Zen-Fire stutter?

[higler]

RJ - I hope so too. You can be sure that that we will let you know. Thanks.

It appears that they are going to be down for a long enough time for a substantial and thorough upgrade. Let's hope for the best.

[higler]

I got started late today because I had some errands to run. Currently, the YM is pretty slow so I can't really coment if the ZF upgrade has helped or not. I have written an indicator which compares the elapsed time for a bar based upon the exchange times (which I think is what the NT Time series provides) and the elapsed time to build the same bar based upon my computer clock. So far, the indicator seems to be working pretty well. When things are running well the two line are very close to overlaying each other. When there is a lag somewhere in the network followed by a data burst of bars then it shows up pretty clearly. The "System" (based on my computer clock) plot will show a much longer bar time on the first "problem bar" compared to the "Exchange" times plot and the next few "System" points will be shorter than the exchange plot because of the "data burst". I've seen only one really bad instance in the last hour where the exchange data built the bar in about 4000 milliseconds but "System" plot showed about 20000 milliseconds followed by very short time frame system plots for the next few bars because of the lag followed by the data burst. This doesn't necessarily say the lag/data burst was a Zen-Fire issue but it does show that there are occasional lags/bursts somewhere over the data pathway. I'm using 50 volume YM bars because they are long enough to give some bar build times but short enough to be fairly responsive.

[higler]

Here is a quick and dirty indicator that I have been trying to use to compare time per bar based on differences in NT bar time stamps based on Time[] vs how long my system takes based on my computer clock. I typically use a tick based chart for this indicator and use it by itself in the chart to try and look for lags/bursts in my system that could result from my computer, ISP, internet delay, or data providers. It may be bogus, but here it is. It plots zero except when data is live. Ignore the first few data points.

[higler]

I won't belabor this any longer. I think that things have improved greatly. I have attached an example of a lag followed by a data burst from friday just to show how the indicator works. There the green line, my system, shows a lag in data resulting in the green line spiking up (time in milliseconds) compared to the orange "exchange" line followed by a burst of very short time bars where the green line dips below the orange line. The example shows that this was consistant accross both YM and ES. I am now using the free PingPlotter software to get also get a general idea of my system network provider performance to make sure that its not my PC or ISP. I also got a very "nice" doublet peaks a few hours later.

[NinjaTrader_Bertrand]

Thanks for posting your indicator and for publishing your findings here!

[BigAAPL]

Nice stuff higler, seems like you are the new hall monitor!