Forums  > Trading  > XTX  
     
Page 1 of 1
Display using:  

chiral3
Founding Member

Total Posts: 5087
Joined: Mar 2004
 
Posted: 2016-05-25 14:30
I trade ccy to delta hedge some of my foreign exposure but, apart from this, I am oblivious. Can someone that is closer to FX tell me what the deal is with XTX? I understand that they were spun off from GSA and have ties to DB. Obviously the name is a tip to quant trading.

Nonius is Satoshi Nakamoto. 物の哀れ

HitmanH


Total Posts: 483
Joined: Apr 2005
 
Posted: 2016-05-25 17:31
Spin off from GSA - as a home for their more MM / LP activities (which often, given capacity, infrastructure costs, Sharpe etc - fits more into a 'prop' structure than a hedge fund where you'll struggle to take home more than 20% of perf).

Not just FX - but that is bread & butter - and where Gerko made is mark initially.

Luciender


Total Posts: 75
Joined: Aug 2008
 
Posted: 2016-05-25 22:49
Apparently 4th largest spot fx dealer by marketshare, if the Euromoney survey is to be believed..

chiral3
Founding Member

Total Posts: 5087
Joined: Mar 2004
 
Posted: 2016-05-25 22:51
205 views, 2 replies, yet several private IMs, an email, and a phone call. I guess this is stuff we're all interested in but don't want to talk about in the open. :-)

Nonius is Satoshi Nakamoto. 物の哀れ

Nonius
Founding Member
Nonius Unbound
Total Posts: 12785
Joined: Mar 2004
 
Posted: 2016-05-25 23:16
I'm with ya Chiral. Btw, can someone tell me this. Who is Bill Gates fucking these days? Lemme know

Chiral is Tyler Durden

chiral3
Founding Member

Total Posts: 5087
Joined: Mar 2004
 
Posted: 2016-05-25 23:49
Gates is doing philanthropy to make up for all the fucking.

Nonius is Satoshi Nakamoto. 物の哀れ

Nonius
Founding Member
Nonius Unbound
Total Posts: 12785
Joined: Mar 2004
 
Posted: 2016-05-26 07:40
Big Smile

Chiral is Tyler Durden

il_vitorio


Total Posts: 105
Joined: Aug 2014
 
Posted: 2016-05-26 13:52
A little article

and this is what they are "doing". From what I heard he is just investing money in the foundation.

One of my most productive days was throwing away 1000 lines of code.

chiral3
Founding Member

Total Posts: 5087
Joined: Mar 2004
 
Posted: 2016-05-26 14:09
C'mon il vitorio, I have access to public data too. I have a BBG, can code a web scraper. Tell me something I can only find in meatspace. :-)

Nonius is Satoshi Nakamoto. 物の哀れ

gammaphreak01


Total Posts: 20
Joined: May 2007
 
Posted: 2016-05-26 22:19
Not seeing much by way of response. It would be interesting to understand their capital structure and credit model, ie do their shareholders have the capacity to underwrite trading losses and breakdown in correlations or are their counterparts ultimately taking those risks? Not clear from what I have seen on their site

Nonius
Founding Member
Nonius Unbound
Total Posts: 12785
Joined: Mar 2004
 
Posted: 2016-05-26 22:44
I don't know about XTX but in the HFT credit files I've seen, the "shareholders" do not underwrite any losses whatsoever and the PBs typically assesses the credit risks as relatively low compared with a lot of other counterpart categories. Part of that is due to the HFT nature of it, the lower overnight exposure etc. There are some op risks in the (sponsored) access stuff. Then again been a while since I looked at this stuff maybe things have changed.

Chiral is Tyler Durden

katastrofa


Total Posts: 459
Joined: Jul 2008
 
Posted: 2016-05-27 00:25
I do cash equity HFT. We close almost all our positions at the end of day.

Adeka


Total Posts: 6
Joined: Oct 2012
 
Posted: 2016-06-22 09:53
XTX also states "Our average risk holding time in G10 is 11 mins and in EM is 23 mins" on their products page. Also it was claimed previously at GSA by Alex that instantaneous hedging is not market-making Does it mean that they really do not hedge much of their exposure immediately or not at all?

katastrofa


Total Posts: 459
Joined: Jul 2008
 
Posted: 2016-07-04 23:55
They probably want to capture enough flow on multiple venues to ensure they never acquire a too big position, as well as becoming less aggressive on one side if they become too long / too short.

"Instantaneous" hedging would mean crossing the spread to close the acquired position, which defeats the point of being a market-maker (spread capture).

Jurassic


Total Posts: 244
Joined: Mar 2018
 
Posted: 2018-05-12 11:47
..

Jurassic


Total Posts: 244
Joined: Mar 2018
 
Posted: 2019-07-20 13:11
""Instantaneous" hedging would mean crossing the spread to close the acquired position, which defeats the point of being a market-maker (spread capture)."

@Katastrofa Why would a market maker need to cross the spread??? ConfusedConfused

Adeka


Total Posts: 6
Joined: Oct 2012
 
Posted: 2019-07-20 14:25
They have recently launched data science challenge https://challenge.xtxmarkets.com/

Maggette


Total Posts: 1139
Joined: Jun 2007
 
Posted: 2019-07-20 16:03
How the fuck did they come up with the idea that I am willing to load up my resumee and do a brain teaser test to take part in a competition I know nothing about other than:
" where contestants are tasked to develop a predictive model based on training data."

How do I know if I am even interested in the problem?

Ich kam hierher und sah dich und deine Leute lächeln, und sagte mir: Maggette, scheiss auf den small talk, lass lieber deine Fäuste sprechen...

gmetric_Flow


Total Posts: 15
Joined: Oct 2016
 
Posted: 2019-07-20 22:51
It seems that it is more about people who want to work for them than people interested in problem solving/predictive modeling.

See here

HitmanH


Total Posts: 483
Joined: Apr 2005
 
Posted: 2019-07-22 09:17
BTW - it was this company (https://www.correlation-one.com/) - that came up with the test for XTX (they've also done the same for some other big quant hedge funds)

Really cool playform

Arti


Total Posts: 4
Joined: Jun 2017
 
Posted: 2019-07-26 14:53
You just randomly answer the test questions and proceed to the challenge... but the way correlation one is handling the challenge sucks. Predictions are processed through their file which receives all outputs from your strategy, some ML models produce logs which you can't disable, their file is not able to handle logs since it's a str and they check for only int/float, thus it breaks. I addressed that to them and they failed to correct it after the first attempt, now they told that this is not their priority atm. They lack libraries, do not disclose the size of the test set etc. It is very low quality compared to kaggle and other platforms.
Previous Thread :: Next Thread 
Page 1 of 1