Dark Pool / Spead B/Ds ? (for stat arb)

E

Esq Esq Esq

Hi everyone,

trying to increase volume/decrease slippage on some pair trading. What are some dark pools with very good liquidity and aggressive remove rates (+ price improvements).

Also - anyone know of any good spread b/ds - that will give you real-time pricing on a pair of securities (kinda like when you do options spreads).

ideas?

thx/
 
That depends on what you're trading... some darkbooks are very active on SP500 symbols, some are active only on big names, while others do most of their volume on dry, small symbols...
 
Quote from eusdaiki:

That depends on what you're trading... some darkbooks are very active on SP500 symbols, some are active only on big names, while others do most of their volume on dry, small symbols...
More...

\Mind sharing your findings?
 
Quote from Esq Esq Esq:

\Mind sharing your findings?
More...
Take for example Getco (Wedbush Morgan) they can get you a fill in 1 out of 4 orders in DOW 30 names, but you won't ever find liq in that ECN if you're trading a thin stock.
 
Getco is interesting....especially with its competitive rates. unfortunately I can't reproduce your 1/4 rate even in dow stocks. Trying to use it strictly, the results are mixed -- I'm also doing some kind of pair trading..and in some cases Getco has been fabulous, however in most cases it has been useless: it either wouldn't get you a fill at all..or it would fill one on the spot and chase the other ad infinitum resulting in a loss.

NIX is of *some* use in *some* cases...especially if ur doing odd lots.

Of course you should explore ASUROUX,SROUX,SRDOT and the rest of EDGA family.

NSDQSTGY/NSDQSCAN can give you *some* advantages in *some* cases :D

MLDOT is really fast and timely -- but ofcourse only on NYSE and again..its like with getco in the sense that if liquidity is there - ud be thinking uve hit jackpot! untill the day it isn't and ur slipped so much it erases all your profits.

However after trying just about any ECN in existence and the combinations thereof..I have found BATS to be most efficient commission/slippage wise.

Using a mixture...results in too many tradeoff's..i.e. by the time u reach the FILL you'd have got partials from other ECN's and wasted enough time to deviate from the price you wanted and handling all these events in real-time isn't fun and sometimes isn't trivial at all (i.e. in the code)

If you're trading large sizes and can afford the research/time: I think the way to go might be some probabilistic models that rely on past liquidity. per stock per ecn per time. or..another way to go is to watch T&S in real time and make your best judgment (again probabilistic) on where to route thru.

Overall..your best guess is experimenting with all that is available to you because what is best for you realy really depends on your methodology and the slippage/delay your model can allow itself. you are unlikely to hit any magic solution -- that will last. its an ever cahnging ever evolving process.

Hope that helps :D
 
Quote from DirkDiggler:

Getco is interesting....especially with its competitive rates. unfortunately I can't reproduce your 1/4 rate even in dow stocks. Trying to use it strictly, the results are mixed -- I'm also doing some kind of pair trading..and in some cases Getco has been fabulous, however in most cases it has been useless: it either wouldn't get you a fill at all..or it would fill one on the spot and chase the other ad infinitum resulting in a loss.

NIX is of *some* use in *some* cases...especially if ur doing odd lots.

Of course you should explore ASUROUX,SROUX,SRDOT and the rest of EDGA family.

NSDQSTGY/NSDQSCAN can give you *some* advantages in *some* cases :D

MLDOT is really fast and timely -- but ofcourse only on NYSE and again..its like with getco in the sense that if liquidity is there - ud be thinking uve hit jackpot! untill the day it isn't and ur slipped so much it erases all your profits.

However after trying just about any ECN in existence and the combinations thereof..I have found BATS to be most efficient commission/slippage wise.

Using a mixture...results in too many tradeoff's..i.e. by the time u reach the FILL you'd have got partials from other ECN's and wasted enough time to deviate from the price you wanted and handling all these events in real-time isn't fun and sometimes isn't trivial at all (i.e. in the code)

If you're trading large sizes and can afford the research/time: I think the way to go might be some probabilistic models that rely on past liquidity. per stock per ecn per time. or..another way to go is to watch T&S in real time and make your best judgment (again probabilistic) on where to route thru.

Overall..your best guess is experimenting with all that is available to you because what is best for you realy really depends on your methodology and the slippage/delay your model can allow itself. you are unlikely to hit any magic solution -- that will last. its an ever cahnging ever evolving process.

Hope that helps :D
More...

Thanks for detailed post.

I found ok liquidity in Bats and EDGX families (actually EDGX being better on thicker names).
However, since i mainly remove and pay spread, i need to really control the costs.

NYFIX gives me a price improvement once in a blue moon - but their servers are slow, and its not even worth it sometimes.

i trade 100k + lots on thick names so, liquidity is a big issue for us, and slippage..and spread...and liquidity
 
ASUROUX is your friend. you will get price improvement (quite often) - internal match is 0.18c/(1 lot of 100share) on tape a/b quite often you will not be charged - I have found this to be about 40% of the time in high liquid issues. If you're routed out, its 0.29c (per 100 lot). in liquid issues its relatively fast (faster than arca/bats/edgx/nsx maybe can be outperformed by MLDOT/SRDOT/GETCO/NIX -- again, depends on conditions)

I have not yet figured as for when it would give you zero remove and when it would charge 0.18c -- the liquidity tag it returns is the same. [anyone has an idea????]

