Same reason they step in and bale out your local bank if it’s threatening to go down the toilet and take every penny belonging to you, your neighbours and local businesses along with it. The potential collateral damage is just too big - would you risk another Great Crash just to make sure a bunch of rich gamblers got soaked?
This is the beginning of the end with AI trading. Let’s say, for giggles, that there are 100 ‘top secret’ ways in which to beat the market almost every time. The minute you get your 101st AI computer to jump into the game, then you will have multiple players in a limited field, which will change the results, and effectively limit the margins. At that moment you will have at least 2 AI’s competing against each other to make $…Which one wins? The one with the fastest Internet connection?
There is a huge difference between investing $1,000 and $10,000,000 and $10,000,000,000. A strategy that makes 10% on $1,000 is never going to make 10% on $10b. This is pretty much what has happened to the Hedge Fund market over the past 15 years. In the early 90’s, Hedge Funds were making great returns…they had new strategies, didn’t have to play by the old rules, could jump around nimbly and reap excessive rewards because not everyone was doing it. Then Hedge Funds became the flavour of the decade, everyone jumped into it, managers started trading their ideas…and the Hedge Fund market stopped making excessive amounts of money. Managers were trading against each other, they were walking the price up (if I buy $10 of stock no one notices. If I buy $10m, everyone notices, and the prices start to rise), strategies were acting in conflict with each other (If one strategy calls for buying X, and another calls for selling X, they can cancel make money in isolation, maybe, but if done at the same time one loses).
For a true AI to make money on the market, it would have to be unique. Alone it could make money by finding ways to work the existing system. But once a second AI comes along, both AI’s would have to take into consideration that there is another AI out there, and they would have to think about what the other one is going to do. Now repeat that for millions of AI’s and the market will stabilize and a new ‘sure’ thing will have to be invented.
-Tcat
I see the end result of lots of good AI’s scanning the market to be a closer-to efficient market.
What are they doing, essentially?
Analyzing tons and tons of data, and making instantaneous moves based on combinations of variables that indicate that a stock not priced correctly.
With more AI’s operating, each with it’s own unique set of rules, (but all basically doing the same thing) you’re going to have every(?) possible inefficiency instantly pounced upon. That’s practically the definition of an efficient market.
Although, I guess it’s also possible that new indicators arise, and the best AI’s at finding them are going to benefit. Really, I don’t think I see the market looking much different than it’s looked for the past 100 years.
market analysis is interesting, but what even the ph.D economists (sometime) forget is: the past is no indicator of the future-stocks are not like a harmonic oscillator. for instance, the "beta’ of stocks is calculated on the basis of past behavior; any given stiock may change its beat right away. So i think that we can spend all the computing time in the world, and still come up with “buy low, sell high”.