The decline of Chicago prop trading

It's sad to see how far Chicago prop trading has declined in recent years. Not too long ago, I envied my classmates and friends who were working at the top Chicago prop shops and making more money than me, while dealing with less bullshit and better hours. Now they hate their jobs, gloomy about the future prospects of the industry and asking me for advice on how to get into top MBA programs.

Will things better for the chicago prop shops? Or is the era of sick bonuses for twentysomethings now over?

 
  1. First, strong post wording to get onto the front page
  2. Yes/No - Depends on volume/volatility. Naturally as finance progress it becomes more liquid and transparent hence you always hear "I USE to make tons of money etc etc" - in other words, hot products/niches come and go. Algorithm trading has come and is on it's way out. There is still a use for it, without a doubt it will move into more complex products (FICC here it comes!), but the big money era is over. If anything decline in prop revenues will increase the push into FICC - especially since the OTC is moving onto exchanges

Whoever doesn't accept the above is either:

  1. A prop trader
  2. A prop trader
 

I agree. HFT for the most part has been nothing more than a glorified speed game, with firms fighting for that tiny edge in low-latency trading and getting as close to the servers as possible. In my opinion, that's not "real" finance: just glorified IT. When I hear my friends at top HFT shops talk about their work, it's mind numbingly boring. Even they're bored and are trying to get out of the business altogether, and these are hardcore CS geeks!

 

Honestly,FICC Algo Trading is really just “glorified IT”. Or in other words, a bunch of developers who automate processes. Though if you are a developer, tech companies would be a better choice.

 

Oh, they used to be all the rage. During my senior year at a top target, I went to the on-campus info sessions for getco and jump, and the room was packed with the top math/cs/applied math/physics students at my school. The companies did a great job of selling themselves, convincing some of my smartest classmates that they will be working on "cutting edge" problems across finance and tech. Off the top of my head, i know 1 guy who turned down de shaw for getco, 1 who turned down google for jump, and another who turned down goldman S&T for DRW.

 
mbavsmfin:

Oh, they used to be all the rage. During my senior year at a top target, I went to the on-campus info sessions for getco and jump, and the room was packed with the top math/cs/applied math/physics students at my school. The companies did a great job of selling themselves, convincing some of my smartest classmates that they will be working on "cutting edge" problems across finance and tech. Off the top of my head, i know 1 guy who turned down de shaw for getco, 1 who turned down google for jump, and another who turned down goldman S&T for DRW.

I can't say Getco recruited at my school but SIG and JSC certainly did. Going to JSC would've been a dream... but now I think I'm blacklisted there as well as SIG since I get the feeling that if you don't get hired the first time you get an interview they don't consider you again. I didn't know a single person who got a full time interview from SIG or JSC if they interviewed for an internship the previous year and didn't get it.

I think Jeff Yass was really pissed at any employee leaving the firm because that would mean more competition. I think they had a non-compete when I applied.

 

JSC is the one prop firm that continues to totally dominate. I have pretty good sources that they basically never have a losing day. Not sure how they do it, but what they've done is amazing.

Jeff Yass is a BOSS.

 

To build on my previous points, can someone explain what the following is not feasible?

  1. Centralized markets for all financial products

2. The ability for PM's to click a button on a platform provided by a third party to purchase/sell products without going through a dealer

Look 20 years back, and now view 20 years forward - the question is, when will it occur?

edit: another thing ^^^^^ to the above prop trader - there will always be prop trading, but my position is that there is more competition = less spread/less opportunity to turn a profit. There will always be room for HFT, but again, it was a hot, niche product and it has had it's run. The business models are changing

 
mbavsmfin:

JSC is the one prop firm that continues to totally dominate. I have pretty good sources that they basically never have a losing day. Not sure how they do it, but what they've done is amazing.

Jeff Yass is a BOSS.

Then why did they have their first (ever, I think) wave of layoffs this year? I hear they're not doing so hot.

 
kingoftheotherroad:

To build on my previous points, can someone explain what the following is not feasible?
--
1. Centralized markets for all financial products
2. The ability for PM's to click a button on a platform provided by a third party to purchase/sell products without going through a dealer
-
Look 20 years back, and now view 20 years forward - the question is, when will it occur?

edit: another thing ^^^^^ to the above prop trader - there will always be prop trading, but my position is that there is more competition = less spread/less opportunity to turn a profit. There will always be room for HFT, but again, it was a hot, niche product and it has had it's run. The business models are changing

I've always viewed it like this:

People who want to go into prop trading to make half a mill before they turn 25 are disillusioned now, and they always have been. In the end, it's still a very solid industry to be in if you're talented because as OP said, there's less bullshit, and the pay is still very good, especially for Chicago low COL (since that's specifically what we're talking about here).

100-200k in Chicago is a TON of money for someone in his 20s. But maybe I just have lower standards/less self-respect than everyone on here who seems to want absurd riches by the time they're 25 :P

 

I agree that $100-200K/year is good money for a single guy in Chicago, as do my friends in prop trading. Their main complaint is that the work isn't that interesting, and you don't develop too many transferable skillsets. It is for these reasons that they're trying to get into top b-schools or switch into tech/buyside, etc.

 
mbavsmfin:

I agree that $100-200K/year is good money for a single guy in Chicago, as do my friends in prop trading. Their main complaint is that the work isn't that interesting, and you don't develop too many transferable skillsets. It is for these reasons that they're trying to get into top b-schools or switch into tech/buyside, etc.

