June 6, 2014 - Thank you for that kind introduction. And my sincere thanks to everyone here for joining this lunchtime session. It's a great pleasure and privilege to return to New York. I want to offer a few reflections this afternoon on market innovation and, in particular, the on-going debate around high frequency trading (HFT), which has dominated global media over the last few months.

For regulators of course, the interest here is not simply around the novelty of 21st century technology (as important as this is) it's also in the link between speed, market fairness, efficiency and safety - which is clearly a much longer-running saga in financial services.

And I feel able to speak with at least a little authority on this, having spent much of my career at the London Stock Exchange. And the London Stock Exchange was, of course, the venue for the first high frequency trade.

Now, I sense a little scepticism here - I am sure you are asking how anybody could be so definitive on such a fast-moving topic. Even the experts can't quite agree a firm definition of what a high frequency trade is, so let me explain.

I think everybody agrees that it is made possible by the harnessing of superior technology. The race for speed has been critical. It usually relies on the ability to capture, process and respond to information more quickly than your rivals. Proximity to the execution engine is often a feature.

And, it is alleged, there are practices that are less palatable, such as 'momentum ignition' - essentially, creating a false market so that you can benefit as it rebounds.

All of these factors were present in the very first high frequency trade carried out on the London Stock Exchange on 19 June 1815. Nathan Rothschild, of the great house of Rothschild, had built a spy network over Europe to get the best intelligence as events unfolded - in this case the Battle of Waterloo.

So, he employed faster horses with more frequent changeover points - superior communications and technology; he 'co-located' himself on the floor of the Stock Exchange; and (shockingly) he initially short sold small parcels of gilts to create a downward market (momentum ignition) before proceeding to buy-up - before the position news reached the market.

Now, it wasn't quite the fortune of financiaal folklore. Nor was it carrier pigeons as you once found on Wikipedia - the font of all knowledge. But the driving motivations are exactly the same as we see today.

In other words, while the horses have been replaced by fibre optics, the broad issue here has never really changed - only the speed boundaries, which now seem to be physical laws rather than technical limitation.

And this, I think, creates some very interesting questions for all of us involved in global markets - irrespective of where we stand within them.

One of the most important perhaps: is there a point when the costs or risk of squeezing the last drop of innovation out of the market creates results that are unacceptable?

Or, a slightly different proposition, does it always enhance market efficiency? Many - including Paul Volcker and Paul Krugman - have questioned if there's any correlation between the sophistication of an economy's financial system and its productivity growth.

And yet, almost irrespective of these arguments, this sophistry if you like, we know tomorrow will not look like today in the markets. Innovation is relentless.

Up next, no doubt, more microwave networks, as well as more server farms located next to exchanges. And, perhaps closer than we think, that learning algorithms and self-improving artificial intelligence are the prime decision makers in electronic markets.

And, an important question here for both regulators and governments - how do we keep pace with all this innovation? Can the official sector front foot change? And, if, as I suspect most would agree, it is impossible, how then do you ensure the risks are managed effectively?

Regulators have, I think, been largely 'technology neutral' - for want of a better phrase - up to now. But this doesn't mean it isn't important to analyse whether innovations threaten a fair and efficient market: which is very much in regulated territory.

Benefits and risks

So, many interesting questions here for leaders today, in industry as well as in government and the authorities.

And a priority challenge for HFT specifically, which I'm not sure has yet been honestly assessed by all players, is where the balance lies between the potential benefits against costs - and how we manage that risk/reward equation. Key areas to consider here: market efficiency; fairness; cleanliness and resilience.

So, on the one hand, there are undoubted benefits to HFT. Most notably, the link between competition and market efficiency, as well as liquidity from reductions in bid/offer spreads and reduced transaction costs.

Most of these benefits, of course, have already been very well analysed and researched, with various official reviews by the likes of Sweden and Australia, as well as the UK Government's Foresight Report in 2012.

So, it's important to be clear sighted here about the possibilities of HFT. Even if it's also an imperative to appreciate the potential risks as we move things forward: three key ones strike me.

First, looking at market fairness, there's the inevitable debate around the impact of speed on market fairness, with all those familiar concerns around unfair advantage for the few over the many - as well as nervousness around conflicts of interests.

Second, in terms of market cleanliness, it is sensible I think to ask whether this type of trading opens up greater opportunities for abusive behaviour.

And third, around market resilience, what price on the possibility that automated trading creates flash crash-type scenarios in the future if algorithms spin out of control? Or, indeed, if assumed liquidity were to disappear at the drop of a hat.

