On the 18th November 2015 the New York State Department of Financial Services published the details of the Last Look Consent Order which they had signed with Barclays Bank PLC. The accompanying press release summarised the order as:

NYDFS ANNOUNCES BARCLAYS TO PAY ADDITIONAL $150 MILLION PENALTY, TERMINATE EMPLOYEE FOR AUTOMATED, ELECTRONIC FOREIGN EXCHANGE TRADING MISCONDUCT

This settlement was negotiated in secret. Hence there has been no public defence by Barclays of their use of Last Look. This site corrects that failing. The real facts surrounding this Consent Order have now been scrutinised in a fair legal process as part of a subsequent UK Employment Tribunal case. In that case a large number of disclosures were forced upon Barclays, and hence these documents are now in the public domain. As a result there is the rare opportunity to publicly examine in detail the substance of a financial regulator’s case against a bank. The dissection of the Consent Order below provides insight into how an unwarranted case can be concocted given sufficient motivation.

This case will likely be of interest to the following groups:

• Barclays shareholders
• Barclays employees
• UK taxpayers
• Financial journalists
• Equity analysts
• Foreign financial firms considering locating their US headquarters in New York State
• Financial Regulators
• Lawyers involved in financial regulation

Background to Last Look at Barclays

Last Look is a long standing practice in electronic markets. It provides a small trading advantage to the ‘Market Maker’.¹ Last Look works by holding an incoming order from a client, to buy or sell, for a fraction of a second before deciding whether to fill or reject the order. An order will be rejected if, at the end of the fraction of a second hold period, the market price has moved a specified amount away from the price on the order.

Toxicity is a term used to describe how difficult it is for a Market Maker to make money from an incoming order or orders. The concept is explained here.

The UK Financial Regulator – The Bank of England – specifically endorsed Last Look as being within best practice in its NIPS code.

“Using a ‘last look’ mechanism is within best practice when showing genuine interest”.

The Hong Kong Markets and Treasury Association state in their Code of Conduct:

“A market participant is not obliged to honour a counterparty if an off-market price is hit without reconfirmation”.

The practice is near universal in many electronic markets. 

Legal obligations around Last Look

‘BARX Terms and Conditions‘ is a legal agreement which all Barclays electronic FX clients must sign up to in writing. The Terms state clearly that the issuing of a quote does not guarantee that Barclays will agree to trade at that price. Barclays only agree a trade when they have sent a trade confirmation message back to the client:

“We are not required to act on any Instruction or to execute any transaction pursuant to any Instruction”; “Acknowledgement shall not constitute or be construed as an acceptance of any Instructions by us and shall not be binding on us. In relation to Proprietary BARX services, a transaction shall only be binding when we send confirmation of execution to that effect to the Client”

At the start of 2015 the following Financial Regulators started investigations into the Last Look practice within Barclays Electronic Foreign Exchange (eFX) business:

• US Department of Justice (DoJ)
• New York Attorney General (NYAG)
• Federal Reserve Bank of New York
• UK Financial Conduct Authority (FCA)
• New York Department of Financial Services (NYDFS or DFS)

Following their subpoenas, interviews and often detailed investigations, the first 4 of these regulators took no enforcement action against Barclays.

Response to the DFS Consent Order

Paragraphs 9-16 contain the first 7 specific alleged failings of Barclays use of Last Look . These assert that Barclays applied Last Look indiscriminately to all clients and did not attempt to ascertain the different causes of market moves. It is a remarkable fact that all of these criticisms apply equally to Barclays use of Last Look today. This was revealed in the UK Employment Tribunal using the ‘Last Look Disclosure’ document which Barclays issued to their clients. The importance of this document stems from the fact that Barclays produced it after all the regulatory investigations. No doubt with great care involving their Legal department. It can be considered to represent Barclays’ views and policies about Last Look today.

Paragraph 34 of the Order states: “Pursuant to a July 25, 2014 Memorandum of Understanding, the Department installed a Monitor at Barclays. Following the execution of this Consent Order, the Monitor will work with Barclays solely on remediation plans concerning the conduct referenced both in this Consent Order and in the May 2015 Order, and will terminate its engagement on February 19, 2016.”

