[smokinn]

When it's just a tiny percentage of the overall trades, it doesn't really matter.

As Goldman Sachs is now proving though, it can be wildly profitable. What happens if it eventually becomes the dominant form of trading? What does the stock market then represent? Surely not the value of a company, based on research and bets on the future. Doesn't it just become a high speed game where computers try to steal pennies from each other?

[cellis]

Ahem. Please don't attribute Goldman's success to trading. Goldman Sachs recent "success" is largely due to them adroitly navigating the political waters of the bailout, without which i suspect they would be in serious trouble.

As for machine trading or stat-arb, high freq, whatever...i don't think it is any better than a random walk. Risk was (is?) just being hidden somewhere and lied about.

[eru]

Not necessarily. E.g. arbitrage is really risk-free.

[cellis]

Frictionless arbitrage?

[eru]

No. You need to find enough of an arbitrage opportunity to overcome the friction. But once you found it, it's a risk-free profit. (You can't do too complex an atomic transaction on the financial markets; but most arbitrage deals take much less than a day --- even less than a few minutes, so your window of exposure is quite short.)

[roundsquare]

If you are trading based on research and conviction then this sort of predatory behavior should only minimal effect. I.e. if you think a company has good prospects for the future, then you probably expect the price of the stock to go up a significant percent. The sort of games described in the paper only alter the price by a small percent over a short period of time. Does it really matter if you buy the stock at 20.00 or 20.10 if you think the price is going to go up to 30.00?

[_cbsz]

For at least "conventional" blue-chip stocks, if the market fails to represent the present value of all future dividends, discounted back to the present, trading based on that metric should be a dominant strategy, since it is a method of extracting value from a stock without trading it.

[Eliezer]

If the stock is underpriced, sure. But shorting overpriced stocks comes with a cost - what if the market gets even more irrational over the next few years?

[_cbsz]

If stocks are typically overpriced, I think you want to be in the business of IPOing.

[eru]

Cover your old shorts with new shorts.