jump in the place of another order that has been withdrawn. Low liquidity and volume means that such chance is lower.
Yes, it seems almost shady/misleading that this site doesn't include the Bitcoin/cryptocoin transaction fee and the exchange fees, which are all available publicly.
Exchange transaction volume should also be fairly readily accessible, but I'm not sure about that. This could be used to calcualate a probability of success to also take into consideration.
Even the various World of Warcraft auction house arbitrage add-ons take this type of stuff into account.
I do not disagree with your point. Is this a reason to leave it out?
It feels like talking to a stock broker / investment advisor who gives all the numbers while leaving out any mention of their cut. In this case the Token Spread site doesn't seem to get anything out of it, unless they somehow tie into exchange affiliate programs wherever they link them up.
Given the efficiency of information exchange over the internet regarding btc prices on numerous exchanges, I'd venture to guess that these arbitrages only exist because of market barriers that average investors cannot efficiently surmount.
Example: a few years ago there was a difference of a few hundred dollars between U.S. and Chinese exchanges. But you couldn't really trade on the Chinese exchange unless you were or knew someone who was a Chinese citizen.
I haven't researched it, but my impression is that arbitrage is only indicative of market inneficiency. And a corrolary is that by the time you get into a position to take advantage of significant arbitrage, someone will likely have beaten you to it.
In other words, efficiency is antithetical to arbitrage. I welcome alternative perspectives.
A hundred-dollar bill is lying on the ground. An economist walks past it. A friend asks: "Didn't you see the money there?" The economist replies: "I thought I saw something, but I must've imagined it. If there had been $100 on the ground, someone would've picked it up."
There were public order books years ago and there were still arb opportunities because the field was niche and people just hadn't written the software yet. In terms of your comment, the technical barrier was what "average investors cannot efficiently surmount." At this point the high-volume pairs are too crowded but I suspect that there are still opportunities to make money where technical complexity is still high (e.g. smart contracts).
Pretty cool site, thanks for showing it. Are you plugged into each exchange's API or using an aggregator?
Taker fees should be shown on all of these potential cross exchange arbs. Exchange withdrawal fees should be presented somewhere as well.
For those looking at this site and seeing the percentages and thinking it's easy money, it isn't. Arbing has become a pretty crowded space and this kind of low hanging fruit is very hard to capture even when fully automated, not to mention the banking logistical issues of using USD (not tether) as a base pair.
If building trading systems in the crypto world is something that interests you feel free to reach out to me, company / contact info is in my profile.
You don't need to be a taker to run arb. You'll be able to push the spread on a side of each exchange, and work with quantities of BTC on each one that get replenished.
Ok sure, show both then, the maker rebate on some exchanges will open up more opportunities.
All I want is just one "arb opportunity dashboard" to show the whole picture so people don't think they can deposit some btc and start clicking away generating a profit...
To be perfectly honest, not trying to be snarky, both this comment and the article it links to seem like they are written by AI. To me it's buzz-word laden gibberish, I can't make any sense of it.
I wrote and ran an arbitrage tracking site for a few years and eventually stopped because there did not seem to be viable way to take advantage of cross-exchange spreads. Sell-Move-Buy didn't work because moving coins took days. Having a pile of each kind of coin on each change is fast but defeats the point of arbitrage because the pile itself it subject to the price changes that arbitrage is made to avoid.
The chart below shows the result of looking at the orderbook of different BTC exchanges, seeing how much was actually offered at a winning price on other changes, computing the total, and taking into account exchange fees (more or less)
At 10:10pm there is a $16USD spike - probably between BTC-E and Gemini that lasted a short while (refresh rate was every 10 min) https://web.archive.org/web/20160219100013/http://cointhink....
I always thought the code could be generalized enough to look at any number of exchanges and coinpairs. I'll keep an eye on tokenspread, it looks like a good start!