What exactly was the ‘Monitor’ imposed by the DFS on Barclays after the Last Look settlement doing? Definitely not ensuring all the alleged failings in the Order were corrected. If the DFS genuinely felt these behaviours were wrong, why did they allow them to continue? It is clear that once the fine had been paid, both the DFS and Barclays were content to return to business as usual. In fact this is the only way in which that business could be conducted.

Paragraph 9 states – “Barclays did not distinguish toxic order flow from instances in which prices merely happened to move in favour of the customer and against Barclays after the customer’s order was entered on Barclay’s systems”.

It is quite simply impossible to determine the reason for any instantaneous market price move when an order is received. There is no information which would allow a Market Maker to make that determination. The market price movements are inherently statistically random events. Even the most generally abusive² counterparty will on occasions obtain a favourable instantaneous price move when they are not behaving abusively. Likewise they will sometimes behave abusively and there will not be an instantaneous price move in their favour.

Most strikingly, Barclays’ present policy shows they still apply Last Look to all clients and are ‘agnostic’ as to the cause of market movements.

Paragraph 11 states – “From 2009 to 2014, a large number of the trades Barclays rejected were not truly examples of latency arbitrage or other toxic order flow.”

This complaint is in part untrue and in part requires the impossible. ‘Toxic order flow’ is simply any order received where the market immediately³ moves sharply against the Market Maker. Last Look can only reject toxic flow as that is how the rule is defined. This criticism misunderstands and conflates the 2 distinct concepts of:

•  Toxic flow / decay – which can be accurately measured for every order once it is received
• Client behaviour or intent e.g. latency arbitrage – which is completely unknowable for any order

When Barclays rejected orders they remained necessarily ignorant as to what the cause of the adverse market move was as described above.

Paragraph 12 states – “Barclays’s (sic) assumed the use of the API/FIX system was itself a clear indication of customer’s ability to engage in latency arbitrage and create toxic flow. Thus, all customers routing orders through such platforms were assigned an undisclosed latency or “hold” time before their trades were executed”.

This is completely incorrect. Barclays made a conscious and open business decision to privilege the service offered via their proprietary BARX GUI system compared to the API/FIX system. This was a public and perfectly justifiable commercial decision. In addition, the BARX GUI system was the only channel Barclays traded through where they knew with certainty that there must be a human being on the other side pressing a button to trade. Therefore it would be much harder for them to carry out certain types of trading behaviour that would create toxic flow. This was not ‘assumed’ it is simply a fact.

It is also a true that the flow received via the FIX/API channel was on average significantly more toxic than that received through the BARX GUI. This was not ‘assumed’ it is a fact.

Any business is entitled to offer different levels of service to different types of client calibrated to those clients own choice of behaviour.

It is a further remarkable fact that still today Barclays do not attempt the discrimination between clients that this paragraph implicitly demands.

“Can I trade electronic spot FX with Barclays without Last Look being applied to me? No. Last Look is applied to all electronic spot FX trading.” (Last Look Disclosure document)

Paragraph 13 states- “Whenever prices within this holding period moved against Barclays and in favour of the customer beyond a certain undisclosed loss threshold, Barclays treated the trade as toxic flow”. This statement just repeats the mistakes of paragraphs 9 & 11 above.

Paragraph 14 states – “Barclays thereby evaluated and applied its Last Look rejection protocols almost entirely in reference to the profit or loss the trade would bring to the Bank.”

This statement is partially correct, but how could this constitute a valid criticism in itself of a commercial activity? The majority of the actions in the eFX system and all other parts of Barclays was in some way directed towards the profitability of the firm as it is in every commercial enterprise. The only purpose of Last Look is to give the Market Maker a slight revenue advantage against the Market Takers (clients of the BARX system). This can partially counterbalance the many advantages given to the Market Taker in the trading relationship.

The attempt to construe this attention to profit and loss as a failing in itself is absurd. This argument implies that the DFS can fine every financial firm for nearly all their business practices.

In any case Barclays did not apply Last Look almost entirely in reference to the profit or loss. They did not apply Last Look anything like as strictly as possible ⁴. They balanced the client experience against their own profit⁵. Barclays chose not to reject a huge number of orders which were instantly loss making as shown in the chart below. The curves represent the average of many thousands of orders accepted by Barclays where they chose not to apply the strictest settings. These orders are highly toxic and could have been rejected but were not.

