6 ms·

> why should you do better? It's a zero sum game, but one might be smarter than others. Some companies doing algorithmic trading consistently skim off their sh

> why should you do better? It's a zero sum game, but one might be smarter than others. Some companies doing algorithmic trading consistently skim off their share of the market's inefficiencies. What I'm wondering is how much of an investment it would be to get into high-frequency trading for yourself. Any idea?

mathgenius 17y ago
Probably 10 - 50 million. You will need 5 - 10 hard core programmers, a network engineer, a trader or three, a bunch of the fastest servers around, etc. etc. This is an insane idea, I would not recommend it! (The margins are being pushed to zero as more and more companies get into this.)
- nearestneighbor 17y ago
  I guess I meant excluding human labor (imagine that a bunch of programmers, etc. with the necessary skills wanted to do it). Would they need to rent a building next to the NYSE and put the machines there?
  - hiteshiitk 17y ago
    Yes they need to buy an office nearby to NYSE, to get fastest feeds(~ us).
    - nearestneighbor 17y ago
      Are there no hosting / cloud services for this sort of thing?
      - adw 17y ago
        The network latency will kill you. You want to be in the same rack, not just the same colo, as the exchange's endpoints.
        nearestneighbor 17y ago
        Are those slots in the same rack auctioned or something? How does having an office in a nearby building help?
        adw 17y ago
        Auctioned: I believe so. I don't do HFT, but with what our company does, I've spoken to a few people who do. So take all of this with a pinch of salt, but as I understand it, your execution and autonomous trading systems live in the same rack as your exchange/dark-pool endpoints. Your slower strategies can be offsite, as is the Big Red Button to close down your positions. Still, faster's invariably going to be better, all else being equal...
arthurdent 17y ago
http://www.algodeal.com http://www.algodeal.com Someone sent me this gem a few days ago. I love the idea (for the guys who run it AND the guys who it opens up opportunities for). Might be something you could play with. Depending on what sort of HF strategy you want to run, I imagine this can be done relatively cheaply. (300K-800k? Pretty wide range I know... that's all I got though). Obviously your strategies are limited, you're not competing with some of the super low latency firms for the fastest of opportunities because of hardware limitations or price prohibitive vendors/deals with exchanges or whatnot. I was just reminded of this: http://www.reddit.com/r/IAmA/comments/9s9d7/iama_100_automated_independent_retail_trader_i/ http://www.reddit.com/r/IAmA/comments/9s9d7/iama_100_automat... If I recall correctly, he got started with under 100k and has a pretty inexpensive setup. "High Frequency" trading means a lot of different things to a lot of different people.
- nearestneighbor 17y ago
  IAMA thread looks very interesting, thanks (although you never know when it's someone pretending to be something he's not). I'm a bit skeptical of algodeal. If they are seriously doing this (actually licensing strategies and following them), wouldn't they be setting themselves up to get ripped off? It seems that if you know what strategy someone is going to use, you can use it against them (effectively changing the market between historical tests and a live run).
  - arthurdent 17y ago
    I actually think AlgoDeal sounds great, but can understand your skepticism. I haven't really looked into it, but what it sounds like to me is they provide you a platform to build strategies and test them, play with fake money. They may optionally take your strategy and run it in the market. If it is successful, they split returns with you. When you say they're setting themselves up to get ripped off, I think you're suggesting that Evil Person X may come up with a strategy A that does well enough in backtests, but is paired with a strategy B designed to rip off that strategy A. Then after AlgoDeal applies strategy A, I put my secret millions to work and make better returns for myself. They don't HAVE to run your strategy. I'm assuming they're smart guys and have some selection criteria other than "performed well in backtests", so they'd try to choose strategies that were less likely to get ripped off by the evil person's counter-strategy. Also, I think of this as being the VC who seems massive dealflow. Half of it is the fun of seeing all the good ideas and picking the top ones. Some will still fail, but choosing from a much larger base of ideas is great. Plus they utilize their own expertise to tweak strategies for themselves. Its basically YC for algo strategies, except they don't have to pay you, they can steal your startup and change it a bit, and they take a much bigger percentage (i'd guess 50-80%, based on what I think are something like industry standards). Its quite common place in finance to have people give you strategies and give them a cut of the profits. This takes it to the next level: crowdsources strategies and removes the barrier to entry for a lot of smart people (trading costs) in exchange for the right to tap those ideas. Love it.
