NexusFi: Find Your Edge


Home Menu

 





Rithmic Latency Calculations (Plus trading remotely)


Discussion in Platforms and Indicators

Updated
    1. trending_up 6,852 views
    2. thumb_up 9 thanks given
    3. group 3 followers
    1. forum 5 posts
    2. attach_file 0 attachments




 
 

Rithmic Latency Calculations (Plus trading remotely)

 
 dsheehan87 
Chicago IL USA
 
Experience: Advanced
Platform: Ninjatrader
Trading: ES CL SB
Posts: 14 since May 2020
Thanks Given: 3
Thanks Received: 18

Ok, first off, currently using jigsaw for trading through rithmic. I have a two part question…



1) Latency calculations



I am not sure if I am understanding the latency measurement within r-trader pro…



I have 2 locations that I can trade.



The first location, measured through r-trader, I am seeing between 15,000-30,000 μ (mu) for orders and prices/data, which if I am converting correctly, is equal to 15,000,000-30,000,000 ns which is equal to 15-30 ms?



The second location, measured through r-trader, I am seeing between 1800-3600 μ (mu) for orders and prices/data, which if I am converting correctly, is equal to 1,500,000-3,000,000 ns which is equal to 1.5-3 ms? This is a vm as well, so possibly latency could be higher (unlikely though)



***update***

Just tested from a physical server instead of a vm at location #2, I’m getting 800 μ (less than a ms?) for price data and under 2000 μ for orders…

Am I calculating this correctly ^ (second location is within the same datacenter as CME Group)



1000 μ should be 1 ms… pretty sure. But, Is the lower latency of any advantage for market depth trading (not hft but manual order entry on the DOM)? I understand the advantage for a HFT, but is there any advantage for a scalper in the treasuries?

2) Remote Trading:



This second location isn’t somewhere I can sit and trade, it houses servers used for other purposes. If I were to put another 1u server there, and connect to that machine via VPN or other means (from 2 miles away), do you think this would work?



Any advice or assistance would be most appreciated!!!

Started this thread

Can you help answer these questions
from other members on NexusFi?
NexusFi Journal Challenge - April 2024
Feedback and Announcements
The space time continuum and the dynamics of a financial …
Emini and Emicro Index
Better Renko Gaps
The Elite Circle
New Micros: Ultra 10-Year & Ultra T-Bond -- Live Now
Treasury Notes and Bonds
Deepmoney LLM
Elite Quantitative GenAI/LLM
 
Best Threads (Most Thanked)
in the last 7 days on NexusFi
Get funded firms 2023/2024 - Any recommendations or word …
61 thanks
Funded Trader platforms
38 thanks
NexusFi site changelog and issues/problem reporting
27 thanks
GFIs1 1 DAX trade per day journal
19 thanks
The Program
18 thanks
 
 
glennts's Avatar
 glennts 
Corpus Christi, TX / Westcliffe, CO
 
Experience: Advanced
Platform: NinjaTrader
Broker: DDT / Rithmic / Kinetick / IQ
Trading: 6E, ES
Posts: 420 since Oct 2010
Thanks Given: 24
Thanks Received: 1,022

If you plan on manually executing trades while using a co-located server to take advantage of the low latency you need to keep in mind that you are essentially now using a web based trading platform. You will still have the latency issues to deal with that exist between your desktop and the server. Let's say it is 50ms. You move your cursor and 50 ms later the cursor on the server based version moves... and 50 ms after that you observe that cursor movement... that gives you a 100ms delay to seeing your screen react to your cursor movement. Now you place an order and there may be two or three back and forth journeys before the order is actually sent for execution at 1.3 ms. You adjust your entry price... more back and forth at 50ms each way needs to take place before anything actually happens.

Remote co-located servers are best for hands off, fully automated trading that does not require your participation. That is the only way you can benefit from the 1.3-3 ms latency.

This is based on my experience spending a few months trying discretionary trading with a co-located server.

If you have gamed online with lag issues... that's what you get.

