I've noticed that Citigroup tends to print a lot of dark liquidity even before it prints edgeA, and I was wondering why this is so. I was under the impression that removing liquidity from darkbooks usually costs money, whereas removing liquidity from edgeA you get a small rebate. Why is it that so much volume goes off on the darkbooks before EdgeA starts printing? Is there a darkbook that gives a better rebate for removing liquidity then EdgeA does, and if so, which one?
There''s lots of inefficiency out there as far as order execution. People trade on regular ECNs such as Arca, Bats, when they can be getting executions on NSDQBX - which pays a huge rebate $1.40 per 1k vs EDGA's $.15. People buy the offer on NSDQBX when there is an ARCA or NSDQ or EDGX midpoint. I think it's just deals institutions have setup with darkpools... but I don't really understand why people throw money away like that.
I definitely plead guilty to this, because it's been easiest to simply settle for the most certainty of execution. In order to do a better job of it, I need to get a better handle on the trade-off between getting the rebate and taking a bigger chance of missing a fill (or getting a poorer fill), and I would appreciate some experienced-trader feedback on this. So let me ask in terms of some basic examples: (1) I have a get-me-out hotkey programmed to load-and-go a limit order that covers my currently open position. It defaults to ARCA. Does it make any sense at all to use EDGA or NSDOBX for the rebate when getting a fill is such a high priority? (2) Say I'm scalping inside the spread with bids & offers using EDGX, and I want to augment that with a protective stop-limit order just outside the spread. If I use EDGA or NSDOBX to get a rebate when the stop-limit order is triggered, am I giving up too much of my safety margin (with respect to getting my protective fill at the best price -- or getting the protective fill at all)? (3) I want to click the mouse and submit a limit-buy order as soon as I see the current price tick above a specific resistance level, because I think it will pop & run upward a good way. If the odds of getting filled are materially lower by routing to take the rebate, then the chance of missing the move (and its associated opportunity cost) would seem to me to far outweigh the rebate $$ received. Am I right about this? I would welcome comments from anyone who has a good, experienced feel for the trade-offs involved in the above situations, because I certainly don't -- and I don't want to waste a lot of money finding out!!
Do you see the dark executions happening at the sub-penny level (assuming the prints allow for the extra decimal places, etc)? If that were the case, my guess is removers are probably hoping that the price improvement they get via dark execution (minus fee) is better than what they'd net by fully crossing the spread on EDGEA and getting the rebate. If you're not seeing sub-penny prints, then, yeah, I'm not sure what the incentive would be for paying a fee to fully cross the spread when you can get a rebate for the same execution price. Reading NYOBScalper's post again, it sounds like this might be part of what he's saying...
You need to watch the LEVEL II and experiment. Clearly you're aware you're throwing some money away, but you don't seem inclined to want to take the time to fix your situation. If you're only throwing away small money and it's not worth your time to fix it, continue on. Given that it's my stock you're throwing money into, I don't see any reason to spend 30 minutes typing out every situation and how you should be routing your order, especially on a public forum, and especially if you won't do the work yourself. Setup hotkeys, stop pointing and clicking. All I'll say is a move in C is a penny... a rebate on taking can be 1/7th of a penny. However, the difference between paying to take and receiving a rebate is $4/1k or .4 penny. Given that a move is 1 penny, execution is clearly important...
I was not seeking or expecting an encyclopedic response, NYOB. I understand the math. And I've done some experimenting. And it seems like, in contrast to rebate trading something static like C, to the degree there is price movement, the trade-off (in math/probability terms) between getting the rebate and getting executed quickly tilts against going for the rebate. But that's not something I can definitively determine on the basis of a few (or even a few dozen) test trades -- which is why I was hoping for feedback from someone who has done many thousands...You mentioned in a different post, for example, that NSDQBX rebates $1.40 to remove. OTOH, ARCA charges $3.00 to remove. How does your fill rate on the one compare to the other? Is the difference enough to justify using NSDQBX only for some very specific situations?