joe_the_user 17y ago
The capital-gains tax is a reason not to get into short-term trading: http://www.fool.com/personal-finance/taxes/2008/09/04/another-reason-not-to-day-trade.aspx http://www.fool.com/personal-finance/taxes/2008/09/04/anothe...
- nearestneighbor 17y ago
  Is this relevant to LLCs?
- arthurdent 17y ago
  >The capital-gains tax is a reason not to get into short-term trading In the same way taxes are a reason to not get into money making ventures. Short term trading encompasses a lot of ideas. Some are good some aren't. Many "short term" trading strategies have significantly higher sharpe ratio's than Fool style investing. Fool investing is great. I learned a LOT from that site. But I also want returns in shorter time spans, so I'm willing to pay that capital gains tax. Motley fool is a GREAT source for long term investing knowledge for anyone interested in that. A lot of other great classic investment texts are getting thrown around in this comment thread (Intelligent Investor, for one). Recommended book for those interested in "Motley Fool"/Benjamin Graham/Warren Buffet type investing, I'd also recommend http://www.amazon.com/Little-Beats-Market-Books-Profits/dp/0471733067 http://www.amazon.com/Little-Beats-Market-Books-Profits/dp/0... Its sort of in the same vein, but slightly shorter term, backed up by a lot of statistics. Basic principle: Choose the best 20 companies every year. Define "best" as top PE or some other really simple metric. You can tweak it a little as well based on your personal risk profile, desired industry exposure. Consistently outperforms the market. The motleyfool is a site that looks amazingly like spam and reminds you of all the "earn 50 bazillion percent every hour using options!" emails you get, but is actually chock full of great information. Greenblatt's book sounds shady as hell, but its a quick, easy read and quite informative, backed by just enough logic to make you want to try it.
sorbits 17y ago
It's a zero sum game It most certainly is not. When you buy shares they represent a share in a company. This company may rise in value due to good management, successful products, expanding into new markets, etc. If you sell your share at a higher price, no-one is going to lose the money you will earn, the company is just worth more than when you bought it. Other securities may make you money in other ways, e.g. if you buy T-bills the FED is paying you interest on the money you effectively loan them.
- nearestneighbor 17y ago
  Relative to the average market performance , making choices as to what to invest in, seems to be a zero-sum game (by definition, as the net total performance of the investments will be average), but I'm not a finance guru, so feel free to correct me.
  - sorbits 17y ago
    What I think you are implying is that if you invest in companies part of some market index¹ and end up above the average performance of this index then someone else, who also invested only in companies in this index, will end up below the average. This however is not zero sum game or has anything to do with markets or investments, it is just how averages work. You could apply the same logic to grades in a school class. ¹ There are lots of market indexes and they only cover a fraction of the investment possibilities and are often trade specific.
    - nearestneighbor 17y ago
      I think what you are saying is that if everyone invested very intelligently, then everyone would be better off, as more money would pour into the more viable ventures (is this what you mean?) I can agree with that, FWIW, but from the short-term algorithmic/high-frequency trader 's perspective (edit), this does not appear relevant (some nth order effect). P.S. Traders often say how their work is useful to everyone, because it makes the markets more efficient. I wonder if it's true. The analogy I'm thinking about is frequent lane swervers on the freeway ("lane arbitrage"). Are they making the freeway more efficient for everyone? I don't think so.
herval 17y ago
it's not a zero sum game. The definition of zero sum games is "A situation in which one's gains result only from another's equivalent losses.". There are many aspects of a stock market that make it NOT sum to zero: the fact that things can go up or down in value with no one buying/selling (aka GAPS) itself makes it a positive/negative sum game. Not to mention taxes, broker costs, IPOs and all other 'game moves' that inject and/or take money out of the 'game'.
adamtmca 17y ago
An interesting note on the zero sum discussion, while the stock market clearly is not a zero sum game options on the other hand ARE zero sum. On the exercise date the payoff for the buyer of an option that's in the money will exactly match the loss of the seller. Vice versa if it's out of the money. Kind of neat.

