Does anyone know exactly what IB shows dark pools when routing an order? For example, say NBBO is 10.01 - 10.05 and I submit a 10.09 bid...does the dark pool get shown a 10.04 bid? Or does IB not show a bid at all and simply request (or check) the best offer from the dark pool? It's been a while since I've looked at it, but I think I remember reading in Barry Johnson's book that both are used/possible. Trader422- I believe you said in another thread that you were going to play around with RPIs and hidden orders...any interesting findings you'd be willing to share? I can offer this from my own studies on my models that trade less liquid issues: No matter what I try, nothing beats sending a marketable order (quickly) and having it get routed through dark pools. That said, when I direct route and take the dark pools out of it, the exact opposite is true and passive orders do better (this is apples to apples, ignoring the different comm rates). I realize this isn't directly applicable to your liquidity provision approach but mentioning it for whatever it's worth.
Forgot to add, I've pretty much given up on adding liquidity as a source of alpha. A number of things have looked solid in testing just to be eaten alive when thrown into the live arena. Also, I don't necessarily consider a resting limit order to always be "adding liquidity" as DeeDeeTwo seems to. For the most part, a fill on a resting limit has come to mean that the NBBO has crossed where my limit was sitting rather me getting a fill on the 'good side' of fair value from an uninformed order. The reality is that my limit is the uninformed order (on that very short time scale). As you've repeated, the bone is picked clean before the flow gets to me. This isn't particularly troubling for me because I started with models that almost exclusively remove liquidity, but I was having fun putting in some work on market-making-like models only to find out they had no chance. Really the motivation for my question about RPI and hidden orders was to find a way to exit trades that are not time sensitive (have a few minutes to exit) as close to neutral (at mid) as possible.
Thanks for stating the almost-but-not-quite obvious ... I would have said it three times already but I am weary of getting DeeDeeTwo riled up and then feeling obligated to respond with my lengthy, boring, turgid prose.
No RPIs, for two reasons: (1) I can't do anything interesting until IB provides API support; (2) I monitor all the subpennies that impact me on a daily basis, partly to see if the RPI thing ramps up, and I have never seen a single NYSE subpenny yet ... plus, more than 98% of all the subpennies I monitor are still Finra prints, with maybe 1-3 per day BATS or EDGE. But I've been kind of infatuated with hidden orders since IB allowed them last August and have a several things to say (keeping in mind that I don't have expert knowledge and would love to hear other elitetraders on this thread correcting me or adding different points of view and new facts): 1. The first thing I did was to blindly switch over to hiddens for a big part of all my trading, just to see what would happen. My conclusions were that it is a different way to trade, with different microstructure stuff going on with the predators and somewhat different patterns of fills and no-fills, but the bottom line was: It definitely did NOT increase my profits or enhance my ability to manage risk, and it may have slightly decreased my profits. 2. Later, I learned that the predators know immediately, if they care at all, where the hidden NBBOs are ... in fact they invented a special order type (NYSE calls it an ALO order) which basically has the purpose of allowing predators to sniff out the hidden NBBO quickly and quietly. Of course, they can also do old-fashioned pinging with 1-lots to find the hidden NBBO. 3. I also read some papers on just how much hidden liquidity is out there in today's markets - it's a lot. So in general if you are trying to be the NBBO (hidden or not) in a less-liquid stock with a wide spread, you can assume you are an idiot because there are hidden levels inside the visible NBBO and the predators will generally have the whole thing sniffed out so THEY are the hidden NBBO all the time, if they want to be, which they usually do for these stocks. 4. There is also a general idea out there that the predators algorithmically guess what the hidden size most probably is, and this is a fairly big number like, say, 1000 shares. This is relevant for a trader like me who often works with 100-300 share orders: If I'm trying to be hidden at the top of the book with a 200-share order, then a few milliseconds after my order becomes the NB it will be sniffed out by a predator, AND the predator will assume it's something like a 1000+ share order - giving him a lot of incentive to get in front of me with his own hidden order. The conclusion is that it's probably better for a trader like me to be visible with my little 200 shares as it becomes less likely the predator will want to get in front of me, at least with size ... if he thinks I might be an iceberg then he can ping me or use another special order type and see I'm not. 5. This is the current thing that fascinates me and which I am actively working on: For my NYSE hiddens back in the order book (that is, they are not NBBO and never were), it appears that someone ALWAYS knows, at least whenever they want to, exactly where the orders are all the time ... predators will follow my hidden orders around instantaneously as I modify, cancel, resubmit, move up and down, or submit brand new hiddens at different levels in the book. Obviously this doesn't happen with every NYSE hidden order, but on bad days I observe it dozens or up to a hundred or so times, depending on how closely I'm looking for the behavior. To my knowledge, this has NEVER happened with any orders in any other order book - just NYSE. I was going to start a thread on this to see if any elitetraders can enlighten me as to what is happening, but I've been waiting until I can open a new non-IB account and experiment there to eliminate the possibility that IB/Timber Hill is coattailing me - a possibility that I consider highly unlikely, but you never know as it would be very easy for IB to notice that I've been an insanely consistent, low-volatility, profitable account for going on 13 years. Would love to hear from other elitetraders on this subject ... but as you can see from my comments, it's just for fun because hidden orders are definitely not anything great, except for the predators.
