has anyone experience NT lagging from R|Trader? just experienced it today. now im not confident trading today.

hey mattz, thanks for the response. i have actually found out that the lagging effect between NT + Rithmic and R|Trader was because my internet was not stable and the platforms had some disconnections.

however, i am interested on the rerouting to the closest server. i think that would help a lot since im currently in the market for looking for fiber optic broadband. internet really sucks here in the PH.

if i get to have data rerouted to/from the nearest server possible, is it guaranteed that data from that server is not far behind from the data directly from Chicago? how many milliseconds is the latency from that server to Chicago? thanks!

I am evaluating different platforms for lower latency (not HFT) API based trading and currently testing Rithmic.

I have understood that the paper account server is not intended to latency/delay testings. It does not represent the real live server latencies which are lower.

Anyway before setting up a real account I did some R|API tests where I sent a limit buy/sell orders for ES contract at CME. I calculated the execution time from calling REngine::sendOrder() to the point where g_bDone flag was set in LineUpdate() callback.

Execution results are quite constantly around 15 ms.

I have a 3 core VPS Windows 2008 server set up in Cermak (thanks to speedytradingservers.com) where the PING gives a value 1ms for Rithmic paper trading server.

Has anybody else done these kind of tests and what kind of results you have received with your paper and LIVE accounts ?

btw. I am not after how fast the Rithmic will execute (I know it is fast) but would like to know the latencies of my whole setup (VPS, communication, application, broker's risk ctrl etc.) how well I am possible succeed in this...

btw2. Linux guys are welcome too. I have not decided the final OS for my setup yet.

btw3. Currently I have 2 canditates: Rithmic and CTS T4. It will be one of them.

One of my former Speedy user did the same kind of test with the R|API but from a Linux VPS, the results were better, between 4ms and 8ms (same VPS location as yours, live account and real trades).
Just FYI on virtual machines the "hardware" clock are not very reliable, but on very short trips like that it should not be too wrong.

Hello Sam and thank you very much for your reply.

I implemented the timing by the QueryPerformanceCounter() and it should work quite well in virual environments but I think you are absolutely right the windows may not be very accurate in timings...

My tests were paper trading account but soon I will test with a live account and real trades. Lets see how it goes then. If it is still double to your Linux values (which I little a bit doubt) then I may switch to Linux. Maybe I will consider dedicated server but it seems your VPS solutions are quite efficient so if I am running just a small strategy box (no chartting etc) then it may be just fine with VPS.

What comes to the communication delays Rithmic support (their server provider sub theomne.net) said their paper trading engine is located in NYC but AMP support said it is in Cermak... So I will rely only real life tests done by myself

It can't be in NYC, you can't cheat with the speed of light .
I also presume Windows or Linux with the R|API should not change the orders latency that much.
It's easy an not risky to test with a real account and a real trade: sent a buy limit order 10 or 20 ticks above the ask, then cancel it when it's in the book and measure how long the process takes.