Recent content by groovychicken

  1. groovychicken

    Sub-20-millisecond CME data

    @MrMuppet, Thanks a ton. I'm slowly executing on your list of suggestions. Ha! That was my initial thought as well. I guess a InteractiveBrokers account won't cut it. :/ Thanks for the heads-up. The cancels per execution is a new metric that I will have to account for (I am currently not...
  2. groovychicken

    Sub-20-millisecond CME data

    @MrMuppet, great reply! Thanks for contributing and confirming some of my thoughts - notably, that TraditionalFinance will have some price barriers to overcome. You mention it may be in the thousands $ per month... Say I were willing to pay a few thousand per month for the highest speed...
  3. groovychicken

    Sub-20-millisecond CME data

    Comparison of local timestamp when you send the message vs the timestamp reported by the exchange when the message is accepted and executed against (or placed in to) the orderbook. Alternatively, comparison of local timestamp when message is sent vs timestamp returned by exchange/broker of...
  4. groovychicken

    Sub-20-millisecond CME data

    @SunTrader, Thanks for your opinion, but I know what I'm doing. Maybe in your experience you consider the difficulties of negotiating wide bid/ask spreads and portfolio management to be tall hurdles; but for me, those issues are decided, and not the topic of this thread. At this point, if I am...
  5. groovychicken

    Sub-20-millisecond CME data

    >> When it comes to Bitcoin futures (full or micro-size) what you should be concerned about, waaaay before speed, is the bid-ask spread. Thanks, I agree, but let's assume I have that figured in already, and let's assume portfolio management is also already baked in: Now the remaining questions...
  6. groovychicken

    Sub-20-millisecond CME data

    I've been trading a bit over the last few years with custom software I've built for the cryptocurrency space.. mainly on Coinbase. With Coinbase I can get a level-3 data feed with sub-10ms latency (as long as my algo is running in AWS), and I can get execution times on the order or 20ms or so...
  7. groovychicken

    storing level 2 data

    I second the flat-file option. Databases aren't made for this kind of an access profile. If the data is large and you will *always* be looking through the old data linearly, just don't use a database. Separate file for each instrument/day or maybe instrument/hour.
See more
Back
Top