Thanks for taking the time with that post - lots of good insight there. If you remember, will you please report back what you find regarding the coattailing. I agree that it's unlikely; not only for obvious reasons but also since it's only occurring on NYSE...however, that would open an incredible can of worms if true. Speaking of things that are fascinating...if you've never submitted a REL order in an illiquid stock, it is something to see. I'm amazed at how quickly the predators can identify the order type. I don't get upset over it...it's perversely comical and I'm genuinely impressed. It's like throwing a hunk of meat into the piranha tank; it sits there for a few seconds and nothing happens but as soon as the first fish finds it, things start to go crazy. Same thing with these orders...it takes a few seconds for the predators to poke at it and then there off ripping up/down the stock before filling it much higher/lower. I'm guessing the movement stops once they reach a legitimate resting limit order on the other side of the book - then they turn around and hit it. I've had an extreme case that ran up ~$.30 in an $8 stock, filled me, then went right back to where the bid/ask was before my order...no other trades taking place. I'm fairly confident I could make money from this behavior but 1) would have to research the legality and 2) they'd quickly evolve and outsmart the strategy anyway.
Someone who can perceive the darkness of a pool must know the intricacies of how the various liquidity providers are structured. Please throw us a bone and explain.
I think (2) is by far the biggest practical problem now, although fragmentation itself and the multiplicity/confusing nature/buggyness of order types on the exchanges is also causing problems. But internalization/PFOF is the worst -- there's just too much money in it for retail brokers probabilistically fleecing their own customers (while in some cases ironically calling it "price improvement" -- "we'll toss you 1/100 of a cent per share while occasionally bilking you out of $10,000 for a bad execution") -- and it seems to be getting worse: http://www.tradersmagazine.com/news/off-exchange-volume-spikes-to-record-levels-110792-1.html Also note that the "minimum tick size" increase will play into the hands of internalizers -- it will make internalizing just that much more attractive, increasing adverse selection for those willing to put out lit bids and offers, thereby undermining the (stated) purpose of the proposal in the first place: http://www.tradersmagazine.com/news/sec-discuss-minimum-tick-size-friday-110796-1.html
Yesterday i would have thought that HFTs saw my order before the exchange, but here's what happened. Stock had bid 17.16 and ask 17.22. I sell limit at 17.21 to post the best ask and get filled immediately, IB charges me as a market order. I'm happy i got filled at a good price but not for the lack of rebate. I check the feed, and someone was fluttering the bid previously. Fluttering means it goes from 17.16 to 17 to 18 to 19 to 20 to 21 and then back to 16 in a matter of seconds. Later they switched to millisecond intervals so a lot of the feeds don't even see them, that's when i got filled. It's not cheating, anyone can do it. This is on a lower liquidty stock also. You guys need to understand that not all HFTs are dedicated HFT monsters moving at light-speed. On lower liquidity some funds (organizations) have traders and their single job is to get the best execution for a large order and this can last for days in some cases. They use unsophisticated HFT-like algos to do things like what happened to me - fluttering, stuffing, etc. They are often not fully automated, and some are even not colocated, at least when dealing with lower liquidity. In essence, with a budget of around $5k you could be that "HFT" on low liquidty, and of course some software. As far as HFT MMs are concerned, i'm not seeing them (again, on low volume stocks). I consider myself an MM because i do arbitrage and 90% of my fills are limits on lower liquidity. I also thought about implementing algos and it wouldn't be difficult. I'm already using non-HFT algos, so that's different. So it's not a grand conspiracy, there's a few traders with less than $1MM doing that as well.
Please tell me if I'm wrong I'd like to know, but my understanding is: You can't ask a dark pool what their best bid and offer are. That's what makes them dark. The only information you can get is from executions (although it seems to me not all of them report executions immediately, not sure about that) and that your own orders didn't execute. (Obviously the people running the dark pool know all the orders within it, and who knows if they are playing fair with that information or not.)