On July 31, 2017, TradingSchools.Org published a blog post titled “Battle Of The Datafeeds.”
The blog post asked a very simple question – which Futures trading data feed and order routing was the fastest for the ‘average’ retail trader?
Many folks probably do not put much thought into which data feed they select. Instead, they put more thought into which Futures brokerage they should use. And more importantly, which Futures brokerage is offering the cheapest commissions.
We all want to pay the lowest commissions for our Futures trades. In fact, I have yet to find a Futures brokerage that did not swear to Jesus that they, in fact, are the lowest priced Futures brokerage on planet earth.
However, once you have selected a Futures brokerage, the Futures brokerage is going to ask you a very important question… “which data feed and routing do you want to use?” Most folks react to this question like a deer gazing at the headlamps — as the car is wildly moving in its path.
It’s an important question. And truth be told, brokers always proclaim that “Datafeed A is better than Datafeed B” etc. But how does the broker really know which data feed and routing is the fastest? Do they have any empirical data to offer? Nope. At least, I have never found a broker willing to offer any such data.
What is the Futures Data Feed and Routing?
After you open your Futures trading account, you will have to select data feed and routing. What does this mean?
Datafeed is the raw data that is being delivered to your computer. It’s those numbers that slide across the screen, quickly posting prices higher and lower. Nobody wants a slow data feed. Imagine for a moment if you are making your decisions based upon data that is 1 or 2 seconds slower than everyone else. This puts you at an information disadvantage. Trading is a game where everyone is jockeying for the most relevant data to base a decision.
A slow data feed is akin to making your investment decisions based upon yesterday’s newspaper. It is pretty much useless. So, you need fast and fresh data without interruption.
Additionally, you also need a fast order routing pipeline. What does this mean? When you place your order on your trading screen, where does that order go? Have you ever really thought about it? Does it go to the broker in Florida where its parsed for “risk management” and then forwarded to an internet pipe in Los Angeles, and then is finally sent back to the Chicago Mercantile Exchange in Aurora Illinois? These are important questions. Remember, the faster your order can hit the order book at the Chicago Mercantile Exchange, then the faster it can be filled.
Getting your order filled at your price is devastatingly important, especially if you are day trading. If your order entries and exits are even a fraction of a second slower than everyone else, then you are pretty much fucked. You are the last guy in line at the buffet. By the time you get to the unlimited shrimp cocktail, it has disappeared. And all you get is the unlimited orange chicken. Not cool.
So now, we have talked about data feed and order routing — for Futures contracts. Lets now talk about a recent test that TradingSchools.Org performed, in hopes of sussing out the truth.
Testing “The Big Three” Futures Data Pipe and Routing
For our test, we thought it appropriate to test the most popular, in terms of retail traders. Yes, we could have tested some of the smaller brands, but their market share was either too tiny or too expensive for the average user to contemplate using.
With this in mind, we tested Rithmic, Interactive Brokers, and CQG. These are undoubtedly the largest players for the common retail trader.
The next step was to create a controlled environment. This controlled environment included three brand new laptop computers purchased at Walmart. Nothing fancy here. In fact, each laptop was purchased at $150 each.
Once we had the fresh laptops on the table, we installed the Trade Navigator program, which would be used to execute the trades.
A very simple trading system
Next, we created the worlds most simple automated trading strategy.
The strategy was as follows:
- At the open of any 1-minute bar, enter on a limit to buy one Emini SP500 futures contract, at 2 ticks below the open of the one minute bar.
- Exit the trade, on a limit of 2 ticks above the fill price.
- Repeat the process every 1-minute.
In essence, we were simply wanting to see if Rithmic, Interactive Brokers, or CQG could “pick off a tick” while the others were not able to get the order filled.
For the test, we executed 300 trades.
The results
First, let me state for the record…this is not a truly controlled scientific test. There are a myriad of factors that can, and do, influence a data feed and order entry system. These results really need to be taken “with a grain of salt.”
A true scientific test would have required that we house a server from within the data center inside the Chicago Mercantile Exchange’s Aurora data center. However, the cost would have been thousands of dollars and would have required thousands of trades being executed. The more data you have, the more reliable the statistical inference. I don’t have that sort of money.
Instead, this test should be taken as from an unsophisticated moron operating in his garage. (which is true)
So what did we find? Over the course of executing 300 trades…(drum roll please)…we discovered the following…
There was absolutely no difference between Rithmic, Interactive Brokers, and CQG. Each provider was able to deliver the exact same result.
Isn’t this just depressing? You read all the way through this article, and you were expecting some sort of grand climax. And all you got was a pithy draw.
Which should you choose?
Considering that we could not find a real difference. The question then becomes, “If they are the same, why should I spend a higher fee for data and order routing?”
Yes, the fees between all three are comparatively equal, so then the question becomes, “which broker gives me the best service?”
And now, we are back to ‘square one’. Thanks for reading. And sorry for the unclimactic ending.
