Well… this is a no-ending series. The very last part of “HFT in y backyard”, titled “Miscellanea”, is nearly finished, but before posting it some few words about… hawks. Thanks to Twitter and some algos I run to know the last informations concerning the HFT microwave networks in my backyard, I came accross a new firm named “Haawks”. The company “provide highly specialized trading and it services for prop shops, money managers and banks”, the most interesting being this “market research on low latency arbitrage trading” – and it’s all about microwave. When I read on the website that the company is incorporated in… Malta, I wondered if the people behind Haawks were involved (somehow) with another firm from Malta I talked about in Part IV (there is no name on the website) but no, Haawks seems to be a newcomer in the small world of law latency trading. That’s interesting to know that the firm was incorporated on September 15, 2014, a few days before I posted the first part of “HFT in my backyard”. That means I was not alone to work on these networks during last summer. But there is a big difference: I have made available all my data for free (including the Google Earth map) whereas Haawks is a consulting firm and, well, a consulting firm doesn’t work for free.
The Market Research on Low Latency Arbitrage Trading was announced to be released on February 4, 2015. This is said to be a “powerful research paper with the following up-to-date facts” regarding what you need to build a microwave network (routes, licenses numbers, antennas, company names, towers, etc.). I subscribed in order to pre-order the paper but got no news, so I contacted Haawks to know more about the paper, saying that we could share information. I received a kind answer (“we like your blog and wish you all the best”) but “unfortunately [Haawks] cannot share any information about [their] research”. I was not really surprised as some data may be really bankable. But which data are we talking about? We can only speculate on that by reading the summary of the paper.
Haawks writes that “nowadays POP latencies of commercial network providers are more and more publicly disclosed on their website”, citing Quincy Data (the sister company of McKay Brothers). It is true that Quincy Data latencies are public but I didn’t find another microwave firm offering such a transparency [correction: Perseus also publish their latencies]; what’s more, Quincy latencies are from rack to rack, that means from a colocated server (in Slough for example) to another colocated server (in Frankfurt for example), so the latencies include transportation time from the dishes installed on the roof of the exchanges (or on a tower near the exchange) to the server where algorithms are implemented. One issue raised when I was investigating the microwave network: who is the fastest? That’s nearly impossible to answer (even if some say Optiver may be the fastest in Europe) as calculating the exact latencies is a very difficult task. But the Haawks paper is said to include the “calculated latencies”, so I would be very interested to know how the company can calculate with precision those latencies.
“Are there any big investment opportunities?” Haawks asks, “the answer is YES, there is a huge investment opportunity in microwave networks in Europe but the opportunity is narrowing by the minute”. That may be true: in my (humble) opinion the microwave business is nearly dead now, in the sense that various competitors have built different networks, McKay Brothers is currently completing its own and it seems Nexcom is also trying to build one between Cornwall and Frankfurt. Given that most of the competitors need more or less the same towers, some of them may be saturated by dishes. Also, some licenses may be not available anymore. “To our knowledge one big player already bought 2 licenses and installed antennas on a very crucial route late in 2014.” How interesting. Who is this big player? I don’t know but I wouldn’t be surprised if the “very crucial route” would be around the Channel. Jump Trading asked for a new Ofcom license on December 2, 2014, to cross the channel from the now famous Houtem tower but I doubt Jump is this “big player”. Most of the licenses used to cross the channel were granted before the end of 2014. There is another possibility: “The potential NY4 [Secaucus data center in the US] to LD4 [Slough in the UK] latency is around 29.6 milliseconds. This would be a reduction of 1.5 milliseconds compared to Quincy Data latency of 31.109 milliseconds.” If I understand correctly, that would mean this big player is not McKay/Quincy and will be faster than them. “Due to technical limitations of this route there are probably only one or two licenses left”. What are these “technical limitations”? The complex issues to cross the channel? Well, I’ll never know…
But what I know is the Haawks people are meeting various HFT competitors these days, trying to sell their paper to who may be interested. I wish them well. Given the price they ask for the research, I wonder if I made a mistake by making my own research free and accessible. I learnt that my little investigation may have been really bankable, so it is therefore possible that I am foolish. I’m not sure I’ll have another opportunity to make money with high-frequency trading, and it’s a real shame!