But does this mean that the electronic platform, LME Select, and “one-lot wonders” are the way forward for metal traders and the industry, and that larger business flows will diminish? Not necessarily – it could also be a case of “back to the future,” with the inter-office phone market and, importantly, the open-outcry ring still having a key role to play on the London Metal Exchange.
Guy Wolf, head of market analytics for LME ring dealing member Marex Spectron, said in an interesting presentation in Hong Kong that LME volumes have climbed while technology has taken over the market but that humans are realizing that this marketplace is not a suitable venue for executing larger orders – of 20 lots, for example.
Specifically, average transaction size on LME Select has fallen to 1.4 lots from 1.7 lots in the last four years while the number of 10-lot transactions now accounts for 3.4% of turnover, down from 11%.
A random mid-morning look at the copper order book on the Select platform illustrates this. Market depth down to 15 inputs, encompassing a $9 top-to-bottom range, shows that the majority of resting bids and offers are of one or two lots, with the largest being a bid for seven lots. And within a 10-minute time span, all the transactions were of one lot. This pattern is largely the norm now.
Where algorithmic trading is concerned, the thinking process is straightforward – the trade is executed almost instantly in line with how the code is programmed. But these programs have become much more sophisticated and technically advanced in the 21st century – savvy, if you like – with artificial intelligence.
The computer is now geared to analyze, discern and respond to inputs scraped from news developments and headlines. These could be wider political and economic trends – what US President Donald Trump says or does. Or, in future, even sector-related news – inventory movements and warrant cancellations can be forensically analyzed to generate a quick “buy” or “sell” one-lot trade.
Again, this takes place almost instantly. This is fine for a computer but for a trader holding a mouse and looking at several screens by the time he or she points and clicks, even for just one lot of copper, the window of opportunity has open and closed – perhaps many times.
And it will be much harder trying to get larger lot-size orders away without spooking the market – placing a 10-lot or 20-lot order in a typical pre-market would almost certainly create some ripples.
Timing plays a part, too. For a trader, reaction times at the end of a long, busy day will almost certainly be a bit slower, which is human nature. The computer, by contrast, won’t get tired, and order react-and-respond times will be the same throughout the trading day.
But Select is not the complete picture as a random look at the ticker for copper demonstrates. Inter-office trade transaction sizes, when they occur, are much larger – here you find average deals of 10 lots or more are the norm.
The open-outcry floor also lends itself to larger-transaction-size trading. Just watch the closing seconds of the official rings for cash business and the end-of-day valuations on the ring. Human traders are in the ascendancy and the lot sizes – particularly in an active, fast-moving market – are substantial.
That is not a surprise, as the history of UK equity market trading shows. On the old London Stock Exchange, when there was a floor, the bid/offer spreads were much closer, while the face-to-face nature of broker/jobber relationships meant that larger transactions were negotiated smoothly. But after the so-called “Big Bang” in 1986, when business migrated to screens, bid/offer ranges widened, which had a ripple effect for market depth – the screen, of course, is impersonal. The trading floor had closed so there was no alternative.
This is not the case with the LME because, uniquely in global financial exchanges, it still has its three platforms of screen, phone and floor. Although it is not possible now to drill down in the LME data to break down turnovers for each trade, some rough numbers were issued at times before the 2012 sale of the exchange to HKEX.
These sometimes showed that inter-office business for outrights and carry trades accounted for 40-50% of total daily business. The likelihood is that off-screen business remains a significant makeup of overall daily business.
So what could this mean for metal traders in future?
Over many years the very nature and structure of LME trading practices has evolved considerably and, as the LME’s Strategic Pathway shows, even more change is to come in the next 12-18 months.
Where Select is concerned, the algorithmic trader will surely become even more dominant with the plethora of one-lot deals. But an algorithm cannot pick up a phone and talk – not yet, anyway. So the inter-office arena looks like the favored venue for larger orders to be made smoothly and without undue disruption or spooking.
That leaves the open-outcry floor, which has seen ring dealing members decline to nine firms. Here, the LME has an intriguing trial coming up in 2019 of moving nickel’s end-of-day close-out to the screen for volume-weighted-average-price settlement for a three-month period. Like the 12:30pm-13:10pm official rings, the closing kerb is a pinch point for large amounts of business being transacted in the final seconds. It will be illuminating when the data is analyzed from this experiment to see the interplay between transaction sizes, buy/sell quotes and price fluctuations, and how it compares with the current model.
But traders are adaptable and flexible. As Wolf pointed out, humans have accepted they cannot battle this technology.
And because metal traders – unlike their counterparts in all other financial markets – have non-screen alternatives available, the chances are that these will continue to attract respectable business flows. They may even help to sustain open-outcry trading into the next decade.