Automated trading through API. Need advice.
- Thread starter forse
- Start date
They are two different front ends so you would need to read each of the API documentation.
When I auto traded on Sterling it was 2009/2010 and a lot has changed since then. The biggest issue I had was the number of orders per second and that the front end took about 120ms to populate a new symbol locally. Meaning if you have not yet traded that symbol for the day it took 120ms for the local database (installed on your computer) to populate the information. Overall we found Sterling to be stable and reliable, as long as we accepted the limitations.
More recently I was using Laser and found it a little buggy. It would crash for unknown reasons and it was a bit of a pain - but, we were using a whitelabel or re-branded custom version of Laser, not just the stock canned laser. The firm we were executing through had a deal where they rebranded Laser so for all I know something could have been de-stabilized that way... or it could have been our code and programming. We ended up just using Laser to watch positions and P&L and for some manual trading while the auto system was colo and sent orders directly to the execution engine, not through Laser's API. Even without pushing orders to the Laser API we still found it buggy and a bit of a pain but again that could have been the version we were using.
When I auto traded on Sterling it was 2009/2010 and a lot has changed since then. The biggest issue I had was the number of orders per second and that the front end took about 120ms to populate a new symbol locally. Meaning if you have not yet traded that symbol for the day it took 120ms for the local database (installed on your computer) to populate the information. Overall we found Sterling to be stable and reliable, as long as we accepted the limitations.
More recently I was using Laser and found it a little buggy. It would crash for unknown reasons and it was a bit of a pain - but, we were using a whitelabel or re-branded custom version of Laser, not just the stock canned laser. The firm we were executing through had a deal where they rebranded Laser so for all I know something could have been de-stabilized that way... or it could have been our code and programming. We ended up just using Laser to watch positions and P&L and for some manual trading while the auto system was colo and sent orders directly to the execution engine, not through Laser's API. Even without pushing orders to the Laser API we still found it buggy and a bit of a pain but again that could have been the version we were using.
Yes, I understand that these are two different API.
API Laser bit confused, a little information on the Internet and, judging by the reviews, does not work very stable.
Sterling operates through the API more stable than the laser? I understand you correctly?
If one can work through the IPA Sterling, then why do we need laser API?
"Overall we found Sterling to be stable and reliable, as long as we accepted the limitations."
What are these limitations?
API Laser bit confused, a little information on the Internet and, judging by the reviews, does not work very stable.
Sterling operates through the API more stable than the laser? I understand you correctly?
If one can work through the IPA Sterling, then why do we need laser API?
"Overall we found Sterling to be stable and reliable, as long as we accepted the limitations."
What are these limitations?
it's been a long time since I've used Sterling so hopefully someone else can chime in.
I believe that there is a limit of something approx 100 orders per second on Sterling. I really forget so I shouldn't even be guessing or saying that. I do know that sterling has limits to either a specific overall number of orders per second or a a set number of orders every so many milliseconds. For us that meant if we were basket trading and trying to send out 100-400 orders at a time it would not allow that so we had to have multiple accounts (and pay multiple account fees) to run the same strategy spread across multiple accounts. You'll have to read the API documentation to find out about the number of orders it takes and how often, etc.
The other limitation was with risk and the platform's database. Each time you traded a new symbol it took up to 120ms for the ticker to be run through risk (to check if you could trade it) and then be populated onto your local machine. This means if you are pairs trading or trading news (like oil & gas numbers or economic events) then you had to pre-trade each symbol you may use (daily) just so that it would be populated or else you uwould get crushed trying to trade with a 120ms delay.
Hope that helps a little bit. You should be able to call up each of Sterling and Laser and request the API documentation to clearly see these things. That would be much more accurate than me trying to remember from years ago.
I believe that there is a limit of something approx 100 orders per second on Sterling. I really forget so I shouldn't even be guessing or saying that. I do know that sterling has limits to either a specific overall number of orders per second or a a set number of orders every so many milliseconds. For us that meant if we were basket trading and trying to send out 100-400 orders at a time it would not allow that so we had to have multiple accounts (and pay multiple account fees) to run the same strategy spread across multiple accounts. You'll have to read the API documentation to find out about the number of orders it takes and how often, etc.
The other limitation was with risk and the platform's database. Each time you traded a new symbol it took up to 120ms for the ticker to be run through risk (to check if you could trade it) and then be populated onto your local machine. This means if you are pairs trading or trading news (like oil & gas numbers or economic events) then you had to pre-trade each symbol you may use (daily) just so that it would be populated or else you uwould get crushed trying to trade with a 120ms delay.
Hope that helps a little bit. You should be able to call up each of Sterling and Laser and request the API documentation to clearly see these things. That would be much more accurate than me trying to remember from years ago.
There is such a wide range in what people consider automated trading that it would be very hard to recommend you a platform without knowing what requirements you have (latency, data, securities traded, frequency of trades and style of trading/strategy).
TWS works great for a lot of automated traders. I don't think there are any better or worse platforms out there - just some are more or less appropriate for the circumstances.
Without posting a little bit about what your intent is only you can decide what is appropriate or not.
TWS works great for a lot of automated traders. I don't think there are any better or worse platforms out there - just some are more or less appropriate for the circumstances.
Without posting a little bit about what your intent is only you can decide what is appropriate or not.
If you're going to work through IPA, I'd recommend one with a high IBU -- and even then I don't think you're going to make any money through it
.In considering automated trading systems, you're proposing a pretty involved endeavor and probably need to do a lot more homework before asking questions like those you're attempting to pose here. Google works wonders in this department -- "sterling api", "laser api", etc.
If you're serious about automated trading, FIX may be your best option. It's "broker neutral" and has far more robust features (retransmission, etc.,) than "platform" API's, which are usually just a library that hooks into a pre-existing, and perhaps unstable, point-and-click "trading platform".