Paragraph 15 states – “Barclays did not perform an analysis to ensure Last Look was limiting its rejections to trades that in fact reflected “latency arbitrage” or other truly toxic flow”. 

This statement repeats the mistakes of paragraphs 9 & 11 above. The allegation implies it is possible to determine the reason for a given market move – no such analysis exists. Indeed today Barclays still do not limit Last Look to trades that reflect latency arbitrage. They state:

“Does Last Look specifically target only certain types of trading behaviours and flows? No. Last Look is agnostic as to the causes of market movements and will reject a trade request whenever the market moves beyond the price tolerance in place during the prescribed time delay” (Last Look Disclosure document).

So Barclays still contend that Last Look does not target certain types of client behaviours (as indeed it cannot) – it is necessarily agnostic as to the cause of market moves. 

In reality Barclays use of Last Look was highly selective in targeting toxic flow – whatever its cause – as shown in the chart below. The contrast between the toxicity/decay of the orders that were rejected vs. those that were accepted could not be more striking.

Further evidence of how discriminating the use of Last Look was is presented below.

Paragraph 16 states – “Thus instead of employing Last Look as a purely defensive measure, Barclays instead used it as a general filter to reject customer orders that Barclays predicted, based on price movements during the hold period, would be unprofitable to the Bank.”

This statement: “a general filter to reject customer orders that Barclays predicted, based on price movements during the hold period, would be unprofitable to the Bank” is an exact description of Last Look per se. This is always what Last Look was and is throughout the market including:

1. when the Bank of England described it as within best practice
2. today at Barclays in their present use of Last Look 

The creation of this “defensive measure” concept is incoherent and disconnected from the facts. When all clients submit some proportion of toxic orders related to their own choice of trading behaviour (see charts below) and determining the cause of an adverse price movement is impossible, the implementation of a “purely defensive measure” is likewise impossible.

The real facts about Last Look at Barclays

For the DFS’ criticisms to be valid it would be necessary that a Market Maker could distinguish between 2 types of client: toxic and non-toxic. In fact there is a continuum of degrees of toxicity across the clients of BARX. This fact was proven in the UK Employment Tribunal using data which Barclays were compelled to disclose despite fierce resistance. The data showed that in 2010 96% of Barclays clients were toxic to some degree and in 2014 94% were. This proportion is also true within each of the individual client groups: Real Money (95%), Non-financial Corporates (96%), Hedge Funds (95%), Brokers (96%) & Banks (96%). In addition there is a continuum of degrees of toxicity across the clients – they are not separated into distinct toxic and non-toxic groups.

All a Market Maker can do is measure the average toxicity of a client over some previous time period. This average toxicity will change due to changes in client behaviour and purely random influences. Hence any attempt to try to classify clients as either toxic or non-toxic on this continuously varying spectrum of toxicity would be arbitrary, rapidly out of date and create countless further problems in itself. Instead, the way Barclays implemented Last Look automatically gave a very low number of rejects to the least toxic and progressively more rejects to more toxic clients in a graded fashion.

The ‘indiscriminate’ claim

The Consent Order states “WHEREAS Barclays employed its Last Look order holding and rejection protocols broadly⁶ and indiscriminately”. In fact the opposite is true, the use of Last Look was very highly discriminating. Using data covering 5 years from 2010-2015 for all US clients, the proportions of Last Look rejects received by the (on average) more benign client types – Real Money and (non-financial) Corporates were both 0.05% i.e. one in every 2000 rejects came from each of these client types. While the (on average) sharper⁷ client types of Hedge Funds and Brokers received over 97% of the total rejects, despite making up only 34% of the clients. Just the top 10% of clients sorted by no. of rejects, accounted for 98% of all the orders rejected. The majority of clients did not have Last Look applied to their orders at all⁸ and hence received literally no rejects. The pie chart below shows this data.

The reject ratios (fraction of orders received that were rejected) for Real Money and Corporate clients were both less than 0.5% i.e. less than one in 200 orders were rejected.  This is shown in the bar chart below.

