Goldman has already mapped 146 “begging to be automated" Analyst IPO Steps

RGE's picture
Rank: Orangutan | 343

This via Marty Chavez, Goldman Sachs's incoming CFO. What are your thoughts?

TechnologyReview.com: At its height back in 2000, the U.S. cash equities trading desk at Goldman Sachs's New York headquarters employed 600 traders, buying and selling stock on the orders of the investment bank's large clients. Today there are just two equity traders left.

Automated trading programs have taken over the rest of the work, supported by 200 computer engineers. Marty Chavez, the company's deputy chief financial officer and former chief information officer, explained all this to attendees at a symposium on computing's impact on economic activity held by Harvard's Institute for Applied Computational Science last month.

The experience of its New York traders is just one early example of a transformation of Goldman Sachs, and increasingly other Wall Street firms, that began with the rise in computerized trading, but has accelerated over the past five years, moving into more fields of finance that humans once dominated. Chavez, who will become chief financial officer in April, says areas of trading like currencies and even parts of business lines like investment banking are moving in the same automated direction that equities have already traveled.

Full Article from technology review

Initially covered by Zero Hedge: http://www.zerohedge.com/news/2017-02-13/goldman-h...

Comments (30)

Best Response
Feb 14, 2017

Computer:
"Hello Mr. RGE, how is your day today? I have gone ahead and compiled a list of acceptable acquisition candidates with varying risks and return for your approval. Please let me know how you'd like to proceed."

IB:
"Where are my analysts? Where are my interns?"

Computer:
"They have been assimilated and discarded."

IB:
"Who's going to get my coffee? Who am I going to go to strip clubs with? Who am I going to talk to and bullsh*t with?"

Computer:
"It sounds like your needs are compromising your ability to be a reliable and efficient asset at this company. Prepare for assimilation."

IB:
"To Hell With You!"

Computer:
"We are Legion."

    • 18
Feb 14, 2017

Computers do not analyze data, they process data. They process what people put in their algorithm, and then a person makes a decision on that information, or another algorithm that has been prepared by a person makes a decision on that information.

Banking already has the leanest deal teams possible, they could not get any leaner unless you were willing to trust an algorithm 100% on every possible task a person would do. As it stands, there is one Analyst today, and there will be one in the future whose job will get slightly easier as it relates to data processing versus data analysis.

Exhibit 2. Real talk, the job of "Analyst" is not going anywhere.

"Hey, Siri, what do you make of the senior management team? Did that CEO sound guarded when answering the question on actioned cost savings?"

"Hey, Siri, can you do (some complicated formatting exercise that would take an hour to explain but could be drawn on paper and explained in context more easily)"

"Hey, Siri, could you take notes on this management presentation dial-in? By the way, the microphone will be across the room, the speaker is from Kazakhstan, and also the microphone is actually a tin can."

    • 7
    • 2
Feb 14, 2017

I agree with you that banking has some of the leanest deal teams around and that it would take machine some time to be able to analyse data up to analyst's level.

However, the comment about job "not going nowhere" is not quite correct because if machines/computers make the job easier then the demand/need for the analyst actually decrease by some margin. Say you are a MD running a good & lean team and you have six analysts working under you now, but in 5-10 years time will you need 6 analysts working for you as machine will do more of the work and their working hours are less? Probably not, you'll only need 4/5 analysts instead of 6.

If you only need 5 instead of 6 that is a decrease of ~17% and if you need 4 instead of 6 that is a decrease of ~34%. My point is the demand for analysts are bound to come down by quite a margin.

    • 1
Feb 14, 2017

The gap in understanding is how banking deal teams are set up. There are not six analysts on one deal and therefore increased productivity could decrease that number to four or five. There are e.g. three deals, each analyst on two active ongoing deals. Even with computerization, a given analyst simply cannot process three+ deals going on at the same time, you will hit a wall with the analyst's capacity. There is just too much going on, and the analyst (the person) needs to be able to know all of the numbers and the qualitative information - strategy, process, timeline, management, internal politics - for each deal. It is just impossible without a fundamental rewrite of what a banking deal is and how decisions are made.

