Hacker News Reader

Comment by lusr | Hacker News Reader

In the context they're referring to, the idea is to enter two positions simultaneously (not to enter and exit one position - in that case you'd have to predict which direction the market is going)

1. You buy and sell at the opportunity price (two positions, not one).

2. Prices converge.

3. You close out your positions in #1 and make a profit regardless of which direction the market went.

View on HN

Ah, I see, thank you. I assume my mistake was thinking of assets in an old-world fungible sort of way, where if you buy an asset in one market you can sell it to someone else in a different market, rather than a new-world sort of way with highly complex financial products that you perform complex transactions with. Is that the explanation for why you can't buy the asset at the cheaper price and sell it at the more expensive price immediately, rather than waiting for the prices to converge? Or is it that by performing two independent transactions you can do both simultaneously and so expose yourself to less risk?

Well I'm no expert (only what I know from university 10 years ago and working in related areas of finance), but it's the latter.

One further piece of the story is that the positions are in different markets for related (or the same) financial instruments, e.g. if two different markets have different prices for e.g. USD/EUR you have a profitable arbitrage opportunity when the difference is large enough to cover finance/transaction costs.

In other words, we're talking about situations where two markets have priced the same thing differently and obviously both cannot be correct -- therein lies the opportunity for arbitrage, before the prices converge.

If it is true arbitrage (that is, no price risk), it is very rarely in the same market.

e.g, let's say you can buy 1 EUR by paying 1.2 USD, buy 1 GBP by paying 1 EUR and buy 1.3 USD by paying 1 GBP, you would be left with 0.1 USD with no risk - in this case you've made an arbitrage profit of 10 cents, and you're likely to be able to do that "in the same market", and won't have to do more than one thing simultaneously.

However, everyone is looking out for these, so if such an opportunity presents itself, everyone tries to take advantage of it, thereby changing the price; These opportunities last milliseconds or even microseconds these days, the profit that can be extracted is very small, and you need to be very well positioned (technically) to be able to make it.

Another form of (mostly true) arbitrage is when the same thing gets traded in multiple venues - e.g. Gold in NY, London and Asian markets. In this case, you CAN'T buy a bar in one place and sell in the other place at the same time, because you need time&money to transfer the metal. So you go long in one place, go short in the other place, and then wait for the prices to converge WITHOUT trying to move physical gold around, and do the reverse transactions. There is no real price risk here (if the prices never converge, you CAN move the gold bar around for a small cost), but there's a lot of procedural and counter party risk.

There's what's known as "stat arb" (statistical arbitrage), which is all statistics and no arbitrage; Let's say Gold and Silver tend to both move together (when Gold moves up or down by 1%, silver tends to move in the same direction by 1%). Now, you see Gold moved up 2%, but silver hasn't. You assume either gold will move down 2% back to silver, or silver will move up 2% to match gold, or gold will move down 1% and silver up 1%. The way to take advantage of that is to sell gold for an amount $X, buy $X of silver, and wait. Whatever scenario happens, your balance (when you sell your silver and buy back the gold), you will make money. Unless ... gold and silver stay divergent, which can happen. In which case, you'll lose money.

Thanks to financial wizardry, you can very often sell things you don't own, and "buy them back" later to make things whole again.

(And finally, there's something known as "risk arbitrage" - it is all risk, no arbitrage, but it sounds like you're not just gambling if you call it "risk arbitrage")

Reply on news.ycombinator.com