7 ms·

machine trading has been in heavy use for a decade. a little late to wonder about its efficacy now

by ohgrayt 17y ago

smokinn 17y ago
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?
- swolchok 17y ago
  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 17y ago
    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?
    - eru 17y ago
      Cover your old shorts with new shorts.
    - swolchok 17y ago
      If stocks are typically overpriced, I think you want to be in the business of IPOing.
- cellis 17y ago
  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 17y ago
    Not necessarily. E.g. arbitrage is really risk-free.
    - cellis 17y ago
      Frictionless arbitrage?
      - eru 17y ago
        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 17y ago
  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?

machine trading has been in heavy use for a decade. a little late to wonder about its efficacy now

by ohgrayt 17y ago

machine trading has been in heavy use for a decade. a little late to wonder about its efficacy now

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?

swolchok 17y ago
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 17y ago
  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?
  - eru 17y ago
    Cover your old shorts with new shorts.
  - swolchok 17y ago
    If stocks are typically overpriced, I think you want to be in the business of IPOing.
cellis 17y ago
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 17y ago
  Not necessarily. E.g. arbitrage is really risk-free.
  - cellis 17y ago
    Frictionless arbitrage?
    - eru 17y ago
      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 17y ago
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?