I am wondering whether anyone can share experiences and longer term average probabilities of fills when quoting inside the spread at various fx ECNs. I need to make an assumption of the probability of getting filled as a function of fx ECN venue, how aggressive I bid and offer inside the spread (in terms of how far away from current best bid/offer and how frequent to modify), and as function of current micro market dynamics (momentum vs mean reversion). I am just looking for an approximate figure that averages out over a longer period (days, weeks,...) and ignoring micro market dynamics and how aggressively the quoting is performed. I intend to eventually quote concurrently on both sides at several ECNs I have access to (FxOne, Integral FxInside, FastMatch, LMAX, FxAll, Hotspot). Not all lend themselves to quoting inside the current spread (due to last look provisions and what have you). I can colocate/proximity locate in the same building as most of mentioned ECNs.

Edit: I am not looking for a specific number but rather wonder how others model fill probabilities when testing market making algorithms.



I think it will also depend on the amount of the orders you will entering. In FXInside it will also depend if you are just aggregating or using a HUB, and even if you use the HUB it will depend if you are enable to "make liquidity" otherwise you will be only sending an agressive watch order waiting a market move.

I don't have any number to share with you, but by manually trading I could said hotspot has good fill rations for bids and offers inside the spread.

Don't know neither if Bloomberg Tradebook is available via fix, but I can remember a lot of small amount fills, if your are planing to send small orders, maybe it could help. Its easy to get filled if there is a "fishing algo" in the other side taking small amounts to see if there is real liquidity.

The other way around with FastMatch, because there you will gain priority based in the order size.

And the last but not less important, obviously it will depend in the currency pair, but for sure you are taking that in account already.

  • $\begingroup$ I did not expect a specific number, I guess I just wonder what assumptions some of the underlying models make or how people model fill probabilities when testing new algos. I will edit my answer accordingly, thanks. $\endgroup$ – Matt Nov 28 '14 at 6:16

Wow, this is the 1 million question. No answer available to that. From what I saw is that that question varies from ecn to ecn. You will be very depending on each ecn and how they connect you to liquidity takers.

  • $\begingroup$ indeed you are right it heavily depends on the structure each ECN operates on and its liquidity providers and takers. I guess I was more trying to share/gather thoughts on the models, used, to assess fill frequency, quality, and subsequent dynamics specifically among fx ECNs. I found very early on that quoting and fill dynamics in fx space are entirely different than in equity space. I did not find models of equity micro market dynamics at all useful in fx space. I can by now probably fill a book with my experiences but would like to do so on the basis of sharing experiences/thoughts. $\endgroup$ – Matt Feb 2 '16 at 8:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.