NexusFi: Find Your Edge


Home Menu

 





The Wall Street Code


Discussion in Traders Hideout

Updated
      Top Posters
    1. looks_one artemiso with 36 posts (30 thanks)
    2. looks_two Big Mike with 19 posts (41 thanks)
    3. looks_3 Outlier with 13 posts (5 thanks)
    4. looks_4 nanex with 11 posts (16 thanks)
      Best Posters
    1. looks_one Big Mike with 2.2 thanks per post
    2. looks_two nanex with 1.5 thanks per post
    3. looks_3 artemiso with 0.8 thanks per post
    4. looks_4 Outlier with 0.4 thanks per post
    1. trending_up 44,758 views
    2. thumb_up 107 thanks given
    3. group 21 followers
    1. forum 96 posts
    2. attach_file 6 attachments




 
Search this Thread

The Wall Street Code

  #11 (permalink)
 
Big Mike's Avatar
 Big Mike 
Manta, Ecuador
Site Administrator
Developer
Swing Trader
 
Experience: Advanced
Platform: Custom solution
Broker: IBKR
Trading: Stocks & Futures
Frequency: Every few days
Duration: Weeks
Posts: 50,398 since Jun 2009
Thanks Given: 33,173
Thanks Received: 101,537


artemiso View Post
I'm disappointed that individuals such as Peterffy, Lauer and Bodek were glorified in this documentary, while the unsung heroes of the market were unfairly bashed.

1. Peterffy is by far the richest person in the entire film, for a good reason: Interactive Brokers trades against its own clients through Timber Hill, and sells them out to dark pools and ECNs (https://gdcdyn.interactivebrokers.com/Universal/servlet/Registration.formSampleView?ad=order_routing_disclosure.html).

2. During the Flash Crash, Lauer was at Allston, a "greybox" shop that did poorly in the past years and therefore had to cut its staff. (Besides, they are a JVM shop, by no means latency-sensitive.) He has absolutely no authority about this matter. The majority of 'mini-crashes' and bottoms are formed by retail (click or electronic) traders and slow traders aggressing against low latency electronic traders, e.g. through stop loss orders. I only say "the majority" for political correctness because honestly I don't pay attention to these things and my actual memory of these incidences was that 100% of them were caused by low-frequency traders. Here's an example where a market lost about 4% of its value in an hour where I had to supply liquidity all the way down to the bottom. The pink bar shows the contribution of low latency electronic traders and the two largest bars clearly show that the slow traders were responsible for most of the market damage and imbalance.



3. I believe Direct Edge had made a public statement that they had email evidence of discussions with Bodek about the hide-or-slide order type well in advance. Besides, it is your own undoing if you choose to compete in an area where specific order types matter and don't bother to keep yourself updated. Bodek had no serious experience before he started Trading Machines (I know to the common public, a short stint at UBS and Hull is a lifetime of financial experience, but practically everyone that you talk to on the phone from the largest 10 FCMs and prime brokers will have a better resume.)

Where is the journalistic standard nowadays?


No one is systematically disadvantaging your limit orders.

The exploits that Haim refer to only exist in the equity markets where Reg NMS is in force. Even so, the equity markets are a fair game because these strategies only degrade the returns for other low latency electronic traders.

<deleted by request>

Eric's recommendations in Flash Crash Analysis - May 6'th 2010 - Part 1 - Empirical Analysis - Answers - Nanex and Nanex - Flash Crash Analysis - Flash Crash Summary are uninformed.

Regarding his first suggestion, it's fundamentally impossible for the public feed to display the timestamp simultaneously as the match takes place with the present exchange architecture. I think he has some vague idea that they should eliminate the round trip on the order channel, which has unintended latency consequences that we waste a lot of time solving. I agree that this is a good idea as it would make for a simpler API to develop against, and therefore reduce latencies and improve the market quality across in general. I feel, however, that Eric's motive for suggesting this is so that data vendors do not have to pay that extra $28,550 per month for exchange licenses; I doubt this will achieve the intended goal, as the license fees for the SIP feeds will undoubtedly increase if unchecked. You can do this and make the SIP feeds free too - but that will become a cost on taxpayer money. You can't have the best of all worlds.

