Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
Hi,
I don't believe there was a thread covering this already, so I wanted to punt it out here. A few years back the CME released MDP 3.0 - Market by Order - Book Management which includes a new feed that has quite a few perks to traders.
From what research I have done it seems that the retail industry has been slow to implement this as a lot of the retail data feeds and platforms still don't have this integrated.
I use NinjaTrader and Continuum myself and don't have the ability to receive MBO data for example. So here are my questions at this point.
1. Is there a plan for Continuum / Ninjatrader to integrate MBO data in the near future?
2. Are there any other data feeds that NinjaTrader works with that already have this integrated? I am not sure if the bottle neck is with the data provider, NinjaTrader or both?
3. Are there any other retail solutions (Data Feed + Platform) that anyone is currently using that have the MBO data feed integrated already? If so, I would love to hear how it is working and any specifics you can share.
4. Are there any custom solutions out there that anyone built their self using some type of direct connection to the CME feed. For example:
https://www.cmegroup.com/confluence/display/EPICSANDBOX/CME+Direct+Installation
Am I correct in assuming that any solution that taps the CME feed directly would be out of scope for anyone with a retail level margin account? I know there are $500K barriers to entry for DMA, but I don't know for example if there are any ways to tap the CME data feed without going the DMA route. Any advise here would be appreciated.
I believe Rithmic api exposes MBO, though I have no experience with them.
As far as I know, cme has no plans of discontinuing it’s market by price feed, so I highly doubt mbo will ever make its way to ninjatrader.
I agree that’s it’s mostly out of the scope of retail trading, but the way around the financial barrier to entry would be to join a prop firm that already has DMA.
The firm I work for uses mbo, but it’s difficult to attribute any explicit financial advantage to it over market by price.
There's maybe 3-4 vendors I know who can provide it over WAN even if you don't have an account with a clearing member, but they're also about 10x the licensing cost for NT so the use case is quite niche.
The other problem is that NT's data structures and API will just expose the level deltas to you, so you'll need to use a different platform.
By the way, I have a decent architecture for distributing MBO over WAN but I don't pay the $96k per year for the licensing rights to distribute the data realtime. If there's enough interest for it, I'm happy to distribute it at-cost for the community. I have thought of this idea before, but back of the envelope suggests that I need at least 60 subscribers to breakeven at IQFeed-like rates.
I do have the licensing rights to distribute the historical market data, but don't think that's of much use unless you have your own custom platform and are already connected to CME and just need backfill.
For full disclosure, I'm not a vendor and make no revenues selling data, just mentioning this out here in case there's interest to come up with something together as a community.
Thank you for offering this, even if this is a long shot, I think this would be a great value. I genuinely believe there would be significant interest from the community here, provided there was proper education and everyone understood what this could possibly mean for their trading.
I am sure something like this would work best on an optimized custom platform, maybe not at the FPGA level yet, but to get the full benefit of this type of data feed, the platform would need to be a little faster than most retail platforms. I could envision a small group of those of us interested, working to build something, or re-purpose an existing custom platform for this.
There are definitely performance and speed thresholds that most people in retail trading will never get passed, but something like this, combined with an optimized custom platform would be a great option, for those that want more.
Obviously I don't think most people on here, including myself, would fully have a line of sight on what all would be needed to get something like this off of the ground. But if you would be willing to point us in the right direction, I would definitely try to drum up the interest in the project.
I think getting 60 people would be possible provided everyone clearly understood the value.
Thanks,
Ian
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.
I wouldn't be much help with a full open source platform.
Unfortunately, that's something that hundreds have tried to open source (search github) and dozens have to commercialize (e.g. Quantopian), and it's a competitive space with little room for differentiation.
It's actually very possible to overcome the performance and speed limitations of the current retail offerings.
But I think what would really benefit retail users is ease of use and performance + pricing transparency.
As @addchild mentioned, the key issue is the financial barrier to entry. A 1 year commitment to provide a data service better than IQF will cost $130k+ in hosting and licensing costs, up to $36k in insurance and legal costs, and while it can be a part time endeavor, at least 3 man months of time from a developer who quadruples as a sales, web development and support person. It's at least $230k. My barrier to entry is probably half of that since I already have most of the above, but it's still a sizable endeavor.
The technical difficulty is mostly one of
i. stability - what server architecture and application design will keep high % uptime, followed by
ii. scalability - how can your software handle hundreds of concurrent connections over WAN, and finally
iii. the messaging architecture between the colo facility and your public gateway - which could be the same host, but that's probably undesirable.
The typical retail options, to their credit, seem to have put some consideration into the first two issues, but not the third. You mentioned Continuum for example, which already pays significant latency and throughput cost because of their protobuf serialization layer. I wouldn't design something like that.
My implementation actually does sit behind 2 FPGAs. It's overkill for the use case, but it's already there for other purposes. By that very virtue I can probably get it faster than $5k/month data feeds that Bloomberg/QuantHouse might sell.
I guess another problem is that the business model wouldn't make sense.
Most of the institutional vendors I know focus on providing only the software licensing, this greatly reduces their own operating costs and at the same time they're able to breakeven with just 2 corporate customers paying $10k/month.
Your costs are actually higher to run a public feed because you are running infrastructure, and managing 100+ retail customer relationships is a lot more taxing than 2 corporate clients.
The only way you can make this profitable is to land a sweet licensing deal with a big firm, like Yahoo.
To elaborate on the 'performance transparency', what I mean is it's hard for retail users to understand the consequences of where their computer is located, how much bandwidth they need, what's the best practices for speed/uptime, how realistic their backtests will be with the production data that they're receiving etc. Retail vendors are extremely opaque about this, maybe because their numbers will actually be quite atrocious, and maybe because they aren't actually sophisticated end users of their own data.