mathgenius 17y ago

Probably 10 - 50 million. You will need 5 - 10 hard core programmers, a network engineer, a trader or three, a bunch of the fastest servers around, etc. etc. This is an insane idea, I would not recommend it! (The margins are being pushed to zero as more and more companies get into this.)

nearestneighbor 17y ago
I guess I meant excluding human labor (imagine that a bunch of programmers, etc. with the necessary skills wanted to do it). Would they need to rent a building next to the NYSE and put the machines there?
- hiteshiitk 17y ago
  Yes they need to buy an office nearby to NYSE, to get fastest feeds(~ us).
  - nearestneighbor 17y ago
    Are there no hosting / cloud services for this sort of thing?
    - adw 17y ago
      The network latency will kill you. You want to be in the same rack, not just the same colo, as the exchange's endpoints.
      - nearestneighbor 17y ago
        Are those slots in the same rack auctioned or something? How does having an office in a nearby building help?
        adw 17y ago
        Auctioned: I believe so. I don't do HFT, but with what our company does, I've spoken to a few people who do. So take all of this with a pinch of salt, but as I understand it, your execution and autonomous trading systems live in the same rack as your exchange/dark-pool endpoints. Your slower strategies can be offsite, as is the Big Red Button to close down your positions. Still, faster's invariably going to be better, all else being equal...

arthurdent 17y ago

http://www.algodeal.com http://www.algodeal.com Someone sent me this gem a few days ago. I love the idea (for the guys who run it AND the guys who it opens up opportunities for). Might be something you could play with. Depending on what sort of HF strategy you want to run, I imagine this can be done relatively cheaply. (300K-800k? Pretty wide range I know... that's all I got though). Obviously your strategies are limited, you're not competing with some of the super low latency firms for the fastest of opportunities because of hardware limitations or price prohibitive vendors/deals with exchanges or whatnot. I was just reminded of this: http://www.reddit.com/r/IAmA/comments/9s9d7/iama_100_automated_independent_retail_trader_i/ http://www.reddit.com/r/IAmA/comments/9s9d7/iama_100_automat... If I recall correctly, he got started with under 100k and has a pretty inexpensive setup. "High Frequency" trading means a lot of different things to a lot of different people.

nearestneighbor 17y ago
IAMA thread looks very interesting, thanks (although you never know when it's someone pretending to be something he's not). I'm a bit skeptical of algodeal. If they are seriously doing this (actually licensing strategies and following them), wouldn't they be setting themselves up to get ripped off? It seems that if you know what strategy someone is going to use, you can use it against them (effectively changing the market between historical tests and a live run).
- arthurdent 17y ago
  I actually think AlgoDeal sounds great, but can understand your skepticism. I haven't really looked into it, but what it sounds like to me is they provide you a platform to build strategies and test them, play with fake money. They may optionally take your strategy and run it in the market. If it is successful, they split returns with you. When you say they're setting themselves up to get ripped off, I think you're suggesting that Evil Person X may come up with a strategy A that does well enough in backtests, but is paired with a strategy B designed to rip off that strategy A. Then after AlgoDeal applies strategy A, I put my secret millions to work and make better returns for myself. They don't HAVE to run your strategy. I'm assuming they're smart guys and have some selection criteria other than "performed well in backtests", so they'd try to choose strategies that were less likely to get ripped off by the evil person's counter-strategy. Also, I think of this as being the VC who seems massive dealflow. Half of it is the fun of seeing all the good ideas and picking the top ones. Some will still fail, but choosing from a much larger base of ideas is great. Plus they utilize their own expertise to tweak strategies for themselves. Its basically YC for algo strategies, except they don't have to pay you, they can steal your startup and change it a bit, and they take a much bigger percentage (i'd guess 50-80%, based on what I think are something like industry standards). Its quite common place in finance to have people give you strategies and give them a cut of the profits. This takes it to the next level: crowdsources strategies and removes the barrier to entry for a lot of smart people (trading costs) in exchange for the right to tap those ideas. Love it.

