## Why don't people buy at one exchange and sell at another?

91

49

Bitcoinity.org shows that for the past 30 days, bitstamp's price is high at 101.00, low at 77.00, and current price is 98.04. On contrast, mtgox's price for the past 30 days is high at 115.00, low at 86.20, and current price is 110.50.

In other words, someone or anyone can just buy from bitstamp and sell to mtgox to make profit.

What's stopping people from doing so? What's the catch?

It can't be that people aren't aware that they could do it right?

11Because of Mt Gox's current withdrawal problems, a dollar at Mt Gox is worth less than a dollar at Bitstamp. – David Schwartz – 2013-08-15T19:47:56.017

@DavidSchwartz, what problems do you mean? – Pacerier – 2013-08-15T19:49:46.890

1

Google "Mt Gox withdrawal" to see various reports of difficulty withdrawing USD and EUR from Gox.

– David Schwartz – 2013-08-15T19:57:13.737

3http://bitcoin-analytics.com/#arbitrage provides realtime arbitrage panel which supports history of arbitrage opportunites for last 24h between all major markets. – Yevgen Martynov – 2013-08-16T14:31:05.063

76

Many people already do this with bitcoin. In finance, this is called arbitrage trading, or simply arbitrage, sometimes even abbreviated arb.

The reason for the price differences are fees for transferring between the bitcoin exchanges (you have to transfer both, bitcoins and fiat currency for a complete cycle) and fees for trading bitcoins against fiat currencies.

Because of the arbitrage traders, who make profits as long as the price difference is big enough, the prices on different exchanges quickly adjust to be right at the border between where the fees eat up the earning and where you make profits.

There is an excellent page, http://bitcoin-analytics.com , which can tell you where there currently are arbitrage oppurtunities. Simply go to the website and click on "Arbitrage".

1Do you mean that for one whole month no one bothered to do arbitrage trading even when the spread is 10% (quite high)? – Pacerier – 2013-08-15T15:32:50.273

2It's not about the spread alone. As explained, a lot of your profits from the spread get taken away from the fees of transfers and trading. Make your own calculations of the fees and I bet you will end up somewhere at 10%. This seems high, but remember that you have to transfer the amount multiple times between exchanges to make a complete roundtrip. So if you have e.g. 2.5% fees for each trade and for each transfer on avergage, this nicely lines up with the 10% total. – Daniel S. – 2013-08-15T15:38:49.627

2Example: transfer fiat to bitstamp, buy at 100, get bitcoin out of bitstamp, transfer bitcoins to mtgox, sell at 110, get fiat out of mtgox. these are 6 steps for a roundtrip. you have utilized the spread of 10% only once, but for more than one of the steps you have to pay a percentage on the full amount. Btw, if you find a nice table with the exact fees outlined for a roundtrip, let me know. – Daniel S. – 2013-08-15T15:40:42.337

4@DanielS. : Normally all these steps add up to much less than 10%. However, right now it's simply virtually impossible to withdraw funds from Mtgox. – Meni Rosenfeld – 2013-08-16T14:11:10.443

2When I used to do arbitrage I built such tables for myself. The problem is that such tables are personalized. Your exchange commission is based on your volume and your transfer fees depend on your bank and/or payment system. Also I heard that professional traders include their risks into prices (e.g. "As I make mistakes or there are various delays I need to make at least X% on gox-bitstamp route to have it profitable on average"). So such table cannot exist or at least it will have limited applicability. – nponeccop – 2013-12-20T08:10:46.610

40

I've writen software to arbitrage on some US exchanges.

I couldn't arbitrage without software because:

1. It was hard to account for all fees to understand if an opportunity is profitable.
2. It took a couple of minutes to evaluate opportunities (query an exchange's order book, query another exchange's order book, do an evaluation, execute a sell, execute a buy) and by that point the evaluation may be invalid.

Here are the factors that make it hard for me to arbitrage with software:

1. Small (fractions of a percent) profit margins. As pointed out earlier everyone in the loop is taking fees.
2. Small (seconds to minutes) profitable time windows
3. Slow (2-8 days) cycle times to move USD back into position for sale
4. Large Capital requirements - since the margins are so small for arbitrage to be worth it on an annual basis takes a lot of assets. For example I might make a few cents or dollars on a $1K sale and purchase of Bitcoin. 5. Risk - with large capital requirements comes risk of loss. What if a trade becomes unprofitable before the arbitrage completes? What if Bitcoin tanks? I am interested in non-US exchanges since the spreads are bigger but I expect it will increase some of the issues above, especially fiat cycle time since my base is USD 23 What's stopping people from doing so? What's the catch? The catch is that few people have been able to get US dollars out of Mt. Gox since June 20, 2013, when Mt. Gox imposed a "hiatus" on US dollar withdrawals. Mt. Gox has a long list of excuses for not paying their debts, which you can find on their site. Some of their excuses strain credulity. This has stopped arbitrage, which normally uses the same money going round and round from exchange to exchange. Mt. Gox is still paying out Bitcoins on demand, most of the time. So people with US dollars in a Mt. Gox account are paying a premium to convert them to Bitcoins, get them off Mt. Gox, and sell them on another Bitcoin exchange. Right now, the premium is about 15%, and it's been up to 20% briefly. It seems to be increasing about 5% per week. This reflects the market's collective opinion of whether funds deposited on Mt. Gox will be lost. http://bitcoin-analytics.com has paid tools for tracking the spread. 8 One of the major reasons standing in the way of profiting from arbitrage opportunities has to do with "volume". The volume for either exchanges is not high enough yet to support big trades. Large profits require large trades (in arbitrage). Making$1,000 may be feasible between exchanges, but that's an extreme best case scenario and using 10k in capital! Moving trades upwards of 100k would be just plain stupid if even possible in such a low volume market. Chances are if your arbitrage trade is big enough to profit from, it will move the market against you.

