sysot1t 

very nice... how about some tests to your TT Gateway @ CROSSLAND, that would be an interesting one to see what latency exists over that fiber network as it exits off your vendor ... better yet, get VisualRoute and map it out, curious to see what EU is like in terms of latency within their networks before the cross the ocean...

While running the test, the shorted RTT was 6,1ms and the longest one was 272ms. That's to crossland.

That pingtest seems interesting. I use MBT with NT for charting and Oanda for entering trades. Is there is a specific server I should connect to through pingtest to test how low/high my ping is with MBT and Oanda servers?

Agree with the other posters that Ping time or latency is the most important issue.

But ZenFire don't let you Ping their data servers to test this.

However, don't forget that the web access version of ZenFire does have a latency reading in the top right hand corner. From memory (I am at work) there is a green 'on' light and if you hover your mouse pointer over it you will get the latency reading from the actual ZenFire data server.

Now this may not be exactly the latency you will expereince through NT, but in the past when I have had Web Zenfire up at the same time as a Zenfire connection through NT the signal/trade data is almost identical - sometimes the web is slightly ahead and other times NT will be slightly ahead.

For MB use the server in Los Angeles. Oanda use NY.

I'm using NT7 and MB Navigator on a W2K Server on a VPS from Solar VPS that I just set up recently. I just did pingtest on it and it was 164ms & 89ms jitter. A D rating! So I opened a support ticket and told them they better clean up their connection or they'll have one less customer in the very near future...

Although, I have been trading on this system for several days without any latency related issues, it's totally responsive. I click(or it hits my limit), it executes... I don't do auto-trading though, it's all manual.

My connection at home is an A!

I can't resist to show my very cheap VPS results (W2008R2/1Gb/bi xeon 2.93ghz, $40/mo)

I'm sure I'm the worst of the bunch here, given my location in a canyon, far from the DSL CO. I have a 3MB service, but get half that typically... But, as some said, latency is more important. Well, that seems to be not so good either.

If I hit a server near home (SF Bay Area):

Ping = 70ms
Jitter = 25ms

To Chicago (OEC):

Ping = 136ms
Jitter = 48ms

I suppose I should not consider scalping with such latency? What, if anything can I do improve this situation, short of relocating?

The only, but I really mean only thing that could help you is to run an automated trading system at a hosted server in Chicago.

Using a virtual server will not help since it will take you the time to see the data coming from that virtual server to your screen.

If you trade on ninjatrader, you also have to add the minimum 100ms delay that is build into ninjatrader on the charts before it shows any data.

One of my good friend is sclaping, with very good results, with a 300ms delay (he lives in an Pacific island).
The real question is your definition of sclaping. You can't fight with HFT sclapers, where the unit is the microsecond, and not the millisecond, but for "human" sclaping, less than 300ms is OK, IMVHO.

TonyB 

I'm sure I'm the worst of the bunch here, given my location in a canyon, far from the DSL CO. I have a 3MB service, but get half that typically... But, as some said, latency is more important. Well, that seems to be not so good either. If I hit a server near home (SF Bay Area): Ping = 70ms Jitter = 25ms To Chicago (OEC): Ping = 136ms Jitter = 48ms I suppose I should not consider scalping with such latency? What, if anything can I do improve this situation, short of relocating?

Thank you MetalTrade and sam028.

In my few weeks of paper-trading, scalping for me has been on the order 15 seconds to 1 minute rt's, generally a couple pips of movement...

I use MC, as opposed to NT, so it sounds like that could be helpful. I'm a long, long ways from running an automated trading system, but if done with MC, it would seem to be as MetalTrade described - on a Chicago area server. If so, that would ideal.

I ran this ping test again moments ago, to SF:

Ping = 33ms
Jitter = 3ms

To Chicago:

Ping = 96ms
Jitter = 15ms

The difference that accounts for this improvement, not sure, but my wife is not on the internet now. It's 8:30 am, as opposed to 8:30 pm. And I did call my ISP regarding my bandwidth, namely upload speeds at 0.3 MB! It had been much higher before...

I ran these tests on my laptop, over a wireless connection. Also, we have an Ooma IP telephony unit in the mix: DSL modem->Oooma->Apple router.

Thoughts? Get one more DSL modem, and go direct (ethernet) to my soon-to-be dedicated trading PC. Split the modem output into two, one to the current path, and the other direct to the trading computer...