Lowering commissions: Traits to look for in a broker/platform for Hi Freq trading | Traders Hideout


futures.io - futures trading strategies, market news, trading charts and platforms


Traders Hideout


Discuss day trading practices and futures trading strategies on this forum for all markets. This forum is also for discussing and reviews for brokers, data feeds, and commercial or third party add-ons




 

Lowering commissions: Traits to look for in a broker/platform for Hi Freq trading

  #5 (permalink)

Chicago
 
Trading Experience: Intermediate
Platform: NinjaTrader
Broker/Data: NinjaTrader Brokerage
Favorite Futures: FDAX, NQ, HG, SI, CL
 
Bullywig's Avatar
 
Posts: 53 since Oct 2015
Thanks: 32 given, 16 received


iantg View Post
Hi Bullywig,

Sure I can give you some information in this area. This is the the world that I am attempting to play in currently .

1. First there are discounts you can get from a broker for hitting a certain volume. For example:
https://www.interactivebrokers.com/en/index.php?f=commission&p=futures2
MB Trading | Offered by Ally Invest

Not every broker will have this published, but there is information out there if you look, or call and ask.

2. The real savings though is going to be getting a seat license at the exchange. There is a great thread here that covers this in detail. https://futures.io/emini-index-futures-trading/7376-becoming-exchange-member-leasing-seat.html But generally speaking for a few thousand dollars (non- refundable up front) and then 300-400 dollars per month depending, you can get your all in trade cost reduced by about half.

3. Also the exchange has a volume tier program of their own. The CME gives you discounts after like 75 trades per day, and further discounts come in for the super high volume players.

I think the net savings you can get with all three items I mentioned will get your all in cost < $2.00 per trade for the CME.


Something to know regarding the HFT world: You will be at a huge speed disadvantage over pros that play in this space, so you will best case only be fast enough to catch bread crumbs. So the first thing you will need to do is co-locate near your exchange of choice / data provider / broker, depending on the specifics. This will help your execution time and give you at least a fighting chance. The pros, can bypass broker risk checks, and not have a middle man handling the data feed. They have direct market access and can plug into the exchange matching engine directly. So how do I the retail guy do this? Just come up with $500,000 and then you can join the club. Unless you are at this level, you will only be as fast as proximity hosting + the highest end of retail platforms allows.

In terms of platforms that could actually achieve a high number of trades and be profitable... I think there are a few retail trading systems out there that might have a fighting chance, but others that certainly would not be fast enough. Anything hosted on the net that requires your trades to sit on their platforms servers before being routed to the exchange would make co-locating useless. There are a couple big retail systems that fall into this category.

Ideally you would want your own system coded in c++, but you aren't likely to go down this road. So here are the retail systems that might give you a fighting chance.

1. Ninjatrader: This is what I use, and while I know it is not the fastest retail option, the fact that it is c# based, allows me to code anything I can think of. Also the unmanaged order class gives it significantly more value to me in pursuing HFT vs. other platforms that have built in rules where you can't do certain things because most retail traders would blow up their account with 10 + open orders on both the bid and ask side... NT let's you do this with the unmanaged order type.

2. Rhythmic API: I believe this is one of the fastest if not the fastest retail level platform. You can co-locate at the CME and in theory get about as fast as any retail player out there. They will cost more per trade though. I think the order routing is 50 cents, vs. CQG / Continuum which is 20 cents.

3. TT Trader / X Trader: I don't know much about this platform but I have heard that it is fairly fast, and several others on this forum that are legit traders use it. They may be able to add more color here.


But here is the big key: Can you make any money in this space in the first place? It is very easy to throw a bunch of limit orders into a crappy simple HLOC bar based back tester and crush it, but most back testers / SIMs will over state your results by a significant magnitude. Playing in this area, especially considering it is a speed game and you will be the slowest guy out there will be difficult.

Here is my advice: Before you spend any money of your own, get a realistic backtesting system from a solid platform, and see if you can beat it with very conservative assumptions. Again, you don't need to make $3 - $5 dollars per trade.
Your break even with all the discounts I mentioned show that anything north of $2 dollars is profit. Start with this goal. Then if you look good, scrutinize the hell out of your results. Look at every trade and try to understand every assumption the backtesting / SIM made. Were these realistic, did these take into account the bid / ask volumes, and your place in the queue, etc. Test and validate, until you feel like you know the assumptions made, and they are a fair representation of the market. If you have any doubt. Get the level 2 data your self and build your own testing platform. I did this, and others on here have as well. You need to be very careful as you test your strategies because I can't tell you how many people crushed a backtester with a Line Break bar, or Renko bar, only to lose real money in the market.

Hope all this helps you out!

Best of luck!

Ian

Thanks Ian:

I am currently using NT8. What version of Ninjatrader are you using? I have briefly looked at X trader, and I have heard of Rhythmic API, but I don't know much about them. I have talked with a colleague about the seat option as well, and your estimate of it cutting the transaction cost appx in half seems correct.

Thank you for your candid comments about backtesting. I am familiar with the ambrosial high one can glean from those all to often fallacious results, that for the briefest moment, make you feel like a genius, but in the end leave you with just a simple sentiment; something like "isn't it pretty to think so" ...I am moving beyond backtesting at this point though and I'm testing my strategy on the live market everyday that I can with my at home NT8 setup for now. I'm not in a great rush, and so far it has worked out astonishingly well. A bit of a surprise, but I will run with it for several months so I can gather as much data as I can, crash and burn, rinse, repeat, and hopefully run through a diversity of market conditions if they present themselves. This forum has been extremely helpful, inspiring at times, and best of all, sobering; it brings me back to Earth. I am trying to get ahead of everything and understand potential costs and pitfalls and what I might be headed for or into, so, again, thank you for your thorough evaluation of my question, and answering so promptly.

Cheers!
Bullywig

Reply With Quote