Morning Coffee: How to perform badly and get promoted past your 50th birthday in banking. And Two Sigma's solution to the black box problem

eFC logo

So many news stories have been published about Garth Ritchie’s imminent departure from Deutsche Bank that it’s surprisingly easy to forget that the 50 year-old head of Deutsche’s investment bank is still actually there.  This fact appears to have upset quite a lot of people; apparently over a dozen “people familiar with the bank” spoke to Bloomberg for a fairly hostile profile of Ritchie’s rise through the ranks, even as the franchise itself fell apart.

Some of the anonymous comments made are pretty unfair – it’s true that the trading revenue has fallen under his leadership, but that’s mainly a consequence of Deutsche shutting so many product desks down.  Some of them are more on target – the Russian equities operation reported to him throughout the mirror-trade scandal and so did Deutsche’s prime brokerage when it was lending to hedge funds involved in cum-ex trades.  Taken together, however, we get a really interesting portrait of how a set of opportunities came along, and of the personal qualities that allowed Ritchie to grab them.  It’s a story we’ve seen at other times and other banks, including Goldman Sachs, which suggests there are some general career lessons about the industry to be found here.

There’s clearly a substantial element of “right place, right time”.  Ritchie joined the bank in the 1990s and was head of European equities by 2005, overseeing the bank’s notorious feast-or-famine approach to the less important and profitable of its two trading lines.  Coming up through the equities side, though, meant that by the time John Cryan arrived, he wasn’t tainted by the crisis-era FICC trading losses, or by the (legal) personal account trades that some of his management colleagues had got involved in.

From then on in, it seems to have been a matter of taking on difficult problems and dealing with bad news in a straightforward and pragmatic manner, a quality that’s much more rare in investment banking than it ought to be.  Ritchie has overseen large staff cuts without losing the confidence of the employees, and dealt with regulators and clients in a way that impressed Paul Achleitner sufficiently to bring him on to the management board.  The economist JK Galbraith said that the essence of a leader was “the willingness to confront unequivocally the major anxiety of their people in their time” and this seems to have described Ritchie’s strengths.  Added to which the simple fact of having been around for more than a couple of years was itself a benefit; he was a rare fixed point in a bank with a serious turnover problem.

But … nothing lasts forever.  The cum-ex and mirror trades scandals were not legacies of the past; they happened on Ritchie’s watch.  And given Deutsche’s struggles to build a profitable banking franchise, it’s understandable that shareholders looked unfavourably on his being given a new five year contract (particularly when a material chunk of his compensation was structured as a role-based allowance for Brexit preparations).  And the very biggest problem at Deutsche – the inadequate and confusing systems and controls – is the one area that Ritchie doesn’t seem to have tackled.  The official word is now that he plans to remain in the job for as long as Christian Sewing wants him, but how long can that be?

Elsewhere, TwoSigma is addressing one of the biggest problems in quant finance and AI– namely that if a trading system has been developed by a machine learning system, how do you explain what it’s doing to the investors?  It’s a specific instance of a general problem called “interpretability”, seen in cases like that of Amazon’s sexist algorithm, or the neural net that was meant to identify tanks but actually identified sunny days because of the photos it was trained on.

In investment terms, there appear to be two main ways of “interpreting” a complex model.  You can either systematically peturb all the inputs and get a more or less sophisticated measure of how much the output changes.  Or you can build a “model of the model”, using a more easily interpretable method to forecast what answer the algorithm is going to give.  There’s no single best solution, according to TwoSigma’s Xiang Zhou – it depends on some technical factors, like whether you’re trying to understand the whole logic of the model or just a single decision.  But more importantly, the approach you choose to explain a machine learning model has to be adapted to “the user’s expertise level” – there’s no point in choosing the theoretical best method if it needs to be explained to the trustees of a local government pension fund.


More turnover from Deutsche Bank.  After less than two years in the job and having been offered a role managing the bad bank, the global head of equities has apparently decided he would rather leave; there is no timescale given and the decision may not be final, but it looks like cuts are coming to the equities franchise. (WSJ)

A first person account from Goldman Sachs MD and communications chief Maeve Duvally, describing what it felt like to come out as a transgender woman (PR Week)

The insider dealing trial of a former UBS compliance officer and a day trader has returned two guilty verdicts.  It seems that the jury didn’t accept their not-very-plausible sounding explanations for why the day trader bought his friend a burner phone to get text messages on (Bloomberg)

Are human resources the “friends of management” or the “friends of the employees”?  In a fascinating interview, the Chief Human Resources Officer of mobile phone company O2 suggests that “if you start deciding which side of the line you sit on you’re in big trouble”.  Which means … probably management, we think. (HR Magazine)

Private equity companies (or at least, those who attend conferences and respond to survey questions) are still planning to hire strongly (Institutional Investor)

Millennial employees are speaking out against their employers’ more controversial practices, leading to Deloitte and McKinsey among others having to pull back from some clients (WSJ)

Don’t give too much information and be honest about how much you’re going to be checking email.  Guidelines for the perfect out-of-office message (The Cut)

And how horrible is your financial centre going to be when the heatwave begins? They’re already running naked through the freezer section of the supermarkets in Germany. (Bloomberg)

Have a confidential story, tip, or comment you’d like to share? Contact: in the first instance. Whatsapp/Signal/Telegram also available.

Bear with us if you leave a comment at the bottom of this article: all our comments are moderated by human beings. Sometimes these humans might be asleep, or away from their desks, so it may take a while for your comment to appear. Eventually it will – unless it’s offensive or libelous (in which case it won’t.)

Related articles