Arbitrage in the finance world is done with millions and millions of dollars and for fractions of a second. That just isn't possible with Bitcoin.

6

Something I am currently looking into, and a huge issue that nobody has really mentioned earlier is blockchain time.

1) Bitcoin is traded on a number of exchanges, however the entire process of buying on one exchange, transferring to another exchange and selling on that second exchange can take on the order of magnitude of 30 minutes(~10 minutes per block verification) if not more. With the volatility of the market, I would definitely not want to hinge a bet that the arbitrage window would still be open that long after it becomes present.

2) There are cryptocurrencies that have shorter verification times, but very few trade on multiple exchanges, and even fewer do so on exchanges that have APIs available. Litecoin is one of the few coins that does trade quite widely, be it still has a verification time of two and a half minutes.

2

I have been wondering about it myself for a long time now. I even thought of developing trading software (bot) that would automatically buy on one exchange and sell on MtGox. The price difference between BTC-e and MtGox can be as high as $100/BTC. The problem is how to get your cash (US$) out of MtGox? Last time I tried, I waited - in vain - for almost 5 weeks and finally created a support ticket to cancel the wire transfer, bought BTCs at the inflated price and transferred them to CoinBase to cash out. The only way I see to make some profit is this: BTC price fluctuates wildly every second of every hour of every day. So

1. buy low at BTC-e (or CoinBase) or any other exchange, say you buy 5 BTCs for $800 each =$4K
2. Transfer to MtGox and sell for $900 each, Now you have$4500 cash sitting in your MtGox account that you cannot get out.
3. Wait until BTC price on MtGox drops to BELOW $800, BUY. Say, for example, at$750, you buy 6 BTCs for the $4500 you have. 4. Transfer back to the exchange you started at (or to CoinBase, where ACH withdrawals to US bank accounts take only 48 hours). Now you've completed a round trip with 1 BTC profit (less transfer fees, which actually should not be more than$100 total).
5. Wait until BTC price is $800 again at this exchange (your starting point), cash-out, or repeat the process. But you see where the problem is. I am assuming the BTC price will swing wildly between$750 and $900 in few short days. What if it doesn't? what if it drops in value and stays below your starting price of$800 for a long time, or keeps losing value? But then again, it's a risk inherit in any speculative trade. And Bitcoin trading is all speculation.

1

there are many similar question on Bitcoin.SE (search for "arbitrage" to read more).

I'll try to resume what I understood until today on this topic (I'm quite interested into it, mee too).

There is room to play some arbitrage buying and selling in different exchanges market, but the real problem is the speed of the transactions and keeping your dollars into all you accounts. You need to keep some dollars and some bitcoins in every market you choose to use (mtgox and few others) and buy/sell keeping an eye on the different prices. The problem is when you want to trasfer some bitcoin froma a market to another (it takes time! and some fees) and when you want transfert dollars in/out a market (it takes fees!!). You should check the transfert fees when you move your dollars and count if it is still valuable...

see also this question and its answers: How can one find how much "room for arbitrage" is there in the Bitcoin market?

0

I can tell you also another thing; the high profit % is due to the temporary closure of bitinstant, and that speed up the process...

This graphic represents the % of profit between 27/1 and 27/6 2013 (6h timeframe) (source: http://www.wolframalpha.com/share/clip?f=d41d8cd98f00b204e9800998ecf8427ecn3c6i81m3)

You can simply reconstruct the events when bitinstant was closed or under maintenence by the graphic.

These are the steps:

1. fee of the bank transfer between your bank and bitstamp (sepa or international, this depends also by your banks)
2. 0.48% fee of bitstamp to buy BTCs
3. minimal fee to send to mtgox+time to verify the transaction (6 confirmations, which can take almost an hour)
4. now a 0.6% fee on mtgox to sell the BTCs
5. and a 1% withdrawal fee+time to receive the money (normally up to 2 weeks, now up to 3!)

ps. I just earned a new battery for my cellphone with this process.

-2

Its not possible due to various settlement mechanisms.Each exchange have their own settlement period and their transaction and transfer fees.

3This is (1) wrong, and (2) pretty hand-wavy. You might include hard numbers and specifics to show how you arrived at your conclusion to avoid more downvotes. – Ben Collins – 2014-01-08T22:15:53.777