Regarding his second suggestion, sure, I absolutely agree. But whoever's under the impression that this will reduce public costs is gravely mistaken. The growing bandwidth requirements is a function of several factors, a few are good reasons, and a few (such as liquidity rebates, and this), are bad.

Thanks @artemiso.

Have you seen these?

Nanex ~ 28-Sep-2013 ~ Shredding Virtu's Response with Science

Nanex ~ 30-Sep-2013 ~ [AUTOLINK]HFT[/AUTOLINK] Front Running, All The Time

Mike

We're here to help: just ask the community or contact our Help Desk

Quick Links: Change your Username or Register as a Vendor
Searching for trading reviews? Review this list
Lifetime Elite Membership: Sign-up for only $149 USD
Exclusive money saving offers from our Site Sponsors: Browse Offers
Report problems with the site: Using the NexusFi changelog thread
Follow me on Twitter Visit my NexusFi Trade Journal Started this thread Reply With Quote
Thanked by:

Can you help answer these questions
from other members on NexusFi?
Exit Strategy
NinjaTrader
Build trailing stop for micro index(s)
Psychology and Money Management
Futures True Range Report
The Elite Circle
New Micros: Ultra 10-Year & Ultra T-Bond -- Live Now
Treasury Notes and Bonds
ZombieSqueeze
Platforms and Indicators
 
  #12 (permalink)
 artemiso 
New York, NY
 
Experience: Beginner
Platform: Vanguard 401k
Broker: Yahoo Finance
Trading: Mutual funds
Posts: 1,152 since Jul 2012
Thanks Given: 784
Thanks Received: 2,685



There are several pseudosciences and misquotes in Nanex's response that I can talk about another time (I'm busy). I'm surprised and pleased that Virtu did a good job rebutting him before me - that was a good find - thanks.

Reply With Quote
  #13 (permalink)
 Outlier 
Germany
 
Experience: Advanced
Platform: TradeStation
Trading: Futures
Posts: 88 since May 2012
Thanks Given: 53
Thanks Received: 93



artemiso View Post
1. Peterffy is by far the richest person in the entire film, for a good reason: Interactive Brokers trades against its own clients through Timber Hill, and sells them out to dark pools and ECNs (https://gdcdyn.interactivebrokers.com/Universal/servlet/Registration.formSampleView?ad=order_routing_disclosure.html).

I know of two strong indications that this is incorrect and Peterffy's success didn't depend on special agreements with Interactive Brokers.

1. In 1977, Thomas Peterffy bought a seat on the American Stock Exchange (AMEX) and became a member, trading as an individual market maker in equity options. In 1982, Timber Hill Inc. was formed and in 1983, Timber Hill expanded to 12 employees. It wasn't until 1993, over a decade later, that Interactive Brokers Inc. got established as a US broker-dealer.

2. In 2010, Peterffy spoke out against internalization before the World Federation of Exchanges, noting that no major online broker sent more than 5% of its orders to an organized exchange (according to the Rule 606 reports mandated by the U.S. Securities and Exchange Commission) with the sole exception of Interactive Brokers.

The transcript of the speech is worth a read: https://www.interactivebrokers.com/download/worldFederationOfExchanges.pdf


artemiso View Post
If the same quality of delivery is used, it's possible to receive the same packet in Chicago in under 1.6 milliseconds after I receive it in NJ, not 5 milliseconds as Nanex claims.

That would violate the speed of light as Nanex showed in their analysis which also considered third party timestamps for confirmation. I didn't find Nanex's response to be pseudoscience.

Reply With Quote
  #14 (permalink)
 artemiso 
New York, NY
 
Experience: Beginner
Platform: Vanguard 401k
Broker: Yahoo Finance
Trading: Mutual funds
Posts: 1,152 since Jul 2012
Thanks Given: 784
Thanks Received: 2,685


Outlier View Post
I know of two strong indications that this is incorrect and Peterffy's success didn't depend on special agreements with Interactive Brokers.

1. In 1977, Thomas Peterffy bought a seat on the American Stock Exchange (AMEX) and became a member, trading as an individual market maker in equity options. In 1982, Timber Hill Inc. was formed and in 1983, Timber Hill expanded to 12 employees. It wasn't until 1993, over a decade later, that Interactive Brokers Inc. got established as a US broker-dealer.

