Will Psomadelis, Head of Trading, Schroders Australia and Stuart Baden Powell, RBC Capital Markets discuss the effects of artificial liquidity providers on the trading costs, speed of trading and market structure.
Execution venues are frequently judged on market share, but perhaps genuine value-added to the price formation process (not just the spread), reliability and overall liquidity quality should also be factored in. Bringing these points together, a question to pose is what the purchase price for Chi-X Europe would have been if it were itself publicly traded, exposed to full analyst transparency and recommendations, or even not principally owned by investment banks and HFT?
Dropping a level, practice-based publicly available knowledge on HFT is still limited. ESMA has noted the primary techniques being “usually either quasi market making or arbitraging”, some would note that the arbitrage could refer to an evolution of ‘non high frequency statistical arbitrage’, others would add latency and rebate arbitrage to the mix. Either way, these are mere high level descriptives and are not representative of what happens underneath; like algorithms it is not the flyer or website description that is key and like many things in life, it is the details and depth that matter. One of those details discussed in more cutting edge circles is the concept of artificial liquidity.
Will, you run a trading desk for a highly respected institutional fund manager; could you talk us through what artificial liquidity is about and how it impacts on you and the market?
Will Psomadelis: Thanks Stuart. Artificial and natural liquidity, being polar opposites in terms of the quality spectrum, are the result of two different strategies and are generally characterised by differing trade durations (or holding periods). Natural liquidity providers are generally investors that deploy capital in the market to extract a return from an underlying business. These include traditional long-only funds, retail, hedge funds and some fundamentally driven quant funds that invest for periods longer than a few minutes.
Artificial liquidity providers (ALPs) ignore company fundamentals and therefore make their decisions on metrics such as price momentum, correlations or by extracting a return through rebates or commissions to name a few. Market maker liquidity is a prime example of this type of churn that is held up as the Holy Grail by some regulators and the one that I believe doesn’t improve our transaction costs. Whilst arguments have been made that the retail investor can benefit from increased churn courtesy of HFT artificial liquidity primarily through tighter spreads, we should remember that no market-wide benefit can be extracted as spread compression is a zero sum game.
It can be then argued that increased artificial liquidity, which is what we are seeing globally, contributes to the deterioration of price discovery. When market volumes are dominated by trades that have no fundamental basis, stocks can move independently of underlying fundamentals. Empirical evidence shows that stocks now tend to overshoot fundamental news (X.F. Zhang, 2010), ultimately detracting from market efficiency and adding to volatility.
At the institutional order size level, remembering we are really just representing pools of retail investors, we can add to the points discussed above our belief that HFT cannot reduce market impact. The typical institutional order will usually have an order duration of greater than the entire holding duration of a position by a market maker, meaning every trade where we are trading against artificial liquidity, creates a competitor…..
At some point during the duration of the institutional order, the market maker will need to cover their position and therefore compete for stock. This is a real issue when ~70% of volumes are HFT in nature (Tabb Group, 2010 on the US market). Put simply, in the absence of someone willing to withdraw capital from the market at price, it is impossible to inject capital without incurring market impact, even if millions of shares are being churned around you during a game of high frequency ‘pass the parcel’.
The fact that market makers vanish in situations like the Flash Crash and create liquidity vacuums proves that natural investors should not be relying on them for liquidity as liquidity is not really what they provide. Non-predatory artificial liquidity may not necessarily add cost in theory, but it also cannot improve our transaction costs.
Another point is regarding the market share numbers of HFT. In Europe, many lean on the widely accepted 35% of the overall market number. AFME recently released an insightful study into the OTC Market in Europe to follow the UK version conducted by TABB Group (in part using AFME data). The report noted that approximately 60% of “all MiFID OTC equity trades were duplicate trades” or “reporting events” such as give ups/ins and principal trades and were thus “non-executable”.
The key point here is that this non-executable business should now be stripped out of the divisor for the market share calculation for HFT. Indeed as the TABB report itself notes “it is the executable liquidity that is most relevant and points to the true size of the market”. If we take the TABB piece as an example and strip out its own non-executable OTC numbers from the total size of the market, HFT in the UK is no longer 35% of the ‘true’ market but rather in reality higher than 50%. This is a sea change and one that could reset understanding and impacts.
Will, parts of Asia are experiencing their own execution venue competitions. What are the key points that can stimulate artificial liquidity and is there anything the buy-side can do to combat what must be a cost?
WP: The primary objective of a stock exchange should be to provide a stable platform for sellers and buyers to trade. Unless the incumbent exchanges are working at capacity (which they aren’t) or gouging prices then the entrance of a new exchange cannot increase the number of natural liquidity participants as there are no existing capacity constraints. By definition, new exchanges that promise to improve liquidity can only really stimulate the ‘artificial liquidity’ we mentioned earlier.
New exchanges are mostly citing speed as their advantage, under the pretence that speed itself increases liquidity and efficiency. Microsecond trading should not be a focus of a capital investor so, in essence, the only people that need ultra low latency to generate returns are artificial liquidity providers or – if we look at these participants more cynically – those taking advantage of information opportunities.
Whilst new exchanges may have lowered the explicit costs for brokers and improved their margins, these savings can be matched or exceeded by the implicit costs the buy-side may incur through the most cynical strategies that seem to flourish in this new environment. If new exchanges mostly bring artificial liquidity then the buy-side has to change its behaviour. Clearly, the best way to reduce market impact is to find another investor willing to extract capital that you are willing to inject (or vice versa). Block trading through the upstairs market (sales traders and block dark pools) is therefore the cheapest way to trade at an implicit price level.
Fong, Madhavan and Swan (2002) show that the ‘upstairs market’ is ‘pareto improving’ as investors’ trading blocks can reduce their market impact at no increase in total cost to other market participants. That’s right, block trades are a net benefit to the market! In essence, as artificial liquidity can’t actually reduce transaction costs on block volumes, the buy-side community needs to realize that working ‘over-the-day’ orders is not the lowest cost alternative they think it is (QSG, ‘Beware the VWAP trap’, Nov 2009). Block volume trades should be manufactured in the upstairs market or traded in block dark pools to reduce interaction with churn. The upstairs market needs to become the primary weapon in the buy-side trader arsenal to kill transaction costs and improve efficiency at the block trade level.
SBP: As a final word, a lot of people are talking about ‘intended’ and ‘unintended’ consequences in relation to HFT and/or any pending regulation. As Scottish economist Adam Smith himself noted “it is not from the benevolence of the butcher or baker that we expect our dinner”, “but from their regard to their own self interest”. Smith went on to state that “we are led by an invisible hand to promote an end which has no part of his intention”. That end is self-interest.
With that and concepts such as artificial liquidity in mind, there is a genuine need for regulators and the buy-side to be cognizant of those points when engaging with members of the industry. Both should move deeper into the inner mathematical workings to unpack HFT trading techniques. Understanding this will help answer the question of whether HFT really does feed the market with liquidity or rather deters natural liquidity, harms the end investor and increases the fragility of a fracturing market micro structure.