This is a highly discriminating treatment by any criteria and certainly disproves the DFS’s claim to the contrary. It was achieved without any explicit discrimination between these categories of client. The very nature of the Last Look rule automatically produced these results which were highly discriminating in both intent and effect. What is more, the discrimination provided by Last Look was laudably fair in that it was strictest on the sharpest clients and most gentle on the most benign clients. This can be contrasted with the discrimination in the bid-offer spreads charged to different clients. Where the sharpest clients got the best prices and the most benign clients (retail clients, small UK businesses), the worst prices – up to 1000 times worse.⁹

Paragraph 17 states – “After the commencement of the Department’s FX trading investigation, Barclays revised Last Look, in September and October 2014, so that its rejection filters would operate symmetrically. That is, rather than reject only those orders that became sufficiently unprofitable to the Bank, trade requests sufficiently unprofitable to customers (but profitable to the Bank) would also be rejected.”

This is an example of  misrepresentation by omission. The statement implies that Barclays cynically tried to change the system once they knew they were being investigated. To achieve this, the DFS have omitted 2 vital facts. First, that the management who made the decision to change the system were not made aware of the DFS investigation in September/October 2014. The written records prove that the first time any indication was given to the relevant staff that there was any investigation was in January 2015.

Secondly, the decision to make Last Look symmetric had in any case been made a full year earlier in January/February 2014. The IT development work to implement the change was carried out through early/mid 2014, Compliance approval was eventually obtained at the end of the summer 2014, which delayed the release date to late September 2014. There is a large paper trail confirming all these events clearly showing they long predate any DFS investigation.

Paragraph 18 states – “In executing these revisions to make Last Look symmetrical, Barclays neglected to update one of its trading platforms, causing 7% of its trading volume to continue under the asymmetrical paradigm until August 2015. Barclays has since updated all trading platforms”

An implication of this statement is that Barclays decided not to update one system in their own interests. This is completely untrue. The work to make Last Look symmetric was specified right from the start to be applied to both IT systems as the contemporaneous documents prove. However an innocent oversight by an IT project manager meant that the work to update the second system was not carried out at the same time as the first. As soon as this oversight was identified in late 2014, work was scheduled to rectify the oversight before there was any knowledge of any future regulatory investigations.

Both paragraphs 17 and 18 of the Consent Order contain the implication of misconduct on the part of Barclays (why else are they included) when the written evidence proves that both implications are baseless.

The general issue of symmetry for Last Look is instructive. No regulator has to date published any requirement that Last Look is applied symmetrically and certainly had not at the time Barclays chose to implement symmetry in their system. The decision was driven by a pro-active internal review looking to improve the (perceived) fairness of business practices – an approach that regulators frequently demand of banks. Instead of getting credit for this however, the DFS has omitted crucial facts, leaving the implication of bad faith on the part of Barclays. Regulatory action like this will deter firms from proactively improving their practices.

Paragraphs 19-22 relate to the application of Last Look to Stop Orders – all these statements are correct and they are wholly justifiable and necessary components of an electronic system executing Stop Loss Orders. The need for this functionality was made brutally clear on 15 January 2015 when a violent market disruption caused by the removal of an intervention by the Swiss National Bank caused the triggering of a large number of stop losses which were filled at incorrect prices causing Barclays to lose many millions of pounds.

If the above functionality had not existed the losses would have been worse. If Barclays had applied the above functionality with tighter settings, a large amount of those losses could have been avoided. Ironically, Barclays have subsequently accused certain employees of not having tight enough risk controls around this event.

In addition it is widely accepted that crudely implemented automatic Stop Loss Orders are responsible for many of the ‘flash crash’ price dislocations which disrupt markets. These ‘smart’ Stop Loss Orders implemented by Barclays protect against this type of market disruption.

Paragraph 24 makes the criticism: “Barclays did not disclose the reasons for not accepting trades in its post-trade reporting to clients. Instead, the client would receive a simple rejection message: “NACK” which stood for “Not Acknowledged””

However the very first Terms of Service Legal documents signed by clients, state in the section concerning order rejects:

“We are not obliged to act on any instruction, or to execute or otherwise enter into any particular transaction, and need not give any reasons for declining to do so”.