Clearly, it's no easy task for global leaders - whether in regulation, politics or industry - to find a perfect balance to the risk/reward equation here. And this perhaps explains why the international response differs across borders.

So, for example, Italy has introduced a tax at two basis points on orders when the order-to-executed trade ratio exceeds a certain threshold. Effectively, requiring HFT users to pay for the additional noise they're creating.

In Germany, similar charges have been introduced but legislators have gone further. Firms engaged in HFT activity are expected to apply for a licence to operate; fees are being charged for excessive system usage; and other tools, like minimum tick sizes and order-to-trade ratios, have also been introduced.

As we can see then, broadly similar HFT risks playing out differently in terms of global response: reflecting (to some extent) distinctions in markets and market structures as we jump from border to border.

And for me, one of the clearest of those distinctions has to be market fragmentation, which - judged purely by share of primary trading venue against total market - is an enormously varied picture.

So, on one end of the scale, we have the New York Stock Exchange at around 29%. On the other, Hong Kong at 100% - there are reasons for this. And, lying somewhere in between, a wide dispersion of percentages from London to Paris, Milan to Frankfurt, Sydney to Singapore.

Hard to see then, in that context, how a one-size-fits-all regulatory response could realistically function. Indeed even between the two most strikingly similar financial centres: the US and UK, where there are similarities in so many areas - not least shared strengths and philosophies - in HFT, the physiognomy looks different in several important respects.

Again three key areas stand out for me. First, the fact that the US equity market is generally more geographically dispersed than in it is in the UK and EU. Second, that investor best execution requirements are also handled differently. And, finally, dark trading in the UK is comparatively low as a percentage of trading compared to the volumes we see over here.

A disparity that is, we suspect, going to increase with the advent of MIFID, which will place a cap on orders that can be executed away from lit orderbooks.

So, different territories, facing similar types of risk, but with different market micro-structures - leading, in turn, to subtly different solutions.

UK policy

And this confusion is, of course, amplified by the great unknown of technology. So, if history is any guide - it will be neither easy to guess where innovation bubbles up next. Nor the outer limits of its reach.

Many years ago now, the Chair of IBM famously guessed that 'there is a world market for maybe five computers'. Just 60 years later and the billionth personal computer had been shipped.

Now, I'd argue that if the tech-industry sometimes struggles to spot the big trends, it's unlikely the financial markets or official sector will do significantly better. But this doesn't mean it's impossible to manage innovation or unimportant to react to change.

And, given the high stakes involved here, I'm not surprised to see world regulators taking the risks posed by HFT very seriously. In the UK this is certainly the case - with what is effectively a three-pronged response in place.

So, we approach the kind of risks posed by market activities, like HFT activity, through a blend of analysis-led policymaking to get rules in shape, as well as through day-to-day supervision of relevant firms and markets, and active market surveillance.

On the first of those three, so policy-led analysis, Europe has already set out, through a comprehensive set of European Securities and Markets Authority (ESMA) guidelines, detailed expectations around systems and controls for both trading venues, as well firms using these techniques.

The bulk of policy work now will be absorbed in gearing the UK up for MIFID II implementation in two years' time - which is bringing in a long, and complex list of new rules. And this high level of involvement will, clearly, continue across the whole dossier through the various stages, including implementation and beyond.

Where will this leave us? In the words of the EU Commission, with the 'toughest' package of measures in the world. Time will tell what others do, of course, But at the moment, I'd say this assessment looks about right.

MIFID effectively builds an extra layer of security onto the existing ESMA guidelines for firms. And for me, there are a few areas to flag that look as if they'll make the system around HFT safer, or more resilient, if you will.

So, for example, there'll be significant strengthening of algo testing prior to deployment between firms and venues. There'll also be more focus on systems and controls in firms, with the objective of making sure they understand and take responsibility for the risk they import to the market.

On top of this, venues and firms will be required to have circuit breakers, or 'kill switches', to stop runaway algos. And, to mitigate the risk that HFT's high messaging rate overloads the system, European regulators will impose order-to-trade ratios and minimum tick sizes: helping to control noise created by ephemeral orders.

Finally, a few other priority areas worth mentioning I think:

First, rules around market making agreements, which will impose liquidity provision requirements.

Second, broader measures designed to give regulators a better overview of what's happening in the marketplace, which include requirements for pure HFT firms to be authorised, as well as better audit trails for orders and trades - by firms and venues.

Finally, there'll be regulation of trading venues' fee structures; publication of data related to the quality of execution for each financial instrument traded; and caps on the amount of 'dark trading' that can be undertaken.

