I don't think it's that easy. There's a time and currency cost to transferring coins between wallets so that's not a good idea.
Instead you'd need to keep large balances in multiple exchanges and then sell coins on the more expensive, buy on the cheaper, and then through the power of algebra, you may make a profit? maybe. Since the direction/stability of the market isn't known you don't know who's the dog and who's the tail. Does it matter? Yes. If the market is genuinely trending and you have the insights to exploit arbitrage, there's better strategies than arbitrage available to you.
But here's a different, more exploitable model. Pretend two exchanges had price X. And you know for a fact you can make one X-1 temporarily and you can make the other X+1 temporarily ... and then they will elastically close back to X in a predictable time Y.
So you "stretch" the exchanges, do the scheme described above, in volume, and all things being equal and nothing else dramatically affecting the market, they will eventually snap back like a rubberband. Since you aren't pushing everything the same direction, you probably won't trigger a trend.
Now you have USD in the +1 account and BTC in the -1 account so you stretch it the other way and repeat.
This is a much better strategy since you are essentially bending the delta back and forth and exploiting it each time.
Also this strategy, which I'll call twanging, is generally strongly knowable. A reliable, repeatable, predictable, strategy with a profit>0 is almost always worthwhile independent of the profit amount because the other factors, risk and predictability, have been reduced to effectively zero.
Also the time to twang is probably on the order of an hour.
So you permute around the markets all day long, picking new tuples in some shuffled sequence and you have a reliable profit machine at scale.
This is all theoretical of course - I'm not actually whaling around with $100mil USD in assets spoofing markets. Like most trading strategies, it's likely nonsense.
Instead you'd need to keep large balances in multiple exchanges and then sell coins on the more expensive, buy on the cheaper, and then through the power of algebra, you may make a profit? maybe. Since the direction/stability of the market isn't known you don't know who's the dog and who's the tail. Does it matter? Yes. If the market is genuinely trending and you have the insights to exploit arbitrage, there's better strategies than arbitrage available to you.
But here's a different, more exploitable model. Pretend two exchanges had price X. And you know for a fact you can make one X-1 temporarily and you can make the other X+1 temporarily ... and then they will elastically close back to X in a predictable time Y.
So you "stretch" the exchanges, do the scheme described above, in volume, and all things being equal and nothing else dramatically affecting the market, they will eventually snap back like a rubberband. Since you aren't pushing everything the same direction, you probably won't trigger a trend.
Now you have USD in the +1 account and BTC in the -1 account so you stretch it the other way and repeat.
This is a much better strategy since you are essentially bending the delta back and forth and exploiting it each time.
Also this strategy, which I'll call twanging, is generally strongly knowable. A reliable, repeatable, predictable, strategy with a profit>0 is almost always worthwhile independent of the profit amount because the other factors, risk and predictability, have been reduced to effectively zero.
Also the time to twang is probably on the order of an hour.
So you permute around the markets all day long, picking new tuples in some shuffled sequence and you have a reliable profit machine at scale.
This is all theoretical of course - I'm not actually whaling around with $100mil USD in assets spoofing markets. Like most trading strategies, it's likely nonsense.