When an order is rejected due to Last Look (or any other reason), the counterparty is sent a reject message within in a fraction of a second notifying them that the order is not filled. It is immediate and clear.

Paragraph 25e states – “Barclays employees suggested explaining to the ECN “what last look is”.”

It is totally implausible that any employee of an ECN who had serviced clients and dealt with enquiries about order rejects wouldn’t know what Last Look was, given its wide use throughout those markets for many years. Emails exist from at least one ECN (Currenex) showing that they certainly did know about and understand Last Look.

In relation to paragraph 25 generally – all 7 quoted communications appear to have come from support staff who were certainly not implementing any instructions from the management of the business. However it should be noted that when for example the Head of Sales stated “I recommend we implement LL symmetry ASAP and draw-up a client communication plan”, the firm’s lawyers came back stating: “I have now heard back from external counsel and I share their view that unless the changes [to Last Look] represent a change to the terms and conditions with customers, they see no need, from a legal perspective, to reach out to customers to inform them of the changes”. Following this advice, there was no communication with clients about this, the largest change ever made to Last Look.

This handful of communications constitutes the only valid and very limited basis for criticism of Barclays in the Consent Order. On these few emails, rests the slender justification for a fine of $150 million. These few emails are a tiny, very carefully cherry picked and unrepresentative sample from the 1.2 million communications related to Last Look that were scrutinised in Barclays’ investigation. There are many, many times this number of examples of Sales and support staff discussing Last Look openly with clients.

On 16 November 2015 Barclays own lawyers stated in an internal recommendation :

“(b) There are many examples of open and transparent communications with BARX users about rejects and Last Look, as well as examples of clients negotiating or seeking to negotiate their Last Look settings – suggesting a high degree of awareness amongst customers of Last Look and its relevance to their order flow

(c) Sales were generally aware of Last Look

(d) Last Look has been commonly understood in the market for many years and market participants were well aware that Barclays, and other banks, had a discretion to reject trade requests using Last Look”

These statements so starkly contradict the statements in the Consent Order which Barclays had just signed, that it is hard not to conclude that they knew the Order was not fair but were simply compelled to sign it.

Paragraph 27 states – “In a marketing presentation for the BARX GUI circulated to a variety of clients, Barclays stated: “No last look – what you see is what you get”. This presentation did not explicitly distinguish GUI from FX/API platform, nor did it explain that Last Look did apply to FIX/API clients”.

This is a quite absurd objection. It is equivalent to complaining that the instructions for a toaster didn’t explicitly distinguish it from a kettle. Nor did it explain that a kettle could contain boiling water. The quote makes clear the material was for the ‘BARX GUI’ (the toaster) and those assertions were correct for that product.

Meanwhile in the documentation for the FIX API (the kettle) it is made quite clear that the product is not what you see is what you get:

“All orders are executed as Fill or Kill. They will either be traded outright or rejected” – in the ‘BARX FX via FIX Rules of Engagement’.

So it is quite explicit that Barclays killed/rejected orders over the FIX channel. There is even a detailed section on how Barclays will notify FIX API clients of order rejections: the ExecType field in the Execution Report message sent back immediately to the client is stated to have just 2 possible values: “Filled” or “Rejected”.

Paragraph 28 “Certain senior Barclays employees instructed traders and IT employees not to inform Barclays Sales team about the existence of Last Look”

‘… avoid mentioning the existence of the whole BATS Last Look functionality’.

This is a plain attempt to misrepresent the wholly standard principle of preserving the confidentiality of algorithmic details. It is a specific requirement of Barclays Global Supervision Policy that information should be kept “on a need to know basis” and that “information barriers” are maintained. BATS is a code base containing most of the algorithms controlling Barclays electronic FX system. None of these, Last Look related or otherwise, were ever discussed outside a limited group of quantitative analysts and traders. Indeed there were quite standard, specific permissioning barriers to prevent anyone other than these few people viewing the details within BATS. How much are Google’s salespeople told about the implementation details of their websearch algorithms? I’ve checked, and the answer is practically nothing.

