5 ms·

Actually, it works in the opposite direction. Here is why: if betting against the herd was the right thing to do, algos would learn this behavior and all algo

Actually, it works in the opposite direction. Here is why: if betting against the herd was the right thing to do, algos would learn this behavior and all algos would start betting against a big price down move. This is why algos actually reduce volatility, not increase it. And volatility is exacerbated by humans: trade wars, attacks on the fed and general erratic behavior of our president

xapata 8y ago
No, betting with the herd is the correct bet, except that one needs to place the bet before everyone else does. Algorithmic, low latency trading reduces bid-ask spread under normal circumstances, but increases the risk of extreme events.
- vasilipupkin 8y ago
  We had plenty of algorithmic trading and years of very low volatility at the same time. If algorithmic trading increased risk of extreme events, it would have shown up during let’s say 2012-2017 years of low volatility. No, algorithmic trading actually reduces risk of extreme events such as fat fingers or extreme bets by humans. What drives volatility is macroeconomic environment.
  - xapata 8y ago
    You don't remember any of the "flash" crashes? Turbulence has increased, regardless of the general trend.
    - vasilipupkin 8y ago
      Flash crash of 2010 was extensively investigated and no evidence that algorithmic trading was at fault was found. After that, we had some of the lowest turubulence years on record, while algorithmic trading was going on, so your theory is not supported by evidence.
      - xapata 8y ago
        There have been others. Plus there's some enjoyable literature if you're curious. Mark Buchanan has a nice blog and some links. The gist of it is that in the low latency space the ratio of players to possible strategies is too high.
        vasilipupkin 8y ago
        if yoi try to tie low latency or algorithmic strategies to volatility, you need to be able to explain multiple years of really low volatility despite all those strategies going on.
        xapata 8y ago
        Low volatility according to what measure? I don't keep up with the literature anymore, since around 2010/2011, but it'd surprise me if we entered some new regime. Googling for "flash crash" turns up many more examples than just the one you mentioned from 2010.

xapata 8y ago

No, betting with the herd is the correct bet, except that one needs to place the bet before everyone else does. Algorithmic, low latency trading reduces bid-ask spread under normal circumstances, but increases the risk of extreme events.

vasilipupkin 8y ago
We had plenty of algorithmic trading and years of very low volatility at the same time. If algorithmic trading increased risk of extreme events, it would have shown up during let’s say 2012-2017 years of low volatility. No, algorithmic trading actually reduces risk of extreme events such as fat fingers or extreme bets by humans. What drives volatility is macroeconomic environment.
- xapata 8y ago
  You don't remember any of the "flash" crashes? Turbulence has increased, regardless of the general trend.
  - vasilipupkin 8y ago
    Flash crash of 2010 was extensively investigated and no evidence that algorithmic trading was at fault was found. After that, we had some of the lowest turubulence years on record, while algorithmic trading was going on, so your theory is not supported by evidence.
    - xapata 8y ago
      There have been others. Plus there's some enjoyable literature if you're curious. Mark Buchanan has a nice blog and some links. The gist of it is that in the low latency space the ratio of players to possible strategies is too high.
      - vasilipupkin 8y ago
        if yoi try to tie low latency or algorithmic strategies to volatility, you need to be able to explain multiple years of really low volatility despite all those strategies going on.
        xapata 8y ago
        Low volatility according to what measure? I don't keep up with the literature anymore, since around 2010/2011, but it'd surprise me if we entered some new regime. Googling for "flash crash" turns up many more examples than just the one you mentioned from 2010.