It’s been a couple years since I first read Professor Donald Mackenzie’s paper “Mechanizing the Merc: The Chicago Mercantile Exchange and the Rise of High – Frequency Trading” via Alexandre Laumonier’s excellent and very resourceful Sniper in Mahwah blog, but have refrained from writing about it myself because there was so much to comment about it. My arrival into the futures industry was shortly after the e-mini S&P was introduced in 1997 (almost 19 years ago exactly today) but I did have fortunate timing to be an active trading participant as the eurodollar contract transitioned from the pit to the screen.
Professor Mackenzie’s paper describes many details pertaining to the liquidity of both derivatives shifting to the screen but the paper also has some a few issues I’d like to comment and supplement with my perspective. The paper begins with a background on the Flash Crash of 2010, descriptions of trading within the pit, origins of Globex, the necessity in learning how to find a trading platform, and the launch of the e-mini S&P but for brevity, I’ll only address the final section on eurodollars because that specifically allows me to speak from a great deal of personal experience.
Eurodollars started access for side by side trading (electronic trading allowed during pit hours) in 1999 but it wasn’t until the latter part of 2003 when it started to get traction for a liquidity shift towards the screen. Between those two points, the liquidity followed a standard daily shift between the pits in Chicago and Singapore with Globex filling the gaps. A rough estimate of the shift followed this pattern based upon Central Standard Time: 5pm Globex reopen held liquidity for light trading until SIMEX pits opened around 6pm and the Globex screen emptied out almost entirely after SIMEX opened until it closed around 6am, after which Globex got repopulated w/orders and moderate trading until the 7:20am pit open which shifted liquidity back to the Chicago pit, once again Globex emptied out except the occasional order someone forgot to pull and remained blank until the pit closed at 2pm and liquidity once again shifted to Globex until the close at 4pm. Some exceptions to this schedule occurred such as when SIMEX celebrated Chinese New Year and their trading floor was closed so Globex held liquidity for that entire session.
There weren’t any barriers to keep liquidity trapped in the pits away from the screen but it took a few years before conditions really ripened for the transition to happen. Foremost, Globex was more expensive to trade actively in eurodollars (15-20 cents/contract more than the pit for locals I want to say) so that dampened initiative to execute on the screen than in the pits. E-mini S&Ps were also subject to Globex fees, on top of CME clearing and clearing brokerage fees, but those were capped at an absurdly low $50/day until 2009 (!!!) whereas active eurodollar traders regularly paid thousands each day without the same fee cap once the contract shifted to the screen. Some rebate initiatives for market makers were eventually established but rebates were minor in comparison to fees paid.
While the eurodollars continued to trade in the pit, a new generation of traders was emerging to provide liquidity on the screen. Trading arcades, such as Macfutures, arose with dozens to hundreds of traders in locations like Singapore, Gibraltar, London, Sydney, Montreal, etc… who traded not just Globex products like the e-minis but more so Eurex and LIFFE contracts which were both entirely electronic at that point. Eventually, as was noted within the paper, these European exchanges sought to leverage their user base to take liquidity from the CBOT and CME in Treasury futures and eurodollars, respectively. Now that that conditions were very fertile in late 2003 for a liquidity shift to the screen in eurodollars, all it took was a first mover make it happen.
One of the most frustrating things about reading Prof. Mackenzie’s paper is it neglects specific details on how actually liquidity shifted and makes it appear like it was a top down directive which virtually forced the transition when that wasn’t the case at all. The true story of how eurodollar liquidity shifted to the screen is much more organic: a couple guys started to make markets for the entire trading session on the screen and that was the catalyst which had been missing for so many years.
One random day in early 2004 (or perhaps late 2003), seated at the top row above of the tiers of brokerage desks on the west side of the eurodollar pit were two guys who remained sitting at Globex terminals when others left to walk into the pit. The clearing firm on their badge was 287 which meant they had an institutional connection of some sort because 287 signifies the firm E.D. & F. Man which didn’t clear locals. Looking at the Globex market shown on the wallboards, their intentions were clear as two sided markets were posted whereas previously it’d become blank after the pit opened. Initially they focused on the front couple months which had the deepest liquidity and were most static, even so they made markets a half tick outside the pit bid/ask initially on thin size of 100 or under. As they got more comfortable in the ensuing days, they’d tighten up their markets to the point that it’d match the bid/ask of the pit. Whenever the market would flare up, the market makers would hit their Scroll Lock key which cancelled all orders on Globex and once they’d get a handle on where the market stabilized, they’d proceed to make two sided markets again.
