973800+ entries in 0.741s
mjr_: two questions, are signup fees accounted for in mpoe, and can you
transfer your positions from a broker back
to mpex
gribble: The bot responds when you start a line with
the ! character. A good starting point for exploring
the bot is
the !facts command. You can also visit
the bot's website for a list of help
topics and documentation:
http://gribble.sourceforge.net/ mircea_popescu: i like how
that
thing is getting upvoted into a frenzy
mjr_: so much respect
to
the education system
there in regards
to math and science
mjr_: what also was interesting is
that our system is built by romanians
mjr_: thanks
to something known as "toxic order flow"
mjr_: even
though i
think scalping is almost dead
mjr_: prop
trading can't really work without access
to liquidity pools
mjr_: that is
the specific area where my company shines...
mjr_: ie. your company offers you access
to
these routes
mjr_: hmmm....you have
to negotiate
that with
the company, and
that is usually on a company level of interaction
mircea_popescu: but yeah, i lay no claim
to genius on
the "not instacancellable" part. ffs it's beyond obvious.
mjr_: or somewhere around
there
mjr_: with
the most expensive being $2.50
mjr_: many
times it's around $1 per 1,000
mjr_: in exchange for being able
to bypass
the bots and in fact
the whole order book
mjr_: on one
that we use, it's 30 seconds
mjr_: dark pools can be very specific about how long you have
to wait b4 canceling
mjr_: your implementation of
the solution is already in place
mjr_: admittedly only 11ms into
the future but in HFT
that can be enough
mjr_: he literally knew
the future, because he caused it
mjr_: so he would have a bot place orders at a very specific
time, and
then in chicago have orders in place
to
take advantage of
the move he just caused
mjr_: ie. it would
take 11ms for any information from NY
to reach Chicago
mjr_: my favorite usage of HFT was
the guy who used
the speed of light
to his advantage
mjr_: and most exchanges won't bust
that
trade for you
mjr_: if you get
the ack back...its real
mjr_: total bs, and i don't know how you would do
that
mircea_popescu: mjr_
the problem with hfts isn't
that
they're fast. it's
that
they retroactively change
time.
mjr_: yes changing attitudes is
tough
mjr_: jurov: and if he was against HFT why would he provide API? it's practically
the only use for it, besides building a pretty front end (although maybe you mean a slow bot?)
mircea_popescu: no argument
that
this is kinda hard, but not so sure it's for a good reason
mjr_: i was going
to ask about speed of executions...
mircea_popescu: <jurov> but
that's
tailored
to HFT. and MP is quite anti-HFT. << im not anti hft. im anti stupidity, which is mostly phantom orders.
mjr_: sometimes
the Right Way is also
the Hard Way,
though, and even
though it's well worth
the value, adoption rate is important, so making
the Right Way easier would be nice
mjr_: and i
think you are doing
things
the Right Way
mjr_: i agree completely
that encryption is vital
mjr_: yes, i just mean
that
typically pgp is not
the easiest
thing for most people
mircea_popescu: see
the next article on
the blog about something along
those lines.
mjr_: LOL yes
they certainly are
mjr_: do you
think you could build encryption into
the nic card and switch/router?
mircea_popescu:
https is securing
the conduit, ignoring
the message. so is dedicated circuit. so is everything done currently pretty much
mjr_: well when put
that way, conduit agnostic is very much prefferd
mircea_popescu: everyone is currently doing
the former. i am doing
the later
mircea_popescu: either secure
the conduit, message agnostic, or secure
the message, conduit agnostic.
mjr_: but not with
the same ease as phishing
typically
mjr_: for example, if you want arca market data...you have
to get it from somewhere and
typically you do
that
through a circuit
mjr_: kinda...i
think it is mostly for reliabilty and just plain access
mjr_: just curious, because most big markets and banks do
that...so i don't know how phishing can interfere in
those cases, obviously customer on online banking website is vulnerable
jurov: but
that's
tailored
to HFT. and MP is quite anti-HFT.
mjr_: host in a data center where you can cross connect
to other people
mjr_: not have
to lay fiber
jurov: i don't
think mpex will be able
to afford laying fiber even
to biggest customers in near future
mjr_: but
typically it is a /29 subnet mask...only 2 addresses anyway
mjr_: everything goes
through a physical point
to point connection
mjr_: we get our market data
through instinet and sfti...but it is not really over
the internet
mjr_: i mean,
the endpoints are fixed, and don't most banks rely on
them, in colocations and carrier hotels?
mjr_: also, how would dedicated circuits play into what you were
talking about re: phishing
mjr_: that makes sense
then
that i didn't understand it...i have
to deal with databases a lot and
they can be a pain
mjr_: i'm not sure i understood how you were fixing
that issue
jurov: *snatches
the article away*
mjr_: but i can change
that
jurov: ironically, most actively maintained/arbitraged passthrough is
the litecoin one, run by deprived
jurov: perhaps. but someone might get
the point.
jurov: Our many and
talented graphical artists are hard at work refining
the abyssal beauty
that is MPEx.co, and I dare not interrupt
them.
evoorhees: not sure I'd want
to advocate it as a mixer
mircea_popescu: you've got by far
the best mixer out
there, might as well use it as such
mircea_popescu: possibly make it so
that inputs sent
to
that addy are NEVER 0conf'd and as a result
they never receive back anything based on
the same inputs.
mircea_popescu: you send X btc, you receive 9x% of x back 100% of
time
evoorhees: mircea...
that's kinda interesting
mircea_popescu: o hey, apparently ineedausername resurfaced, and is paying his gippt
thing in full.
mjr_: so
they may not have ramped up production as much as previously
thought