Hi, I am using nodejs to place orders and my apps are hosted on aws and heroku (north virginia regions).
I am experiencing about 200-350ms response time from the alpaca API when fetching price data and account data which results always in a total execution time of arround 600ms to 1s. I was wondering what are your average response times and where do you host your apps? I get roughly the same results when using my own implementation using axios.
Users have been reporting trade data latency on the order of 150-300ms. That is the time from the participant_timestamp (ie the time the trade executed) to when it is received via a web socket stream. That is greatly dependent upon the network. Internally, Alpaca typically sees about a 20ms latency. Of course our servers are colocated on the same premises as the SIP provider and use high bandwidth connections. Submitting orders has generally been reported as taking between 120-250ms. This is by no means authoritative, but I did find several references when googling “API response time” and found this one (others were similar) stating a 200-350ms response was considered fast.
It would be good to hear what other users have experienced.
If I want to colocate a server as close to Alpaca’s datacenter as possible, to minimize quote and order latency, which datacenter would that be?
The Alpaca data servers are located in New York. The account servers are hosted by Google at their us-east4 data center North Virginia.
The data servers are colocated on the same premises as our Securities Information Processor (SIP) where Alpaca gets all market data. It’s a private datacenter.
The account servers are located close to the market partners on the east coast in a public Google Cloud datacenter.
Maybe go with Google hosting in their us-east4 datacenter. Some of our customers have chosen to go that route for the same reasons you asked about.