Surprisingly, there was little animosity from the pit when the arbitrage began because some pit brokers were readily accepting the orders from the electronic market makers. At some point the transition to the screen was going to begin and two sided electronic markets signified the beginning of the end for the eurodollar pit. That first day of these two sided markets, I was standing in the spreaders section of the eurodollar pit and the sentiment was a mix of wondering if others would actually start trading against these markets along w/the resignation that the electronification process was getting a serious push.
Straightforward arbitrage is such a fleeting strategy that I’m not sure how long these original market makers were able to profit from it because within a year the liquidity was sufficient that I stopped going into the trading pit anymore by 2005. In the interim, their markets attracted small orders because executing on the screen was empowering in it’s disintermediation away from the trading pit. Trading is a copy cat business and it wasn’t too long before locals in the pit utilized handheld devices to post their markets on the screen which they were also trying to work in the pit, not necessarily making two sided markets but for the chance of getting hit on something good they needed, it was another tool to utilize. After the CME eurodollars established daytime liquidity on the screen, SIMEX suffered the same fate as the Chicago pit and the primary liquidity pool for the contract transitioned quickly to Globex for the entire 23 hour trading cycle.
One significant development which really added electronic volumes, but also wasn’t mentioned in the paper, is the early 2004 creation of CME’s incentive programs which allowed new participants to trade at a reduced rate but without having to initially purchase a standard membership. These programs were initially supposed to be temporary but like a temporary government program, continue to exist in perpetuity. I don’t believe CME ever released numbers on how much volume the incentive programs contributed but it was clear from seeing so many individuals and firms sign up on the weekly membership bulletin, that it was a gateway for many to expand into Globex markets.
A few additional comments on the paper I’d also like to make:
- CME initiated a proxy vote to members (B share trading rights) in February 2004 about authorization for the exchange to forcefully transition the first two eurodollar quarterly contracts to the screen if they didn’t trade 25% of volume during regular trading hours by mid March 2004. In January, for the entire eurodollar contract, Globex represented 10% of volumes and most of that was concentrated in the first few months and by the time this proxy was presented I recall the volumes being very close to the 25% threshold already, by the time mid March came around I’d estimate well over two thirds the volumes in the first two quarterlies was done on the screen so the vote was nothing but a formality to show intentions. One additional important factor that underscored the passing of this proxy along w/wider acceptance of electronic trading was that CME equity stock was a consistently strong performer and many members held significant amounts which would appreciate in value as electronic volumes rose.
- A bizarre quote is included about how a trader from another pit was amazed at how a eurodollar trader with 100 contracts on could step away from the pit and go to the bathroom. Immediately preceding that quote was a contrary comment about how the eurodollar market was frequently static and barely moved, sometimes remaining the same bid/ask throughout the day. I don’t know if the author misunderstood the speaker but there wasn’t a trader in the entire eurodollar pit who would’ve felt uncomfortable for a bathroom break w/100 lots on. Even someone leaving for a week vacation with 100 outrights on wouldn’t solicit any reaction considering the positions locals had on easily went into the thousands, occasionally tens of thousands and for the elite over a hundred thousand, hedged of course.
- Incorrectly, it’s stated that Globex top of book queue priority wasn’t enabled until after the CME merged w/the CBOT. Since I started using Globex in 2000, there was always top of book priority for turning a market or making the best price in it. This was a huge advantage which was used in dual ways to establish priority and also know when someone else established it to look to knock their priority out so the allocation became pro-rata (for instance, if an offer traded out but didn’t go bid, it was possible to bid a 1 lot and cancel so if the same offer was reestablished, the bid allocation switched to pro-rata, since it was momentarily bettered by the 1 lot, rather than the top of book order getting 100% priority).
I could continue to ramble on about this, and may do so in additional edits. Here’s a few visuals to go along w/the paper.
Above is the chart for monthly electronic eurodollar volumes and you can see an initial rise in late 2003 with a significant acceleration in early 2004. I attempted to do an overlay against pit volume but couldn’t get it the way I wanted to further illustrate. It should also be noted that before liquidity shifted, the volumes shown were entirely done in the gaps between SIMEX and CME trading pits being open.
The once crowded eurodollar pit was reduced to less than half a dozen people on the final trade last summer.
Also, I have all the Globex reference stuff regarding fees and user manual from 1999 which is of the lead screenshot but it’s too much stuff to upload without a specific reason to. Someday I might when more time is available because some of Globex’s other O.G.s might enjoy the walk down memory lane. There’s also so many hilarious stories to share about the early and comparatively primitive days of Globex, it used to be such a small community trading it and seeing counterparty IDs made it so much more information rich. I’ll leave “true early globex stories” for another post in the future.