Remarkably, Barclays own lawyers don’t believe that this conduct was inappropriate. Another revealing document that came out of the Employment Tribunal case was the recommendations by Barclays Legal team to start a disciplinary process following the Consent Order. It states: 

“The clients affected by rejections were overwhelmingly the most sophisticated BARX counterparties and, therefore, the group most likely to take advantage of any information about Barclays’ Last Look settings to the detriment of Barclays (and BARX pricing). With that in mind, it may have been appropriate to withhold information in order to maintain the integrity of Last Look as a legitimate defensive tool for Barclays.” 

This communication also shows that Barclays did not believe that the use of Last Look was ‘indiscriminate’ as claimed in the Consent Order.

To try to twist the above statement to imply an attempt to hide the whole Last Look practice itself, is wholly unjustified. There is distinct sleight of hand in this attempt to conflate confidentiality about BATS Last Look functionality with the existence of Last Look. No-one receiving this email would have understood it to mean ‘deny the existence of Last Look’¹⁰ given that it was a constant topic of conversation between Sales and Trading and Sales and Clients and that Barclays had developed tools with the sole purpose of informing Sales about the detailed statistics of Last Look rejections.

There are numerous emails showing the author of this alleged attempt to hide the existence of Last Look, talking openly with Sales staff about Last Look. Twenty two of these emails were presented in evidence at the Employment Tribunal. There is the written record of a meeting organised by him with 2 senior Sales managers with an agenda item “Last Look / Rejects”. The same author also directed his report in writing to create a ‘Last Look role’, one of whose duties was to:

“Try and lead the education of sales about Last Look”.

Again, to his reports he wrote:

“Please provide any client who enquires for info on their top level Last Look stats (trades, volumes, %rejected trades) with that info. Obviously the client could calculate these stats themselves from the orders and trades they see but it may be an effort for them to do this so providing them is helpful”.

The above evidence is the tip of the iceberg of open communications about Last Look that put the lie to the charge in the Consent Order that: “WHEREAS there was a lack of transparency both internally and with customers regarding Last Look”. What is more, Barclays lawyers knew this and stated in their response to the subpoena from the NYAG: 

“The rationale behind Last Look was and is widely known throughout the electronic Trading, Sales, Quantitative Analytics and Technology Teams within Barclays” 

Paragraph 28c “On Nov 7 2011, the Barclays Managing Director and Head of Automated Electronic FX Trading wrote: “Do not discuss Last Look with Sales. If there has been a spurt [in rejected trades] just blame it on the weekend IT release and say it is being fixed”.

This is a clear attempt to suggest this manager has misrepresented the truth. The impression is deliberately created that to ‘blame it (the spurt) on the weekend IT release’ is untrue. However the statement is entirely accurate. This email was sent on a Monday morning after an IT release had been put out over the weekend. That IT release had caused a spurt in rejects which was wholly unintentional and was indeed corrected immediately. There are numerous emails which prove the deliberate efforts made prior to this IT release to ensure that the release caused no change in the number of rejected orders:

• From the analyst making the changes: “We are making changes to the delays in Last Look rules… whilst keeping the client experience constant”.
• Business manager to an IT manager: “The reason why we’re changing the bps is precisely to preserve the reject ratios. They have to go in together.”
• Business manager to an analyst: “No-one should see a change in their confirm times with this release.”
• Analyst to IT staff: “we’ve now revised our thinking in order to maintain constant levels of rejection ratio”
• After some clients experienced an increase in rejects (the spurt) a manager wrote: “Didn’t we widen their BP thresholds to stop this happening”

The sentence immediately preceding this one quoted by the DFS (but deliberately not included) makes it completely clear that this spurt was accidental and should be corrected immediately: “What spurt in rejects? There shouldn’t have been a spurt in rejects. Just recalibrate the thresholds…if that has happened”

There is a definite intent by the DFS to create in the mind of the reader that a manager urged staff to misattribute rejects in general to an IT failure. However this is plainly not what the email says. It states correctly that the spurt in rejects was due to the IT release. 

The same manager who wrote this alleged attempt to misattribute rejects to IT failures wrote in another email to the Head of Electronic Sales: “We’re getting the vague sense from some people that they think LL ¹¹rejects are some kind of failure of the machine when in fact they’re often all that’s standing between us and loss-making client flow.” This is an active attempt to make sure Sales did not think Last Look rejects were due to IT problems.