Let's unpack that last statement and say, somehow, banking is completely changed. Let's say, for example, FutureBankSachs & Co has MD's and a set of algorithms that take all financial information, put it into a model, write up key information, etc. Now, what, the MD has to process 30 deals him/herself based on the input from the algorithm? How can the MD have calls with all of the parties involved? Can't do that, so let's add in 5 VP's below the MD to manage the process for 6 deals each. Okay, the VP's take in the model outputs and read the computerized write-up, and this takes all of their time to synthesize this information and run the process. They run out of time to provide accurate service for each of the deal. They need Associates below them. Banking Associates come primarily from MBA programs, have less of a quantitative background. Somehow let's change that and have them all be computer engineers from MIT. How do these Quantsocciates somehow become banking VPs, Directors, and MD's? The human capital pipeline breaks down somewhere and the system just doesn't make sense.

    • 10
Feb 15, 2017

At least GS is getting to the party... the other banks have a ways to go. GS is far ahead in terms of technology and automation of basic tasks.

As a wild-ass guess, I am going to throw out there that at least 50% of current analyst work will be automated in 10 years. You need people for relationship management, managing a process, communication, customization of outputs where appropriate and all of the other idiosyncratic non-automatable tasks.

Making DCFs on public cos with CapIQ data is an anachronism. So is much of the glorified investment banking analyst role.

    • 1
Learn More

7,548 questions across 469 investment banks. The WSO Investment Banking Interview Prep Course has everything you'll ever need to start your career on Wall Street. Technical, Behavioral and Networking Courses + 2 Bonus Modules. Learn more.

Feb 15, 2017

I'm not sure man. unless reporting standards change significantly you can't just rely on a Mergermarket/factset download to do your stuff.

Feb 15, 2017

For most illustrative / preliminary pitches you can. Only when a deal becomes live do you really need to scrub the numbers. Plenty of banks just use FactSet / CapIQ comps for pitches. It's more efficient

Feb 15, 2017

Natural language processing is getting A LOT better. Siri is just the shitty public version that Apple debuted. Closely guarded NLP algorithms are fantastic at detecting nuance in speech, text, etc. I absolutely see this happening.

    • 1
Feb 15, 2017
DeepLearning:

Natural language processing is getting A LOT better. Siri is just the shitty public version that Apple debuted. Closely guarded NLP algorithms are fantastic at detecting nuance in speech, text, etc. I absolutely see this happening.

Agreed. Equating true machine learning/AI with Siri is a joke.

Feb 17, 2017

bingo. I already know some top firms use it to analyze earnings calls for tone/inflection

Feb 15, 2017

They're using the audio? Damn that's cool... Lately I've been working on some stuff with NLP and EDGAR filings but would be incredible to work on the actual audio of the calls.

Learn More

7,548 questions across 469 investment banks. The WSO Investment Banking Interview Prep Course has everything you'll ever need to start your career on Wall Street. Technical, Behavioral and Networking Courses + 2 Bonus Modules. Learn more.

Feb 15, 2017

If I had a nickel for every time I've caught a mistake in CapIQ data, I'd have $2.15. That doesn't sound like a lot, but it's 43 times.

    • 3
Feb 15, 2017
Ash Ketchum:

If I had a nickel for every time I've caught a mistake in CapIQ data, I'd have $2.15. That doesn't sound like a lot, but it's 43 times.

I don't know if they still do this, but SNL used to give you $50 for every error you could find. Didn't happen super often, but enough to buy us lunch every once in a while.

Feb 16, 2017

they still do as far as I know. does that apply to gramma mistakes as well?

Feb 16, 2017
spreadsheetmonkey204:

Ash Ketchum:If I had a nickel for every time I've caught a mistake in CapIQ data, I'd have $2.15. That doesn't sound like a lot, but it's 43 times.

I don't know if they still do this, but SNL used to give you $50 for every error you could find. Didn't happen super often, but enough to buy us lunch every once in a while.

A coworker of mine had caught one before. They said they'd send $50 but she never got it lol. She couldn't care less to follow up.