2. In 2010, Peterffy spoke out against internalization before the World Federation of Exchanges, noting that no major online broker sent more than 5% of its orders to an organized exchange (according to the Rule 606 reports mandated by the U.S. Securities and Exchange Commission) with the sole exception of Interactive Brokers.

Somehow half of my post got lost while I was typing it.

I don't really see your argument here. There's not a lot you can say about a person's wealth today based on two dates several decades ago. I wasn't even alive in 1982: this doesn't imply anything about how I made the majority of my money today. IBKR's K-10 filing (https://www.sec.gov/Archives/edgar/data/1381197/000104746913002439/a2213156z10-k.htm) yields more concrete evidence:

Timber Hill is slow by comparison to say, GETCO or IMC. The majority of their trading advantage comes from customer flow. If this isn't telling enough, consider their trading gains: in 2012, they made $511.5M from prop trading and payment for order flow (I concede this is a slight overestimation as they lump "market data fees" and "account inactivity fees" together with this) in contrast to $412.6M from commissions. In 2011, they made a startling $708.2M from prop trading and payment for order flow as compared to $456.2M in commissions.

IBKR made a whopping 43% and 49% of its net revenue from prop trading in 2012 and 2011 respectively. Over the same period, Goldman Sachs made 17% of its in net revenue from prop trading compared to other activities that improve our society, e.g. client services, lending and investment management. IBKR promotes negative externalities to the market, e.g. gambling, while Goldman doesn't (well, this is up for another debate).

Statistically speaking, IBKR is a worse vampire squid than Goldman will ever be.

Reply With Quote
Thanked by:
  #15 (permalink)
 
Big Mike's Avatar
 Big Mike 
Manta, Ecuador
Site Administrator
Developer
Swing Trader
 
Experience: Advanced
Platform: Custom solution
Broker: IBKR
Trading: Stocks & Futures
Frequency: Every few days
Duration: Weeks
Posts: 50,398 since Jun 2009
Thanks Given: 33,173
Thanks Received: 101,537


artemiso View Post
Notice that I have a standard deviation of 6 microseconds across the 2 states - I'll bet you that Nanex can't even achieve this by pinging their own router.


Are you on Steadfast's network (assuming, since you used a Steadfast IP to test against)?

Impressive deviation. This is from one of my 350 E Cermak servers to another, on Steadfast's network. Keep in mind this is also two public facing web servers, db and app servers, with 400 users online (nexusfi.com (formerly BMT) servers).



Mike

We're here to help: just ask the community or contact our Help Desk

Quick Links: Change your Username or Register as a Vendor
Searching for trading reviews? Review this list
Lifetime Elite Membership: Sign-up for only $149 USD
Exclusive money saving offers from our Site Sponsors: Browse Offers
Report problems with the site: Using the NexusFi changelog thread
Follow me on Twitter Visit my NexusFi Trade Journal Started this thread Reply With Quote
Thanked by:
  #16 (permalink)
 
Zondor's Avatar
 Zondor 
Portland Oregon, United States
 
Experience: Beginner
Platform: Ninjatrader®
Broker: CQG, Kinetick
Trading: Gameplay Klownbine® Trading of Globex
Posts: 1,333 since Jul 2009
Thanks Given: 1,246
Thanks Received: 2,731

Great thread, guys.

"If we don't loosen up some money, this sucker is going down." -GW Bush, 2008
“Lack of proof that something is true does not prove that it is not true - when you want to believe.” -Humpty Dumpty, 2014
“The greatest shortcoming of the human race is our inability to understand the exponential function.”
Prof. Albert Bartlett
Follow me on Twitter Visit my NexusFi Trade Journal Reply With Quote
Thanked by:
  #17 (permalink)
 Outlier 
Germany
 
Experience: Advanced
Platform: TradeStation
Trading: Futures
Posts: 88 since May 2012
Thanks Given: 53
Thanks Received: 93


artemiso View Post
How is this violating the speed of light?

I ran a quick calculation to arrive at 1.6 milliseconds before adjusting for the geodesics. There's a bit of work to make the adjustment but I can do that if you insist: it brings me to a difference in geodesic distance between DC and NJ2 as compared to that between DC and CME's location of about 613.566097km. There are also a few things that are missing from this picture that are responsible for another ~100 microseconds. So I can concede to you that the difference will be under 2.1+0.1=2.2 milliseconds.