I use the Rithmic data feed for order execution and if I hit the bid or ask I get an instant fill, standing stop orders get an instant fill occasionally with 1-2 tick of slippage but that is a market depth issue and not a latency one as standing orders are held on Rithmic's co-located order execution servers. Which, as I understand your post, is what you want to achieve.

Thanked by:
 
 dsheehan87 
Chicago IL USA
 
Experience: Advanced
Platform: Ninjatrader
Trading: ES CL SB
Posts: 14 since May 2020
Thanks Given: 3
Thanks Received: 18


Thanks… Good thing is, I will be able to route directly into this server from my office, so if I’m in the building I can at least get this type of latency, and from what I’m seeing now on a physical machine, < 1 ms for price data. (More incentive to leave my house)

Other good news is that doing some tests, I get roughly 8-10 ms to this location from my office at home (much faster than going to rithmic directly - which is weird). Maybe some weird Chicago routing thing.

So, theoretically, trading remote through a vpn directly into the network, I should see 10-15 ms (less or equal to what I see here).

What I’m after really is to be able to see the orders hit the book even a fraction faster than someone sitting elsewhere. The problem is that many times when a trading opportunity arises, it unfolds over several minutes, but the key is to get a limit order into the book at the opportune time. Sometimes, you know a trade opportunity is unfolding, and when it looks like the market is holding, you tap the bid/offer, and you are too high in the queue because others beat you to it, and you don’t get the fill (I never, ever buy the offer, if I’m buying I’m buying the bid or nothing - same for shorts). Slippage isn’t a thing for me really, as I always exit on limit orders…




Sent using the NexusFi mobile app

Started this thread
 
 
glennts's Avatar
 glennts 
Corpus Christi, TX / Westcliffe, CO
 
Experience: Advanced
Platform: NinjaTrader
Broker: DDT / Rithmic / Kinetick / IQ
Trading: 6E, ES
Posts: 420 since Oct 2010
Thanks Given: 24
Thanks Received: 1,022

I've not looked at JigSaw in quite awhile but I recall it would show where your order was in the queue if that is important to you. Not being familiar with your methods or knowing the instrument / time frame you are trading I can't comment on the importance of shaving ms off of time stamps. Not sure how, if you place a 10 contract limit order in the ES for example where the order book numbers are rapidly changing up and down in milliseconds, that you would know any change is a result of your order landing without being able to flag your specific order. You place 10 while someone else pulls 10 and then along comes a 7 and a 3 in the blink of any eye, you see a net gain of 10 that lasts for a fraction of a second but it's not your order. I don't think perfection in order book information flow is possible, at least to the extent that you want to manage your trades. I also don't think that it matters. Ultimately you are either right or wrong with your trade and if a few milliseconds or 1 tick+/- on a fill is the difference between success or failure I suggest there are other things you need to focus upon.

Thanked by:
 
 dsheehan87 
Chicago IL USA
 
Experience: Advanced
Platform: Ninjatrader
Trading: ES CL SB
Posts: 14 since May 2020
Thanks Given: 3
Thanks Received: 18


glennts View Post
I've not looked at JigSaw in quite awhile but I recall it would show where your order was in the queue if that is important to you. Not being familiar with your methods or knowing the instrument / time frame you are trading I can't comment on the importance of shaving ms off of time stamps. Not sure how, if you place a 10 contract limit order in the ES for example where the order book numbers are rapidly changing up and down in milliseconds, that you would know any change is a result of your order landing without being able to flag your specific order. You place 10 while someone else pulls 10 and then along comes a 7 and a 3 in the blink of any eye, you see a net gain of 10 that lasts for a fraction of a second but it's not your order. I don't think perfection in order book information flow is possible, at least to the extent that you want to manage your trades. I also don't think that it matters. Ultimately you are either right or wrong with your trade and if a few milliseconds or 1 tick+/- on a fill is the difference between success or failure I suggest there are other things you need to focus upon.