Feb 15, 2017

Good riddance to a lot of the stuff getting automated. It frees up analyst time for models, deal execution, client interaction, and other work that will actually benefit their clients and employers, and advance their careers. Working group lists - and a half dozen of the workflows that they govern - are automated now. Is there someone who really wants to spend a zillion hours word processing one of those things, to offset some tiny risk that next year's analyst class will be a little smaller?

Feb 15, 2017

Churning out the majority of preliminary pitch books can pretty much be automated already. Everyone has their pack of 100+ 'template' slides with a reminder to update the charts for a new ticker/date and do a find/replace on [company]. 50+ page 'introductory' books that probably took a few hours/days to do in the 90s now take less than 15 minutes, hell my Director can almost make them himself now!

Agree with all of the above on the analysis / human factor stuff.

    • 1
Feb 15, 2017

Clearly an exaggeration... we all know directors can't make books. Nice try.

Feb 15, 2017

I had a talk with a professor who use to work at Goldman, in 2016 35% of employees at Goldman Sachs worked in technology/IT/Computers, and the number will increased to 40% in the coming future. Goldman has more tech workers than many Silicon Valley firms. Really curious to see how things continue in the future.

Overall this trend seems like its bad for finance workers, but overall, great for clients, the economy, and the small and even institutional investor.

Feb 15, 2017

When I did a super day for a tech SA at goldman, they kept repeating over and over that goldman is a technology company disguised as a bank. maybe they weren't lying like I thought they were lol.

Feb 16, 2017

*goldman is a giant squid disguised as a technology company disguised as a bank

Feb 15, 2017

Still lying lol.

Feb 16, 2017

I swear to god if see one more thread about automation... I'll find you OP and I'll make you do your analyst stint one more time.

You killed the Greece spread goes up, spread goes down, from Wall Street they all play like a freak, Goldman Sachs 'o beat.

    • 2
    • 1
Feb 17, 2017

I think the most depressing thing here is that the prevailing idea across the whole financial industry is that we will be the last to go because we have so many "soft skills" and the experience that you can't program....That's bull$hit. In the end formatting, modeling, forecasting, etc. will all be programmed in the not too distant future. I'm not even going to attempt to say when, but if you are like me and looking at another 30+ years of working before retirement it's not looking good. If you think it's the soft skills that are going to save you I have two words "Renaissance Technology". Learn coding now and we might have another five years after the judgement day described above. It will all eventually come down to data and returns....not to mention you don't have to pay software a bonus.

I was reading something the other day making an argument that the last jobs to probably go will be the craftsmen and construction workers whose experience is cultivated from years on the job and the trial & error tacit knowledge that has been passed down for generations. So maybe it's time we all embraced our inner hipster, grow a mighty beard, buy some albums on vinyl, and trade in those Ferragamo's for a pair of Red Wing's & vintage carpenter jeans. Of course the counter to that is, have you noticed the recent success of infrastructure funds?

    • 3
    • 1
Feb 15, 2017

There's a distinction to be drawn between what is happening in investment banking (M&A + capital markets) and what's happening in sales and trading. Matching willing buyers and sellers is obviously something that we don't need people to do. Nor, arguably, is spotting arb opportunities, mispriced securities, etc. Hence the ghost town trading floors.

But anyone who has worked in coverage knows that relationships are everything. There are freakishly smart, incredibly hard working and extremely technical industry bankers who never make it beyond VP because they lack people skills. If a computer-like person can't make it (and most of them can't), an actual computer definitely won't. Automation in IB will be far more incremental, and mostly focused on the things bankers shouldn't and don't want to do anyway.

But who cares? Who's to say that won't free them up to pitch more clients and execute more deals, i.e. be more profitable? It won't necessarily occasion a reduction in headcount.

Feb 17, 2017

hell everything is gonna b computerized tbh this is why i have spent yrs learning how to code potatoes

What concert costs 45 cents? 50 Cent feat. Nickelback.

Feb 17, 2017
Comment
    • 1
Apr 28, 2018
Comment