While not addressing the question of a role for AI I often find myself thinking we should get away for the frequent trading of financial assets and make them a bit more like the trading of mutual funds. Does all the intra-day trades really give more information or just add noise and the opportunity for the insiders to make money off retail (and even some institutional) investors?
Seem like designing the market to work a bit more like the one often used in the Econ 101 theory—that Walrasian Auctioneer—we could have more stable markets that do better at pricing capital assets than today. In other words, take all the order flow see that the prices are to clear and then all trade occurs at that price.
I suspect you’d still see some gaming the system with fake orders (a bit like the algos have been accused of in today’s markets) but all systems get gamed.
This would have the consequence that if you see that XXXX is trading at $Y, phone up your broker and ask to sell your holding in $XXXX, it could very well end up selling at $Y/2.
That’s a thing that can happen already, but the delay between saying “sell” and actually selling is typically measured in seconds rather than hours, which makes big divergences like that less likely.
Of course you can avoid this by not saying “please sell 100 shares of XXXX” but “please sell 100 shares of XXXX unless the price drops below $Z, in which case don’t”. But this is more complexity than most retail investors want to handle :-).
While not addressing the question of a role for AI I often find myself thinking we should get away for the frequent trading of financial assets and make them a bit more like the trading of mutual funds. Does all the intra-day trades really give more information or just add noise and the opportunity for the insiders to make money off retail (and even some institutional) investors?
Seem like designing the market to work a bit more like the one often used in the Econ 101 theory—that Walrasian Auctioneer—we could have more stable markets that do better at pricing capital assets than today. In other words, take all the order flow see that the prices are to clear and then all trade occurs at that price.
I suspect you’d still see some gaming the system with fake orders (a bit like the algos have been accused of in today’s markets) but all systems get gamed.
This would have the consequence that if you see that XXXX is trading at $Y, phone up your broker and ask to sell your holding in $XXXX, it could very well end up selling at $Y/2.
That’s a thing that can happen already, but the delay between saying “sell” and actually selling is typically measured in seconds rather than hours, which makes big divergences like that less likely.
Of course you can avoid this by not saying “please sell 100 shares of XXXX” but “please sell 100 shares of XXXX unless the price drops below $Z, in which case don’t”. But this is more complexity than most retail investors want to handle :-).