I have CQG data data feed and they have the worse data feed in the World. Their feed doesn’t refresh, knowing that this is the right data, I put my trade with the information and most of the time I loss big time. Ninja Trader 8 Support gave this to me as a new trader, knowing that I could not win with this kind of data feed. NT8 support is partly to blame on my losses for giving this worst kind of data feed.
Thank you for the work!
LOL. You’re looking at ONE-MINUTE charts instead of millisec-level (tick) charts. Of course your strategy or broker doesn’t matter because your time scale is so wide. (Any dull knife will cut a slice of butter). If you tried a more time-sensitive strategy, like scalping, arbitrage, momentum detection, then HELLZ YEAH there’d be a huge difference. And if your Windows begins to download updates or using virtual memory, your algo will be delayed even further. Please do your homework.
The test is useless somewhat since it ignores the hardware limitation. Why did you buy cheap laptops? I know after the test you would take them back to Walmart and get a refund anyway so you should have selected the MOST EXPENSIVE and FASTEST laptops!
Well, the test was so unscientific that it really didn’t make much sense to purchase $2,000 laptops.
In order to properly test, it was not the hardware limitations and software limitations…because even if they are $300 laptops vs $2k laptops — theoretically, they would have all performed the same, if matched. And besides, most people are using cheap laptops, so it just made sense.
A proper test would have been to test all the vendors over a longer period of time, over high-stress market conditions. But again, it’s hard to predict when “high stress” situations are going to appear.
As I discovered, I just don’t have the massive amounts of money to blow on commissions for such a test. I was looking for a quick and dirty test.
At the end of the day, we are retail traders, and the test provided enough information to support the conclusion that the data feed is not really going to make a huge difference.
But if I had to really get dirty, I would say that Interactive Brokers Futures feed is the worst. All of the orders are routed from Aurora to Greenwich data centers, where the legacy providers like Rithmic are housed inside Aurora, or very near Aurora.
Hope that helps.
Thanks for taking the time, cost and effort of comparing these datafeeds, Emmett. Indeed I’d agree it’s not so much the speed difference in the end but perhaps what each broker that uses these feeds provide in terms of other service features, such as offered instruments and amount of leverage for each, and maybe which platforms are supported etc. In general I’d say HFT still screws the retailers as they’ve taken over since 2009 especially during fast moving markets and sudden volatile moves where there can be more slippage as your market orders are frontrunned. But other than that most retailers don’t have to worry about speed and slowdowns (unless there’s an unexpected systemic problem. like there were reports CQG had been stalling a certain time of day a few months ago where hopefully it’s all fixed now and should never occur again)
Could we have this study also on Options trades?
This is a good point. While this article covered futures’ feeds well, I think tradingschools should do a study/article on how stock market brokers are offsetting their lost revenue from the “zero”-commission conformity with making deals to sell customer order flow to market makers and HFT. (https://seekingalpha.com/article/4205379-robinhood-is-making-millions-selling-out-millennial-customers-to-high-frequency-traders)
(https://medium.com/@efipm/zero-commission-brokers-selling-order-flow-are-the-new-intermediaries-who-will-disrupt-them-5e79628b32f1)
While legal, it stinks and sordid similarly to the pump and dumping scammers. Now even daytraders on TDA are complaining about bad or delayed fills during “busy” times likely when HFT is more active.
And it’s irksome to see recent articles by our dumper sham Ross on the subject continuing trying to pretend to be some trading authority with integrity. blech.
What does IQFeed have to do with order execution??? IQFeed is just a market data provider.
When I think of best data feed the first that comes to mind in IQFeed. Didn’t even know CQG was still in business.
IQFeed is a good one. However, after running the test, I came to realize that any differences are pretty minor.
The real question, should you switch if anyone drops their price? Example: What if IQFeed, Transact, etc. were to drop there fees from 50 cents per side to 5 cents per side? I would be all over that.
For higher volume traders, I think that Tradovate offers a great deal. Suppose you create a high-frequency trading system that attempts to scalp single ticks all day long, you could go with Tradovate and pay absolutely no commissions or data fees — only $199 per month. Unlimited trades. And I know that Tradovate uses CQG.
Tom, IQFeed is not an “execution” data feed; CQG, IB and Rithmic are. You cannot compare two different things.
300 trades trying for 1 point per trade. So what are the stats? Profit, winrate, profit factor, etc.
Hi Cyn,
This “strategy” wasn’t actually a trading strategy meant to earn any profits. It was only a test to see if there was any difference in quality of execution.
I simply wanted to see who was the fastest. It turned out to be a three-way tie.
Yes, it was a waste of trading commissions, but nobody has ever attempted this sort of test.
However, the test does provide some value…suppose that Rithmic drops their pricing vs CQG, then it would make sense to switch, without having to worry about quality of execution.
Yes, there was a small profit.
However, this is mainly attributable to the upside bias in the SP500 Futures contract.