Paragraph 29 – the DFS conclude under the title – “VIOLATIONS OF LAW AND REGULATIONS” that the specific law/rule/regulation which Barclays have allegedly broken is: “With regards to the aforementioned conduct, the Bank has conducted banking business in an unsafe and unsound manner.“

It would be hard to imagine a more vague and catch all offence with which to charge anyone. This should be contrasted with regulators such as the US Dept. of Justice or the New York Attorney General who have to actually specify a particular law which has been broken and how, when bringing a case.

 

Paragraph 31 – “A Barclays Managing Director and Global Head of electronic Fixed Income, Currencies and Commodities (“eFICC”) Automated Flow Trading has been suspended but remains employed by the Bank. The Department orders the Bank to take all steps necessary to terminate this individual, who played a role in the misconduct discussed in this Consent Order.”

This clause blatantly violates an employee’s right to anonymity under New York Banking Law Section 41 . Which states:

“Such order and the findings of fact upon which it is based shall not be made public or disclosed to anyone except the director”.

In addition it violates Barclays cited principle of not identifying details of employees’ disciplinary processes.

Moreover this ‘Termination Order’ violates the protections offered to employees of firms under New York Law. In short, DFS does not have plenary power to remove a company’s employee. Nor may DFS violate an individual’s due process rights when removing a person. Instead, DFS’ authority to remove employees is narrowly tailored by statute to provide procedural protections for persons whom DFS attempts to remove. In this case, the DFS made an unlawful end run around these statutory limitations and procedural protections by ordering Barclays to terminate the employee. The details of this flagrant evasion of statutory obligations by the DFS can be found here.

Barclays followed this instruction by concocting and upholding a large number of allegations against this employee who they subsequently dismissed. The employee then sued Barclays for unfair dismissal under UK Employment Law. As part of that case, the Tribunal forced them to disclose a large number of documents which they had refused to disclose as part of their internal disciplinary process and as part of the Employment Tribunal process. Many of these documents proved very damaging to Barclays’ defence as they showed the true nature of Barclays use of Last Look and its acceptance amongst senior management, the Sales, Compliance and Legal departments. Barclays were found guilty of unfair dismissal (Liability Judgement). The dismissal was found to be both procedurally and substantively unfair in both the original disciplinary process and the appeal process ¹². At a subsequent remedy hearing, despite fierce resistance, the Tribunal ordered Barclays to re-employ the dismissed manager, albeit in a different role. Barclays then simply refused to comply with this lawful order including the order to return 21 months of arrears of pay and benefits. A subsequent order from the Tribunal forced them to pay these arrears.

1. the party providing the quotes against which other parties (clients) may trade.
2. sharp, toxic
3. within a fraction of a second
4. The thresholds for rejecting orders were generous to clients. An order had to be loss-making within a small fraction of a second before it would be rejected. Most of the Last Look thresholds were set at wide values, meaning Barclays would accept orders where their entire fee (bid-offer spread) had been retracted plus some further loss amount e.g. £100.
5. Barclays lawyers in their response to a subpoena from the NYAG wrote:

   “Barclays typically determines a particular client’s rejection threshold based on either (a) Barclays’ negotiation with a client regarding pricing and reject ratios, including adjusting thresholds to meet a particular client’s request for a maximum reject ratio, or (b) Barclays’ unilateral adjustments to Last Look settings and spreads to try to find a balance acceptable to both the client and Barclays“

6. the broadly claim is true, but as shown above it is still true today
7. more toxic – trades tended to immediately revalue in the clients favour more strongly
8. nearly all clients using the BARX GUI channel had no Last Look
9. these huge spreads were defended in the Employment Tribunal by John Mahon  who had run the Corporate Banking division which implemented many of them. He claimed that the use of Last Look did not put “client’s interests at the forefront“, but apparently charging a small UK company or Mrs Jones in Swindon 1000 times as much as a Hedge Fund for the same service was fair.
10. This would be like saying “don’t tell anyone in Sales that they work in a bank”
11. ‘LL’ is the widely used abbreviation for ‘Last Look’
12. The “Discussion and Decision” section of the judgement uses the words unfair/unfairness/unfairly 14 times