So basically you're re-inventing the wheel and building another LMAX ? With all due respect, the fact that you have to come to an anonymous forum to ask simple questions as to how to build your Forex exchange means you're really only thinking about the tip of the iceberg as far as problems you'll have to handle. The words "I'm building" are also somewhat concerning ? This is not the stuff you can do on your own in your back room, you need some serious budget for staff and infrastructure if you're seriously thinking about competing with the big boys. You've already alluded to the fact you can't even afford a Bloomberg ("Not sure what Bloomberg uses; if anyone knows please tell me."), so what about all the expensive data feeds you need ?
What's the difference between LMAX and Matchbook FX? There are already plenty of ECNs out there for the professional Buy side (see EBS, RFX, etc...).
Nono, another FastMatch! Not LMAX since they use 'strict price/time' order matching. Matchbook was before my time, but from a quick look at the wiki link, I didn't see anything that mentioned 'randomization' mechanisms for matching. Not surprising though, if it closed in 2009, it really predated the need for something like that. But truth is...if I'm not mistaken, EBS even recently implemented a 'randomization' of sorts to address OP's issue.
OP's impetus for wanting to make this exchange (pretty ambitious task, if you ask me) was that he wanted a venue where it wouldn't be all about speed. To have a mechanism to serve as a speedbump of sorts, to level the playing field for all participants is his idea (sort of an IEX for FX). Most ECNs don't address this issue - there are some, however, that do.
That's where you come in . You're going to tell me the right way to do this. Now, can you please answer the question? How does this look? Code: Order Number Base Counter Type Action Quantity Price 519564758564015_ABC GBP USD LMT BUY 100 124.93
Yes, this is the key point, isn't it? Can a system be designed that levels the playing field, without introducing delay? The only answer that I've come up with is a system designed for algorithmic trading, that equitably runs everyone's algos directly on the server. That invalidates the need for colocation, because everyone's algos are running locally. It doesn't matter if you're a retail trader with a dial-up connection, or a major investment bank with a colocated fiber link. Either will upload their algo, run it directly on the server, and they both have the same access speed and privileges to both market data and the order pools. In a race condition, when two algos are fighting to buy/sell the same financial instrument, the next order to be matched is randomly selected. "Non-display fees" actually gave me the idea. Recall when the exchanges cranked up the fees for algos so unbelievably high? Like on the NYSE, USD70/month/subscriber for a standard display, vs USD6000 per month for non-display? At first I was furious: I almost cracked my monitor with my cursing. That actually prompted me to join ET; check this post: https://www.elitetrader.com/et/threads/non-display-fees.301156/#post-4403250 But then light bulb went on: there is a vulnerability there: algos level the playing field to the point where they must be paywalled. They don't want people writing algorithms; it's too easy for a smart developer/quant/mathematician to rake in profits. Ritchie will do the exact opposite: encourage and support those writing algos. By the way, no "non-display fees."
Forex trading has actually been around since the 90's believe it or not and REALLY flourished with the advent of internet and REALLY REALLY took off after SEC instilled a $25K minimum capital requirement for pattern-day trading that killed off the stock daytrading and all the stock daytraders just flocked over to forex trading. Back in the days before the 2008 crash, the daily move in forex was so high that you could easily make several hundred pips in just 30 minutes. It was common that a currency pair would move like few cents in a day!! There was so much volume; the daily volume was $5 Trillion with retail fx accounting to several hundred billion dollars a DAY!!! So yes a central exchange was needed. I mean none of the financial instruments have such large daily volume turnover and they are all traded on central exchanges and something so big like forex with such high volume are still traded OTC??!! The crux of the issue is NOT an issue of necessity; it's an issue of sharing of wealth. With the forex market dominated by a few big banks who are the ultimate market makers; they have the whole control of the market with complete dominance over volume and thus profit. Do they want to trade on an open exchange to share everything with the joe schemo of the general public or do they want to confine the wealth in the hands of the selected few? Before 2008, because of the tremendous volume and revenue, they didn't mind sharing especially they were paid handsomely with the profit-sharing scheme of the exchange but after 2008, when all of the money were gone, they just don't want to share anymore. And when they are not on board, there is no exchange. So to me the most important is to get the big players on board to get flows, volumes.
I though I'd just clarify what I meant by 'predating the need for randomization' ... Just that in the mid 2000's when matchbook existed...HFT was not yet so pervasive (and certainly not in FX) that it was necessary to institute any mechanisms for randomization (basiclly slowing things down) in order to democratize the playing field (order book). I certainly didn't mean to imply that there wasn't adequate volume, in case that is how you understood it.