Execution delay?

[mgin98]

I have a strategy running on a 1-tick chart and I'm seeing 1 second delay for entry after my signal is generated. The strategy does a market EnterLong() then draws a triangleup, however, the actual entry posted is a second later, which often results in a tick slippage. Thinking maybe the market may be too fast for the demo server (Zenfire) to return a fill confirmation, I ran it this morning during the slow period and I'm seeing the same thing. Is this execution "latency" normal due to the messaging functions between NT and the broker or does NT not support millisecond entries? See attached example.

[NinjaTrader_Dierk]

>> Thinking maybe the market may be too fast for the demo server (Zenfire) to return a fill confirmation
Likely, this has nothing to do with any "milliseconds issue" you're speculating about.

[mgin98]

I've changed a few sections around in the strategy, pulled the indicator code out of the indicator and added it to onbarupdate in the strategy and I'm getting the same results. Below is what I'm doing, running on a 1-tick chart and since the triangle is painting at the correct spot, the entry logic seems to be operating properly. What I don't understand is why the entry is a number of bars behind, sometimes up to 20-30 ticks behind.

if (long_sig >= 30
&& Position.MarketPosition == MarketPosition.Flat)
{
EnterLong(DefaultQuantity, "");
DrawTriangleUp("3"+Time[0].Ticks, true, 0, GetCurrentAsk(), Color.Blue);

[NinjaTrader_JoshP]

mgin98,

Suggest you use TraceOrders = true and isolate the order's placement behavior. Actual filling of an order can take several ticks to occur.

[NinjaTrader_Dierk]

Real simple: As you order is sent to the exchange and filled there it could take several milliseconds. During these milliseconds you may get lots of ticks on a heavy traded instrument (like ES).

[mgin98]

Thanks, I think I figured that out today. I created a simple "enterlong after 20 tick closes on the Ask" strategy and ran it on a 1-tick chart. My roundtrip averages 80ms to the rithmic server but the traceorder and a packet sniff shows there might be more to it. I can handle a few milliseconds, but anything beyond a few hundred ms seems to give a tick one direction or another.

However, in the traceorder snapshot attached... are the numbers I highlighted in red... in milliseconds? The time difference is consistently 500+ whatever that time unit represents. If that's the case then I can stop debugging this part of the strategy and will move on to the next part. I'll obviously have to colo if I need better speed.

[NinjaTrader_Dierk]

>> are the numbers I highlighted in red... in milliseconds
yes