I do etf's strictly, so for me most of the time its either route out or internal match in which case its 0.0 - since about november last year, it has been 0.0 on about 40% of the orders i route thru it, so the avg rate for me comes down to about 0.17c or so.

as all else it has its problems. in *some* issues at *some* times..it simply has NO liquidity! I have been unable so far to determine what conditions effect it. if it has no liquidity...it will route out..but by the time it does that...my train would have left already.
another problem with it, is that since november it has crashed several times -- some time it wouldnt cancel orders...(and if you have few hundreds stuck its a bit of a problem because you then find yourself suddenly filled at surprising prices :>), sometimes it simply would not except orders :D this has happend several times..id say, less than 7-8.

However, even so, many people prefer it and use it as their main tool and for good reasons. In large issues, 100K+ lots are no problem too. Give it a try.

u can check ASURDOX/RDOX etc...which suppose to do the same but they aren't suppose to route you out -- therefore guaranteeing low fees (I ssay supposed, because sometimes -- they do); and they dont give price improve.

I forgot to mention QUID/QUIK...I dont use them as in the issues I trade they dont have liquidity but some people have had great success with them. the D version is dark and would give price improve *most* of the time. check them out.

Another thing that might work for you (if your profit target margins allow it ) is POSITNOW. Very expensive - 2.2$ per 1 lot (of 100 share)..but again for a reason, pretty fast good liquidity and almost never slippage. again - i dont use it, but some people trading less liquid issues with higher bid/ask spreads have had great success with it (that well deserves the fees)

Hope that helps :)

ps: you say u do 100k lots on big issues. how often? what is your timeframe? what is you profit target?
I do intraday with much smaller lots and quite high frequency (target is 3-5cents per trade *after* commissions and slippage);
 
I have not yet figured as for when it would give you zero remove and when it would charge 0.18c -- the liquidity tag it returns is the same. [anyone has an idea????]

****************************
What liq flag do you get?


****************************


I forgot to mention QUID/QUIK...I dont use them as in the issues I trade they dont have liquidity but some people have had great success with them. the D version is dark and would give price improve *most* of the time. check them out.

*****************************
What's QUID/QUIK?

******************************
 
asuroux:
Mar 3, 2009 3:32:04 PM Short PFE W 200 11.92 0.04 0.02 0.015 0.0 ASUROUX

Liq tag is W, 0 commission

then

Mar 3, 2009 3:43:14 PM Sell PFE W 200 11.84 0.04 0.02 0.015 0.18 ASUROUX

Liq tag is W, 0.18c commission

same thing with N tag, sometimes charge sometimes no:

Jan 27, 2009 1:53:44 PM Sell UYG N 200 3.25 0.04 0.02 0.015 0.29 ASUROUX

but
Jan 27, 2009 1:57:08 PM Short UYG N 200 3.22 0.04 0.01 0.015 0.0 ASUROUX

notice .29 for 200shares i.e. 0.145 per 100

now same tag diff price:
Jan 27, 2009 2:28:45 PM Sell UYG N 400 3.21 0.08 0.02 0.03 0.87 ASUROUX

in this order 400 for 0.87 i.e. roughly 0.21c per 100


In other tags it has same oddities, same issues same tag diff price...luckily at least there's a clear pattern here :) X are reroutes and I are removes and both are consistent with pricing. N's are clearly most of the time 0, and B's are add-liquidity and are 0 too (it doesnt give credit)...rest of EDGA family..ASRDOT/ARDOX and friends..have mostly same tags..but are quite consistent about them and their pricing. ASUROUX seems the only black sheep (black box??? :D).

Anyone with more experience on EDGA family to elaborate??

QUID/QUIK: only have old ecn papers on me here..will look up in office if you're really interested. what does google have to say??
 
Quote from DirkDiggler:

asuroux:
Mar 3, 2009 3:32:04 PM Short PFE W 200 11.92 0.04 0.02 0.015 0.0 ASUROUX

Liq tag is W, 0 commission

then

Mar 3, 2009 3:43:14 PM Sell PFE W 200 11.84 0.04 0.02 0.015 0.18 ASUROUX

Liq tag is W, 0.18c commission

same thing with N tag, sometimes charge sometimes no:

Jan 27, 2009 1:53:44 PM Sell UYG N 200 3.25 0.04 0.02 0.015 0.29 ASUROUX

but
Jan 27, 2009 1:57:08 PM Short UYG N 200 3.22 0.04 0.01 0.015 0.0 ASUROUX

notice .29 for 200shares i.e. 0.145 per 100

now same tag diff price:
Jan 27, 2009 2:28:45 PM Sell UYG N 400 3.21 0.08 0.02 0.03 0.87 ASUROUX

in this order 400 for 0.87 i.e. roughly 0.21c per 100


In other tags it has same oddities, same issues same tag diff price...luckily at least there's a clear pattern here :) X are reroutes and I are removes and both are consistent with pricing. N's are clearly most of the time 0, and B's are add-liquidity and are 0 too (it doesnt give credit)...rest of EDGA family..ASRDOT/ARDOX and friends..have mostly same tags..but are quite consistent about them and their pricing. ASUROUX seems the only black sheep (black box??? :D).

Anyone with more experience on EDGA family to elaborate??

QUID/QUIK: only have old ecn papers on me here..will look up in office if you're really interested. what does google have to say??
More...

Strange... according to Direct Edge's website, EDGA flags W & N should always charge 0... so you might want to ask your broker about those charges...


QUID/QUIK... well... google doesn't say much, other than "did you mean QUICK/QUICK?' and a few sites that give paycheck advancements in GBP...
It's not a big deal I'm just curious to know who they are since I've never heard of them before and I usually like to keep track of all the mayor players in the dark space...
 
You must log in or register to reply here.
Back
Top