Because it would be faster than light. How does not adjusting for geodesics let you arrive at 1.6 milliseconds? That's implausible from the checks I did.

Your other numbers are in agreement with Nanex's and show that there's something wrong when the theoretical minimum is established at 2.2 milliseconds but trades reacting to the news release occurred just 1.637 milliseconds apart in the two locations, according to Virtu's accurate and synchronized timestamps. That's what Nanex argued and you basically confirmed.

Nanex ~ 28-Sep-2013 ~ Shredding Virtu's Response with Science

Reply With Quote
  #18 (permalink)
 artemiso 
New York, NY
 
Experience: Beginner
Platform: Vanguard 401k
Broker: Yahoo Finance
Trading: Mutual funds
Posts: 1,152 since Jul 2012
Thanks Given: 784
Thanks Received: 2,685


Outlier View Post
Because it would be faster than light. How does not adjusting for geodesics let you arrive at 1.6 milliseconds? That's implausible from the checks I did.

Your other numbers are in agreement with Nanex's and show that there's something wrong when the theoretical minimum is established at 2.2 milliseconds but trades reacting to the news release occurred just 1.637 milliseconds apart in the two locations, according to Virtu's accurate and synchronized timestamps. That's what Nanex argued and you basically confirmed.

Nanex ~ 28-Sep-2013 ~ Shredding Virtu's Response with Science

1. You've given me no evidence of what checks you've made.

Step 1: The radius of the earth is about 6400km. New York is about -77.5 degrees and Chicago is about -88 degrees. You know these if you've had to apply for a job at a hedge fund or a prop firm as a quant because you have to prepare for Fermi estimate brainteasers. I know the first one because I was a physicist.

Step 2: The most difficult part of the calculation is manipulating the trigonometric functions so you either try to pin the longitude or the latitude to an integer multiple of 1/4*pi. Since I know geographically that the distance between the 2 cities is mostly east-west versus north-south, and that the 2 cities should be nearly on the same latitude, it's easier to avoid trouble by taking both latitudes to be 1/4*pi. This simplifies most part of the calculation - because

sin(1/4*pi) = cos(1/4*pi) = sqrt(1/2)
-77.5/180*pi = ~-1.4. -88/180*pi = ~-1.5.

Step 2: Transform to Cartesian coordinates, where phi1 = phi2 = sqrt(1/2), psi1 = -1.5, psi2 = -1.4.

(x1,y1,z1) = (r cos(phi1) cos(psi1), r cos(phi1) sin(psi1), r sin(phi1))
(x2, y2, z2) = (r cos(phi2) cos(psi2), r cos(phi2) sin(psi2), r sin(phi2))

Step 4: Take the Euclidean distance (notice all of the approximations are to 2 significant figures, this makes calculation fast but also implies that any higher precision can be discarded):

sqrt( (x1-x2)^2 + (y1-y2)^2 + (z1-z2)^2 ) = 660km

So the approximate direct distance (ignoring surface curvature) from Washington DC to Chicago is about 660km. To obtain an upper bound, assume the packet is transmitted over fiber. Light travels in fiber at approximately 2/3*c. The distance between Chicago and Washington DC is about twice that of the distance between Chicago and New York, so the difference in time to receive the packet would be approximately (330e3/3e8) / (2/3) / (1/2) = (110/1e5)*3 = 1.6 milliseconds.

Step 5: Adjust for geodesics... This is difficult to do by hand. Usually, you would just assume that the arc length would be approximately equal to the chord length given the small angular separation. Then we're done. This is actually a stronger assumption to those that I made in my estimates above (biggest source of error is the sensitivity to psi1 and psi2).

2. There is no reason why the first trade immediately after 2:00:00 PM has to be a trade responding to the news announcement at 2:00:00 PM. Cross-venue trading occurs between these 2 venues all throughout the day. Any participant can write a buy program to lift the asks at two venues at a preset time in the future almost simultaneously (<<1 ms). It is poor academic practice to pick just 2 instruments precisely to find the pairwise difference in time of first trade after 2:00:00 PM. If you want to do this, you don't have to pick any special economic event. You can easily run a script and find hundreds of pairs whose times between their first trades after any announcement at an arbitrary time (e.g. 10 AM EST) are smaller than 1 millisecond and claim it is a market exploit.

