5 ms·

This article is pretty terrible. First it talks about high cancel rates (95-97%), as if this is somehow evidence of bad behavior. It isn't. Matt Levine has a ni

This article is pretty terrible. First it talks about high cancel rates (95-97%), as if this is somehow evidence of bad behavior. It isn't. Matt Levine has a nice article with simple math showing that a 95% cancel rate is unavoidable. http://www.bloombergview.com/articles/2015-10-08/why-do-high-frequency-traders-cancel-so-many-orders- http://www.bloombergview.com/articles/2015-10-08/why-do-high... Then it talks about selling order flow to HFTs as if this somehow harms the little guy. It doesn't - it lets HFTs offer liquidity (potentially at better prices, and never at worse prices) to the little guy that they would be too scared to offer to big guys. https://www.chrisstucchio.com/blog/2014/fervent_defense_of_frontrunning_hfts.html https://www.chrisstucchio.com/blog/2014/fervent_defense_of_f... Following all this FUD is a bit of shilling for IEX. Why is this nonsense here?

blazespin 11y ago
Basically, it's just another person who doesn't want AI meddling in the stock markets. Sorry folks, that ship sailed a very very long time ago. Short term, all of our markets now belong to AI consuming alternative data (think weather, satellite pictures of parking lots, etc), economic releases, etc. If you want to trade the markets and not program AI, your best bet it is to make longer term bets where AI doesn't perform so well.
codeismightier 11y ago
I'm not a fan of payment for order flow. The problem is that it lets HFTs pick off the uninformed orders, leaving the orders that actually reach the exchange more "toxic". This lowers the incentive for others to rest orders on the Central Limit Order Book, harming price discovery. So while in the short term no one is harmed, the long-run consequences could be bad. With that said, yeah, the NYT piece is terrible.
- crzwdjk 11y ago
  If the HFTs pick off uninformed orders, wouldn't that mean that the remaining orders are more informed? That does seem like it would make trading on the exchange itself would get more expensive, though, because this is all basically a mechanism for the HFTs to price-discriminate. But the market did somehow work back in the bad old days of odd-eights and such.
executive 11y ago
It's an advertorial written by the CIO of Yale. Yale is backing IEX.
lotsoflumens 11y ago
The very first assertion in the article is "AMERICA’S equity markets are broken." Do you think that's false? I trade the "markets" every day and I think they are almost completely broken. If HFTs supply liquidity, what do you think happened during the flash crashes of recent history? Hint: they stopped supplying liquidity. That's why Virtu can make 2 years of trades with only one negative day - a feat that is almost impossible without the front-running advantage that they have.
- consz 11y ago
  Why do you think Virtu's performance is impossible without "front-running"? Do you not think you would see similar results for any firm with statistically positive EV that makes a large number of trades every day?
  - lotsoflumens 11y ago
    No I don't. Can you name any such "firm"?
    - consz 11y ago
      Sure -- Jump, Tower, HRT, Citadel all have similar results. Why do you not believe such results are possible? Do you think it's impossible for a firm to have statistically positive edge for sustained periods of time?
      - lotsoflumens 11y ago
        I assume you're referring to these guys: http://www.bloomberg.com/news/articles/2014-07-23/don-t-tell-anybody-about-this-story-on-hft-power-jump-trading http://www.bloomberg.com/news/articles/2014-07-23/don-t-tell...
        consz 11y ago
        Of course, I think they're the only trading firm named Jump. Is there something particular in that article you wanted to point out?

blazespin 11y ago

Basically, it's just another person who doesn't want AI meddling in the stock markets. Sorry folks, that ship sailed a very very long time ago. Short term, all of our markets now belong to AI consuming alternative data (think weather, satellite pictures of parking lots, etc), economic releases, etc. If you want to trade the markets and not program AI, your best bet it is to make longer term bets where AI doesn't perform so well.

codeismightier 11y ago

I'm not a fan of payment for order flow. The problem is that it lets HFTs pick off the uninformed orders, leaving the orders that actually reach the exchange more "toxic". This lowers the incentive for others to rest orders on the Central Limit Order Book, harming price discovery. So while in the short term no one is harmed, the long-run consequences could be bad. With that said, yeah, the NYT piece is terrible.

crzwdjk 11y ago
If the HFTs pick off uninformed orders, wouldn't that mean that the remaining orders are more informed? That does seem like it would make trading on the exchange itself would get more expensive, though, because this is all basically a mechanism for the HFTs to price-discriminate. But the market did somehow work back in the bad old days of odd-eights and such.

executive 11y ago

It's an advertorial written by the CIO of Yale. Yale is backing IEX.

lotsoflumens 11y ago

The very first assertion in the article is "AMERICA’S equity markets are broken." Do you think that's false? I trade the "markets" every day and I think they are almost completely broken. If HFTs supply liquidity, what do you think happened during the flash crashes of recent history? Hint: they stopped supplying liquidity. That's why Virtu can make 2 years of trades with only one negative day - a feat that is almost impossible without the front-running advantage that they have.

consz 11y ago
Why do you think Virtu's performance is impossible without "front-running"? Do you not think you would see similar results for any firm with statistically positive EV that makes a large number of trades every day?
- lotsoflumens 11y ago
  No I don't. Can you name any such "firm"?
  - consz 11y ago
    Sure -- Jump, Tower, HRT, Citadel all have similar results. Why do you not believe such results are possible? Do you think it's impossible for a firm to have statistically positive edge for sustained periods of time?
    - lotsoflumens 11y ago
      I assume you're referring to these guys: http://www.bloomberg.com/news/articles/2014-07-23/don-t-tell-anybody-about-this-story-on-hft-power-jump-trading http://www.bloomberg.com/news/articles/2014-07-23/don-t-tell...
      - consz 11y ago
        Of course, I think they're the only trading firm named Jump. Is there something particular in that article you wanted to point out?