Naked Science Forum
Non Life Sciences => Technology => Topic started by: jimbobghost on 05/11/2018 17:49:09
-
I am specifying over a long period of trading.
I spent years developing a mechanical trading system, and another 15 or so actually day trading stocks.
after about 20 years in total, many thousands of dollars in equipment, and millions of dollars trading, I finally gave up when my gains had been reduced back to a break even point.
my problem seemed to be the randomness of the market; and no amount of optimization and back testing could overcome it. the final straw (perhaps coincidently) came with high speed trading algorithms.
anyone involved in computerized mechanical trading systems?
-
My Dad worked for many years as a commodity broker on the London Commodity Exchange (LCE). In the 1980s a computerised / automated trading system was introduced on some of the markets. My Dad said it destroyed the market and no one could make any money. As a result they reversed some of the implementation because people preferred the human touch. I was a bit too young at the time to understand exactly what he meant; perhaps there are visitors to the forum who have experience and can comment?
-
my problem seemed to be the randomness of the market
Predicting the future is an unsolved problem in computing - especially in chaotic systems like the stock market.
An earlier successful mechanical trading system was the Automatic Totalisator, which took bets on horse races and decided the odds so that the "house" always won. Rather than prescience, this relied on knowing the total pool of bets.
Of course, you need a regulatory environment which forces punters to use this system so the house can win (or persuades the punters that all other forms of betting are run by criminals).
See: https://en.wikipedia.org/wiki/Tote_board#Automatic_totalisators