3. Read carefully. Nanex makes a shifting argument here. This was their first post.



This was their second post.



The SPY timestamp should have nothing to do with the measurement of the reaction time in the GC futures contract. This measurement now has a completely different frame of reference from the one earlier implied. If you think this is valid, then why don't you trade with me the difference {EUR 21 - US$20} for the difference {US$21 - US$20}?

4. I notice that you didn't respond to my post about Interactive Brokers and that you gave no evidence of your 'checks'. If you're only interested in arguing for the sake of it, there's no point in me posting about anything if it doesn't help you. Whatever floats your boat. No one is stopping you from using Nanex and Interactive Brokers as your service providers.

Reply With Quote
  #19 (permalink)
 
Big Mike's Avatar
 Big Mike 
Manta, Ecuador
Site Administrator
Developer
Swing Trader
 
Experience: Advanced
Platform: Custom solution
Broker: IBKR
Trading: Stocks & Futures
Frequency: Every few days
Duration: Weeks
Posts: 50,398 since Jun 2009
Thanks Given: 33,173
Thanks Received: 101,537


artemiso View Post
I have to apologize - I noticed that I uploaded the wrong screenshot - I was about to answer your question and did a quick traceroute when I realized it hops to New York. My vague short-term memory reminded me that the pings are about 6.X~7.X and the hardware was the same on this server so I thought this was the correct screenshot. I'll put up another screenshot if not that we're fixing something now. My first reaction was that this would be CH1 to New York but my number would be off. I'll check again. Curious now. The correct screenshot that I was pulling out should be pinging to 208.100.4.54.

 
Code
root@phoenix:~# ping -c 5 208.100.4.54
PING 208.100.4.54 (208.100.4.54) 56(84) bytes of data.
64 bytes from 208.100.4.54: icmp_req=1 ttl=63 time=0.273 ms
64 bytes from 208.100.4.54: icmp_req=2 ttl=63 time=0.215 ms
64 bytes from 208.100.4.54: icmp_req=3 ttl=63 time=0.238 ms
64 bytes from 208.100.4.54: icmp_req=4 ttl=63 time=0.218 ms
64 bytes from 208.100.4.54: icmp_req=5 ttl=63 time=0.235 ms

--- 208.100.4.54 ping statistics ---
5 packets transmitted, 5 received, 0% packet loss, time 3999ms
rtt min/avg/max/mdev = 0.215/0.235/0.273/0.028 ms
Mike

We're here to help: just ask the community or contact our Help Desk

Quick Links: Change your Username or Register as a Vendor
Searching for trading reviews? Review this list
Lifetime Elite Membership: Sign-up for only $149 USD
Exclusive money saving offers from our Site Sponsors: Browse Offers
Report problems with the site: Using the NexusFi changelog thread
Follow me on Twitter Visit my NexusFi Trade Journal Started this thread Reply With Quote
  #20 (permalink)
 
Big Mike's Avatar
 Big Mike 
Manta, Ecuador
Site Administrator
Developer
Swing Trader
 
Experience: Advanced
Platform: Custom solution
Broker: IBKR
Trading: Stocks & Futures
Frequency: Every few days
Duration: Weeks
Posts: 50,398 since Jun 2009
Thanks Given: 33,173
Thanks Received: 101,537



artemiso View Post
Looks correct.

1. You're only paying for the stable power and connection and not the latency benefit if you are using NT7 for execution. The delay on NT7 is >160 milliseconds on a 1.6 GHz dual core Sandy Bridge architecture with SSD, absolutely dwarves whatever reduction in delay that Steadfast brings you.

2. If you're ever looking to pay more to reduce your network latency, let me know. You've helped me the other time with the data storage.

I am using my own code to execute, but it is on IB. I am not exploiting latency in my strategies, I do not scalp and neither do my strategies

Mike

We're here to help: just ask the community or contact our Help Desk

Quick Links: Change your Username or Register as a Vendor
Searching for trading reviews? Review this list
Lifetime Elite Membership: Sign-up for only $149 USD
Exclusive money saving offers from our Site Sponsors: Browse Offers
Report problems with the site: Using the NexusFi changelog thread
Follow me on Twitter Visit my NexusFi Trade Journal Started this thread Reply With Quote
Thanked by:




Last Updated on December 19, 2013


© 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