Hello I am using IB for my trades, I want to develop some kind of automated trading algo. for that matter, I want to check my connection to IB servers: Now, as I use CMD, ping: For USA: For EU: Is this 158ms ok? 100ms? I want to trade futures mainly, I dont "fight" another big players algo's but I need proper fills.
What'll be your average holding time? ~200 ms is very normal internet ping times. Not great, but OK and expected. If your holding time is longer than a day, you can probably survive multiple minute ping times as well.
Yes. It all depends on your trading style and objectives, of course, but if that isn't ok for you, one would have to wonder if your style and objectives are reasonable ones in the first place. This thread may interest you (for all that it's old, now): http://www.elitetrader.com/et/index...ng-latency-connection-speed-in-trading.55173/
Unless you are scalping for ticks, speed in milli-seconds is least of your concerns, and if you think you going to be faster than big players who either own or lease seats, think again. What many retail traders don't know is how many outside touches by either your trading platform you have to stay hooked into, your broker-they checking your account size, your clearing firm sometimes then into Globex. And being close to like Chicago, don't mean anything, you can be routed all over the country. Then you have to consider something else, you using cable, DSL and what speed or T-1 dedicated phone line or whatever else that might be faster. I have T-1, not really for speed but the cable goes out often in my area, so I use cable and DSL as backups. I have a bank of batteries in case the electric goes out, I can run for five hours alone on them, and buried in hard ground is generator when batteries get weak. When it gets windy here, electric goes out too. About the best you can do is rent space on a server with 3rd party at the exchange you are trading at or relatively close to exchanges.
Hey, thanks for the comment I am scalping for ticks actually, the whole idea of using batteries appeals to me as well. I am about to hire tech guy to program my idea, I hope 158 ms will fill the right price. (no I dont need to be faster than the big players!)
I ran those ping tests before and they look ok, but I don't know what the ping test would be if I had like Ninja, I use as backup, which gets data from IQFeed then to broker etc etc etc to Globex, I am unaware if you can do a ping test using all the touches.
There are two things here: propagation delay (i.e., physics) and application delay (the time the operating system on the remote server takes to service the ICMP echo request). A more accurate measure would be the time take to service HTTP requests (you can do this with curl). Results of a ping (from Northern Virginia): ra1n@VH-ATLAS:~$ ping gdc1.ibllc.com PING gdc1.ibllc.com (208.245.107.3) 56(84) bytes of data. 64 bytes from gw1.ibllc.com (208.245.107.3): icmp_seq=1 ttl=244 time=17.1 ms 64 bytes from gw1.ibllc.com (208.245.107.3): icmp_seq=2 ttl=244 time=16.0 ms Results of curl, or the time taken for a HTTP request to TCP port 4000: ra1n@VH-ATLAS:~$ curl -so /dev/null -w '%{time_total}\n' http://gdc1.ibllc.com:4000 0.043 ra1n@VH-ATLAS:~$ curl -so /dev/null -w '%{time_total}\n' http://gdc1.ibllc.com:4001 0.043 You can see that the HTTP request is serviced 4x faster.
Ping times from a command prompt on a Windows OS is not a very accurate way to determine latency. Ever.