Since 2009 high-frequency trading firms, which represent approximately 2% of the nearly 20,000 trading firms operating in the U.S. markets, have accounted for over 73% of all U.S. equity trading volume.
Also note that this excludes lots of other very short term traders in the market like individual day traders operating out of brokerage houses, coffee shops, office suites and living rooms around the nation, and institutional investors who are making some significant trades every day for time horizons measured in days and weeks. Many of these traders are doing essentially the same thing as the high-frequency traders in a less sophisticated way.
The fact that so few firms engage in high-frequency trading, and that so many day traders are economically marginal, suggests that the profits to be made from this kind of activity are pretty marginal. This is classic arbitrage of rounding errors and kindred "substance blind" investment decision making, in some ways the opposite of another "substance blind" investment approach called "index trading" which tries to replicated overall market return rather than trying to beat it.
Once you filter out the high-frequency technical trading and blind index fund activity from the equity markets, you are left with a pretty modest volume of smart money trading on the economic merits of publicly held companies for the relatively long term.
Vassar economist Rajiv Sethi has written a lot on algorithmic trading and kindred issues of market strategy. One of his bottom line conclusions, as I understand it, is that markets have a self-correcting tendency to shift from substance blind follow the leader kinds of trading to investment merit based trading, in what he calls "endogenous regime switching" (hand it to academics to give great insights boring names), as the proportion of algorithmic traders in the market overleverages the outside information driven signals in the market place.
When most of the trades in the market are driven by traders running on autopilot, the market itself becomes unstable, because technical bets can be confused with genuine new information about the outside world and leverage noise into major market movements.