A very simple strategy, which is not supposed to make money, but to measure the delay in this process :
- order send
- order received by the exchange
- order cancelled
- cancel receive and acknoledge by NinjaTrader
An short limit order is sent X ticks above the ask, to avoid a fill, N times.
Each order is cancelled when his status is "working".
A readable results is shown when the strategy is cancelled.
By default it will send 5 orders, at ask price + 30 ticks.
Warning : do not use it live with volatile instruments during volatile times !
Avoid CL/GC for example, and avoid news period.
I would be extremely interested in having some feedback, with different brokers and locations.
The process, after having simply clicked on "buy limit" is complex, and a lot of parameters are involved, but with enough samples we may get an idea of who is fast and who is not that fast (I already have some thoughts ).
Success requires no deodorant! (Sun Tzu)
The following 15 users say Thank You to sam028 for this post:
@sam028 this is a great idea. I am going to give it a try when I get a chance and will report my results. Hopefully others will do the same as this could be very useful data for comparison.
"The day I became a winning trader was the day it became boring. Daily losses no longer bother me and daily wins no longer excited me. Took years of pain and busting a few accounts before finally got my mind right. I survived the darkness within and now just chillax and let my black box do the work."
Broker: Primary Advantage Futures. Also ED&F and Tradestation
Trading: Primarily Energy but also a little GE, GC, SI & Bitcoin
Posts: 3,968 since Dec 2013
Thanks: 3,258 given,
7,773
received
Interesting subject.
I don't use NT so I'm not sure what your video is showing. Am i right in believing that it is showing the time required to send a cancel request and get a cancel confirm from the exchange on the 5 orders had an average time of 0.729 seconds and a range of 0.55 to 0.92 seconds?
Just FYI, the figures shown in the video not very relevant, as I used my IB paper account, and the machine where the test was done is very busy with 2 extremely heavy/not optimised bots (I'm not the author !!! ).
I will asked some of my dear SpeedyTradingServers users/clients/friends to run it few times, as I may have all NinjaTrader supported brokers/data feed, it could give a good picture of execution times.
Success requires no deodorant! (Sun Tzu)
The following user says Thank You to sam028 for this post:
Ran the test and seen orders go through but don't know how to view the time results like your video. Looked under strategies and orders but must be missing something. Sorry I trade discretionary, not familiar : ) Please advise
I think anyone posting results would be helpful to post location and feed using as will help us compare apples to apples. I plan to run test when get chance. I am on a dedicated machine in Chicago using Continuum feed. I will be real interested to see how that stacks up to a rithmic feed on a machine in Chicago. I have used both rithmic and continuum and can't tell if one has any advantage over the other.
"The day I became a winning trader was the day it became boring. Daily losses no longer bother me and daily wins no longer excited me. Took years of pain and busting a few accounts before finally got my mind right. I survived the darkness within and now just chillax and let my black box do the work."
I like this stuff, but in my view I would suggest to have information by witch broker, software and server you will get the best fills and lowest latency for CL colocated near by the exchange.
Crude oil has a faster fill then ES perhaps the execution time for ES is longer as a result that you have to wait in the queue before you get filled. But I thinking you have also to consider the latency from hawaii to Chicago. So you can reduce the total time with the latency time .
The times showed are the time to be in the queue, not being executed.
The name of the strategy is not a good one, it should be "speedqueue" or "speedbook", not "speedfill" .
As far as I remember @neko33 is using a lot of screens (8 ?), using all these screen in a colocated/RDP environment will look sluggish, too many things to refresh with a > 100 ms latency.
I also consider that from Hawaii, it's not that bad !
Success requires no deodorant! (Sun Tzu)
The following user says Thank You to sam028 for this post:
Speedqueue is a better name I agree it is not the time till a order is filled. But do you have any idea why the speed for CL is shorter than for the ES?
Many of my best trades have been with resting limits at areas of interest anyhow. Really think I am underusing this method of entry. Maybe it's all a blessing in disguise. Either that or go freeze in Chicago and fight with bots
The following user says Thank You to neko333 for this post:
I was at the European Trading Architecture summit here in London in December with the CTO of UBS we were discussing their infrastructure latency, and how for them in particular as well as other HFT/Algo trading businesses they need to know the latency at each point in the IT space. They had at the time invested heavily in tools to exactly measure their latency cost for each interface across their business.
Really interesting was that they price it into their service such that they can quote the latency to their retail / institutional clients and for faster round-trips they charge a premium. Obviously its very useful measure of the IT quality of your service provider.
The following user says Thank You to sands for this post: