I believe algo trading is very possible for big institutions... usually involving arbitrage.
Theoretically any one with successful mechanical trading system should be able to translate it into successful coding.
But practically there seems to be peculiar algo cost (simply licensing and trading fees, slippages) ... to translate successfully mech. trading system should generate enough profits to cover these costs. Apart from simple trading cost there is a knowledge cost (knowledge for making a successful mech. system first then expertise to implement the technical plumbing work, which can be hired) which is too big for many retail traders.
So l89, if you happen to jump through these algo chasms and land on other side...usually populated by institutional guys... that's really great!