In the treasuries it’s different… the bid for example, when the market isn’t so volatile, they will hold at a price for minutes on end, but you can see the icebergs and other traders refreshing their orders. They key thing with that is I try not to join that bid price unless I see them also hitting up into the offer (other traders obviously know this as well). Even a blink faster is a huge help. It’s a matter of missing a trade or not really. If I’m late to the game, I won’t get filled, and the market will take off without me (and then 99% if the time it’s too late to enter because I can’t get the edge… and if I can’t trade the edge, I don’t want to be in the trade, if I can trade the edge, I can scratch the trade or make a tick profit as soon as I’m filled, if I buy the offer, I’m already losing).

It’s the same as it was on the floor… you buy the bid and you immediately try and sell the offer. Tougher on the computer but very possible.

I trade for 1-3 ticks (more if the market is moving, I never use targets). So, yeah, if I miss my entry, it’s a coin flip to enter honestly. This is impossible, like you said, in the s&p (though it was possible back in the day for sure, but that market is now too thin, which is why it moves so fast now). At the moment this is tough to do in the treasuries, easy to get crushed, way too thin and volatile because of the war…


The whole reason I thought about this latency thing, is because Last summer, I did a test with cqg in the datacenter, vs rithmic at home (was at home using a remote session), and cqg was ahead of my dom at home (by like a half second maybe, noticeable) even with the remote session. I thought that odd, but just thought maybe that day cqg was faster. I am 1.5 miles from CME and 1.5 miles with the main Datacenter in Chicago (it’s not aurora as people think) which is where my servers are (but not in CME’s space, with another DC but they live on the same backbone…. Now, since trading is brutal, looking for any edge I can get until thing get back to normal lol.



Sent using the NexusFi mobile app

Started this thread
 
 iantg 
charlotte nc
 
Experience: Advanced
Platform: My Own System
Broker: Optimus
Trading: Emini (ES, YM, NQ, ect.)
Posts: 408 since Jan 2015
Thanks Given: 90
Thanks Received: 1,147

The latency you are seeing from Rithmic is the latency between them and the exchange. As long as you are co-located well, this part should be the least of your concerns. The bigger impacts to your latency are:

1. The bid and ask feed has a timestamp down to the microsecond from the exchange. You can test the delta between this timestamp and when your application receives the update (well depending on what you are using, and if you code it to produce this output). I am also co-located in the same building as the CME matching engine, and I built a system using Rithmics API and I tested this latency, and depending on how volatile the market was, this was the largest aspect of my latency chain. I am not sure what the UI is like for R-Trader, but to give you an idea other retail applications I tested were around 250 milliseconds behind the exchange time stamps.

2. Next you are likely hitting 25 to 100 milliseconds processing your application and converting the raw data feed to your computer screen. I don't know how slow R-Trader is, but given that it has to render data to a UI using some type of IO, there is a cost here that will likely be your second biggest problem.

3. Next you have to physically react, click a button and get an order from your keyboard to your systems order routing section of code, so add another 50 to 200 milliseconds here.

4. Then Rithmic will take your order and send it to the exchange in the time that you have been seeing quoted.


I am sure you know you are bringing a knife to a gun fight here already, but I don't know what you could do to be faster, smarter here honestly. Most algorithms place orders WAY ahead of time at every price level, and use the MBO data feed to see their exact position in the queue, and they only trade orders once they get to an ideal spot, else they cancel. And they have the luxury of waiting until the last possible second to cancel. You likely couldn't land a cancel unless you were pretty far away from the top of the book.

As long as your alpha isn't latency sensitive, you may be able to pull something off and the added co-location may help. But scalping the top of the book really is exclusively the domain of HFT unfortunately.

In the analytical world there is no such thing as art, there is only the science you know and the science you don't know. Characterizing the science you don't know as "art" is a fools game.
Visit my NexusFi Trade Journal

 



Last Updated on May 6, 2022


© 2024 NexusFi™, s.a., All Rights Reserved.
Av Ricardo J. Alfaro, Century Tower, Panama City, Panama, Ph: +507 833-9432 (Panama and Intl), +1 888-312-3001 (USA and Canada)
All information is for educational use only and is not investment advice. There is a substantial risk of loss in trading commodity futures, stocks, options and foreign exchange products. Past performance is not indicative of future results.
About Us - Contact Us - Site Rules, Acceptable Use, and Terms and Conditions - Privacy Policy - Downloads - Top
no new posts