joe_the_user 17y ago

The capital-gains tax is a reason not to get into short-term trading: http://www.fool.com/personal-finance/taxes/2008/09/04/another-reason-not-to-day-trade.aspx http://www.fool.com/personal-finance/taxes/2008/09/04/anothe...

nearestneighbor 17y ago
Is this relevant to LLCs?
arthurdent 17y ago
>The capital-gains tax is a reason not to get into short-term trading In the same way taxes are a reason to not get into money making ventures. Short term trading encompasses a lot of ideas. Some are good some aren't. Many "short term" trading strategies have significantly higher sharpe ratio's than Fool style investing. Fool investing is great. I learned a LOT from that site. But I also want returns in shorter time spans, so I'm willing to pay that capital gains tax. Motley fool is a GREAT source for long term investing knowledge for anyone interested in that. A lot of other great classic investment texts are getting thrown around in this comment thread (Intelligent Investor, for one). Recommended book for those interested in "Motley Fool"/Benjamin Graham/Warren Buffet type investing, I'd also recommend http://www.amazon.com/Little-Beats-Market-Books-Profits/dp/0471733067 http://www.amazon.com/Little-Beats-Market-Books-Profits/dp/0... Its sort of in the same vein, but slightly shorter term, backed up by a lot of statistics. Basic principle: Choose the best 20 companies every year. Define "best" as top PE or some other really simple metric. You can tweak it a little as well based on your personal risk profile, desired industry exposure. Consistently outperforms the market. The motleyfool is a site that looks amazingly like spam and reminds you of all the "earn 50 bazillion percent every hour using options!" emails you get, but is actually chock full of great information. Greenblatt's book sounds shady as hell, but its a quick, easy read and quite informative, backed by just enough logic to make you want to try it.

sorbits 17y ago

It's a zero sum game It most certainly is not. When you buy shares they represent a share in a company. This company may rise in value due to good management, successful products, expanding into new markets, etc. If you sell your share at a higher price, no-one is going to lose the money you will earn, the company is just worth more than when you bought it. Other securities may make you money in other ways, e.g. if you buy T-bills the FED is paying you interest on the money you effectively loan them.

nearestneighbor 17y ago
Relative to the average market performance , making choices as to what to invest in, seems to be a zero-sum game (by definition, as the net total performance of the investments will be average), but I'm not a finance guru, so feel free to correct me.
- sorbits 17y ago
  What I think you are implying is that if you invest in companies part of some market index¹ and end up above the average performance of this index then someone else, who also invested only in companies in this index, will end up below the average. This however is not zero sum game or has anything to do with markets or investments, it is just how averages work. You could apply the same logic to grades in a school class. ¹ There are lots of market indexes and they only cover a fraction of the investment possibilities and are often trade specific.
  - nearestneighbor 17y ago
    I think what you are saying is that if everyone invested very intelligently, then everyone would be better off, as more money would pour into the more viable ventures (is this what you mean?) I can agree with that, FWIW, but from the short-term algorithmic/high-frequency trader 's perspective (edit), this does not appear relevant (some nth order effect). P.S. Traders often say how their work is useful to everyone, because it makes the markets more efficient. I wonder if it's true. The analogy I'm thinking about is frequent lane swervers on the freeway ("lane arbitrage"). Are they making the freeway more efficient for everyone? I don't think so.

herval 17y ago

it's not a zero sum game. The definition of zero sum games is "A situation in which one's gains result only from another's equivalent losses.". There are many aspects of a stock market that make it NOT sum to zero: the fact that things can go up or down in value with no one buying/selling (aka GAPS) itself makes it a positive/negative sum game. Not to mention taxes, broker costs, IPOs and all other 'game moves' that inject and/or take money out of the 'game'.

adamtmca 17y ago

An interesting note on the zero sum discussion, while the stock market clearly is not a zero sum game options on the other hand ARE zero sum. On the exercise date the payoff for the buyer of an option that's in the money will exactly match the loss of the seller. Vice versa if it's out of the money. Kind of neat.