Why Barclays lied: Thoughts from a dark pool operator

As the former head of electronic trading product management and quantitative strategies at Deutsche Bank, responsible for their dark pool in Europe, I wanted to add my own 2 cents to the noise surrounding Barclays and their dark pool, LX.

Eric Schneiderman, New York Attorney General, has some serious beef with Barclays because they filled their dark pool, Barclays LX, with high-frequency traders (HFTs) and then lied about the highly predatory nature of their trading strategies to their institutional clients. They erased key data from marketing material and gave private, sensitive information to HFTs about other participants in the pool.

As us brits would say, 'a rather silly move'.

What were their clients afraid of?

Institutional investors have a regulatory duty to prove best execution to their clients.They don't understand, nor can they truly measure the effect of HFTs on their trading performance. There are too many unknown unknowns. Furthermore, HFTs are highly secretive in nature and everything that is said about them tends to be pretty horrific.

Tradebot was (and possibly still is) one of the largest HFTs in LX. They feature in Schneiderman's complaint because Barclays deliberately removed them from a graph which profiled the participants in LX before presenting it to their institutional clients. Barclays knew Tradebot operated toxic trading strategies. Tradebot's homepage says the following:

The stock market is tough. It owes us nothing. It punishes our mistakes. Others have more money, more power, more connections. We are underdogs. We keep learning. We innovate. Every day is a new fight. Technology is our weapon. We make millions of small trades. We cut losses. We identify opportunities. We focus. The market can be beaten. We love the game.

Does that sound like the kind of friendly individual that you would choose to trade your apples and pears with? How about your hard earned retirement fund?

If Barclays' clients knew the extent and nature of HFT activity in LX, a lot of them would have opted out of trading in the pool. That is not to say that all HFT trading is 'bad'. I would argue there are 'good' HFT strategies and 'bad' HFT strategies. It's much easier to be safe then sorry though, isn't it?

So why did Barclays let predatory HFTs into LX in the first place if they knew it wasn't in the best interests of their institutional clients?

And why did they lie to stop their institutional clients from opting out of trading in the pool?

Bloomberg's Matt Levine suggests that:

  • The commission paid by HFTs to access LX made virtually no impact to the revenue of Barclays' equities division, so Barclays didn't do it for the extra cash.
  • Given the unidirectional trading nature of institutional clients, HFTs were needed to provide opposing liquidity within LX.
  • Without opposing liquidity from HFTs, there would be far less trading activity within the pool meaning the orders of institutional clients would not get traded and Barclays would therefore get less commission.

I only partly agree with Matt on this one.

Barclays pursued this strategy for 3 reasons:

Reason 1: Money

If HFTs were not allowed access to LX then it's true that there would be considerably less trading activity. However, most of the institutional orders within LX are placed there by execution algorithms that are using LX as a first port of call before sending the order out to external, lit exchanges such as NYSE or BATS. If there was less liquidity within LX then those algorithms would simply have traded more externally. The client would still have executed their order and Barclays would still have received their commission.

However, when trading on external exchanges, Barclays has to pay a fee. Aggressive trades on NYSE cost $0.0026 per share. And they will be trading against many of the same HFTs that are already present in LX (passive orders on NYSE receive a $0.0015 per share rebate which is a major incentive for high-frequency market makers). By inviting those HFTs to trade passively in LX for free, Barclays saves $0.0026 per share in execution fees on aggressive trades.

For passive trades, when trading in LX, Barclays lose the $0.0015 per share rebate they would get trading on NYSE, but they make up for some of that by charging HFTs $0.0002 to $0.0005 per share for aggressive trades.

If you assume that the ratio of passive orders to aggressive orders is 50:50, then Barclays saves approximately $0.0014 per share by trading institutional orders against HFTs in LX versus trading on NYSE.

$0.0014 per share cost savings doesn't sound like a lot. However, with client commissions for electronic orders in the order of $0.01 to $0.02 per share, this can have a significant impact to the bottom line of an electronic trading business. LX matches 285 million shares per week. This equates to approximately $20 million a year in cost savings. Not a huge amount compared to the $4 billion in equities revenue in 2013, but, due to constant client pressure to reduce commissions, equities is an incredibly low-margin business, and this is certainly enough to incentivise a heavily-siloed electronic trading business.

Reason 2: Market share

By opening the doors of LX to HFTs and other brokers, Barclays now has the second largest dark pool in the US by traded volume. This was Barclays' plan the whole time. They want(ed) to create the world's largest dark pool. This is down to old skool thinking, still popular amongst the ranks of investment banks, which says:

Create the perception that we have the biggest market share of client orders and our competitors' clients will want to trade with us in the hope of finding more natural, opposing liquidity from our other institutional clients. Net result: we make more commission dollars son

The banks with the largest dark pools use it is a major marketing tool to attract more business, talking endlessly about their huge market share to clients. The banks without the largest dark pools try to grow their dark pools. Banks themselves have defined 'size of dark pool' as a standard measure of market share and they are all competing to create the biggest. Correlation is different to causality, but perception trumps reality.

Funny thing is, it worked. Clients believed it because their own, old skool thinking, told them the same thing. In the pre-electronic-trading days, the more internal block trading a broker did, the more likely they were to reduce the market impact of your orders and deliver superior trading performance. Dark pools are the direct, electronic interpretation of block trading. Therefore a larger dark pool must equal superior trading performance. Happy days!

The difference here is that Barclays were not crystal clear with their institutional clients about the sources of liquidity present in LX.

Reason 3: More money

Barclays has their own, profitable, in-house HFT operation.

Barclays is in a vulnerable position

In many areas, banks have focussed too much on competing for the sake of competing and on short-term profitability, without stopping to questioning why. Instead of innovating and creating real customer value, they've reduced value and trust. This has led to some bad decisions. We've seen this before with the LIBOR scandal and the FX fixing scandal.

In this case, Barclays has actively destroyed customer value and broken regulatory standards.

Clients know this. Regulators know this.

The thing to keep remembering though is, amongst all this short-sightedness, there is opportunity. Institutional clients still have huge, trading-related problems and needs that they are desperate to solve and that they are willing to pay for. They are just waiting for someone to deliver a great solution. Recent events have done damage to the loyalty of customers and have opened the door a little bit further for innovative competitors.


Subscribe by email. Click here then click 'Subscribe'

Find me on Twitter and LinkedIn

Dave HunterBuilding a fintech startup. Former head of electronic trading product mgmt and quant strats at DB. Living in London.

comments powered by Disqus