Supervision

Now, this may all sound rather dry and technical - welcome to my world - but it is very important in terms of reducing our risk of exposure to the likes of flash crashes - or indeed so-called 'hash-crashes'. Fat finger mistakes, if you will.

Having said this of course, policy is generally only as effective as the supervision around it. And this is one of the reasons why I have colleagues overseeing trading venues, and why the FCA has, for quite some time, focussed on big-ticket issues like market resilience; fair and open access to trading facilities and data feeds; and making sure appropriate systems and controls are in place to ensure market integrity and stability.

So, for example, there's already been significant work across areas such as making sure venues have the ability to suspend trading - and also have risk-based circuit breakers in place.

In terms of how this operates on the ground, the FCA maintains a relationship with the range of firms undertaking HFT activity - right across the spectrum from universal banks to very small prop traders.

And, post-crisis, much of our firm specific supervision is focussing on the business models of those firms, as well as the broader sector context.

Effectively then, what we have now in the UK is a risk based approach. So, depending on the level of risk-posed, oversight might range from general monitoring for compliance with ESMA guidelines on automated trading - all the way through to 'deep dive' assessments of specific firms. Perhaps, say, to understand the risks associated with the development of algos for use in HFT.

On top of this, we also undertake broader investigations into subjects that can relate, directly or indirectly, to HFT - such as market fairness and market resilience.

Key questions here for me in all of the above: what is the right level of oversight in a market like this? How do we make sure controls are consistent? How do we make sure they are both comprehensive and robust?

And, perhaps most important, how do we make sure firms, heavily motivated by financial return, accept responsibility for the significant risk they import into the financial markets?

Market surveillance

Which brings me on to the final area I want to mention this afternoon, market surveillance. How do you monitor what some describe as the 'wild west' of HFT?

Clearly, this is not easy. Nor should we pretend otherwise. When the number of trades is significant, speed is high, and market structures are complex - it is invariably difficult to perform surveillance effectively.

In HFT, the sheer number of determinants used by algos compounds this challenge. So, while the nature of market abuse risk presented by HFT activity is not necessarily new - the speed and volume of messages (both order and trade) does add complexity, especially in a cross asset/cross border/cross market world.

In other words, pattern detection alerts need to be able to detect suspicious behaviour irrespective of the speed of trading. And these alerts are becoming ever more complex in order to maintain the accuracy.

Alongside this of course, the higher messaging flow also means the capacity of surveillance systems has to increase in line with the trading technology and consequently the costs of surveillance rise too.

So it becomes important, I think, not to let the best become the enemy of the good. Perfection is, frankly, an impossibility and over the course of many, many years, it has become clear that achieving market cleanliness is perhaps one of the toughest challenges facing regulators.

Having said this, significant progress has been made in recent years by regulators like the FCA. So, essentially what we have now in the UK, is a mix of exchange-led monitoring, with the regulator analysing risks such as cross-market techniques on the one hand. On the other, industry itself reporting suspicious activity - so the challenge here becomes a shared one.

And clearly there have already been some important victories for industry in terms of enforcement. In the UK, for example, the Michael Coscia case, which involved the use of algorithmic programmes to layer thousands of false orders on the ICE Futures Europe exchange.

Looking further ahead, detection of this kind may well become easier. We know, for example, that MIFID and the Market Abuse Regulation will help the UK collect more data and information and, as a result, help us look into more areas and cases.

The key issue here: venues will be required to keep data in standard formats which they will then be obliged to make available, on demand.

On top of this, there'll also be a requirement to submit suspicious order reports (to complement the existing suspicious transaction reports).

More difficult perhaps from our perspective, is the pressure created by increasing regulatory scope, with a whole tranche of financial instruments are brought into range for market abuse and these, in turn, are likely to put some extra pressure on regulatory supervision.

Conclusion

So, to wrap up here, a great deal of activity taking place around the world in the face of market innovation in areas like HFT. Just as Rothschild used advantages he spotted in 1815 - people will always look for an edge.

But I would also argue, as I look at the very specific UK market, that it is important to take a step back, and reflect on automated trading coolly and not get carried away by headlines.

We do not want to be in a position where acting in haste precedes repenting at leisure. Innovation will always bring with it some risk. Sometimes too much risk.

But it also creates serious possibilities to explore. Finding the balance between those two poles is one of the great financial challenges of our time, with the most far-reaching consequences for the many billions of investors, savers and pension holders around the world.

So, no easy task. But an important one and one we should all be excited to be a part of.