I know two people at two of the three firms you mentioned that liked their work and thought it was interesting. What are the people that you know doing on a high level? I can't speak to what one of them is doing since he says it's pretty secret (but involves creating an algo ofc) and the other I haven't talked to in a while but said he liked the work last time we spoke.

 

Even within finance, there are SO many more interesting stuff out there. For example, investment associate at bridgewater, trading with google's cash reserves, commodities hedging for a large corporate such as Pepsi, VC branch of a large tech firm such as Intel, distressed debt and special stituations investing for a large hedge fund, etc.

 

I've been a prop trader for 7 years (woah). I trade at one of the few places that somehow operates on the old Chicago model - firm gives you money, do what you want as long as it makes money, split profits. The landscape has definitely changed. The guys that are still here have changed with it. Guys/firms just trading off order flow or some other gimic (nothing wrong with gimics, I know people that have made millions on stupid shit) have mostly washed out. The way I trade now is absolutely nothing like I traded at the beginning of my career.

Prop trading has afforded me a pretty sweet lifestyle throughout my 20s. I'm financially secure - no debt and a net worth in the top 5% of people my age. Even with the changing trading landscape, I feel pretty confident that I can continue to make good money going forward.

That said, I wouldn't recommend the prop trading industry to most people right now. The odds of success are extremely low, earnings potential is anyone's guess, and job stability is pretty much non-existent.

 
spoonfork:
he odds of success are extremely low, earnings potential is anyone's guess, and job stability is pretty much non-existent.
Whats changed so much and why is the probability of success lower now?
 
monkeyattack:

kicking your trading bretherns while they are down as you exit for greener pastures Brady? cold blooded haha

Lol, just seeing the thread title I thought "what, brady is back?"

 
kingoftheotherroad:

Algorithm trading has come and is on it's way out.

How do you even remotely justify this?

Some of the old, established funds are struggling... that's just natural progression and does not mean the industry overall is doing poorly.

Do any of you work in the business? So far, 2013 has been a good year for many, many quant/algo firms.

 
kingoftheotherroad:

Again I meant, algo prop firms - there is a place for them, but overall the niche has had its run. I do not work in the business, but many of my friends do. Justify your statement that 2013 has been great for quant/algo firms? - on purely prop plays

"Algo trading" is a vague catch-all describing many things, which is why you don't hear it applied to the buyside very much.

What's a "purely prop play"? By play do you mean trading/investment strategy? More quantitative strategies, like HFT, Index and Stat Arb, CTA, etc tend to be very sensitive to execution and trading costs, so they're tightly coupled with the idea and can get conflated together. Also, one can trade/execute via algos no matter what the strategy or product is.

 

I do actually agree with the premise of this thread. I am in a quant hf so not really trading but our holding horizon is short enough where there's some overlap. But I am increasingly dismayed at how narrow the skillset I am developing are becoming and how disconnected they are from the reasons I initially came to wallstreet for. Sometimes I do feel that business school would be a great way to reset my career trajectory.

 

Thanks for your pm. I agree with you that much of trading is VERY narrow; you develop expertise in a specific product/strategy, and it will become difficult to transition into something more interesting and with better exit opps. I highly recommend smart traders to target top b-schools.

 

I think that what is over is the era of simple ultralow-latency & ultrahigh-frequency strategies capitalizing primarily on fast technology and interconnections.

I might be wrong, but I think there is a new era of "big data" strategies coming up - involving heavy machine learning, stats and map-reduce algorithms - which may require minutes or more to compute, but still very exploitable for intra-day trading.

In general, I believe that in contrast to the over-saturated horizon of sub-milisecond strategies the horizon of heavyweight intra-day strategies yet waits to be unlocked.

Some evidence: I know that GS has recently formed an internal machine learning team advising their trading desks. Also, Citadel hired some machine learning people (btw, here is a counterexample to the algo "decline": Citadel HFT had 25.7% net return in 2012). And you have new "smart" shops like Teza technologies that are supposedly doing more research-based strategies.

But again, I might be wrong...

 

I agree with what you said. The technology race in HFT has turned into an arms race, it is kind of humorous imo watching firms compete for miniscule amounts of time. It shows how far technology has come "I think that what is over is the era of simple ultralow-latency & ultrahigh-frequency strategies capitalizing primarily on fast technology and interconnections."

edit: Quant macro? The market has turned into a quarterly/short term forza race.

 

Look, some people love this stuff. But as I said earlier, I think spending your time thinking about and working on ultralow-latency trading is extremely dull and intellectually ungratifying. Others will obviously disagree. In my opinion, HFT is glorified IT, not real finance. And as someone interested in higher-level investment strategies, what these prop trading shops do have no appeal to me. Lot of my friends at these shops feel the same way and think the gig is up.

 
alpha001:
I think that what is over is the era of simple ultralow-latency & ultrahigh-frequency strategies capitalizing primarily on fast technology and interconnections.

I might be wrong, but I think there is a new era of "big data" strategies coming up - involving heavy machine learning, stats and map-reduce algorithms - which may require minutes or more to compute, but still very exploitable for intra-day trading.

In general, I believe that in contrast to the over-saturated horizon of sub-milisecond strategies the horizon of heavyweight intra-day strategies yet waits to be unlocked.

Some evidence: I know that GS has recently formed an internal machine learning team advising their trading desks. Also, Citadel hired some machine learning people (btw, here is a counterexample to the algo "decline": Citadel HFT had 25.7% net return in 2012). And you have new "smart" shops like Teza technologies that are supposedly doing more research-based strategies.

But again, I might be wrong...

Several years on, this turned out to be largely true. Always like to visit old threads on the prop industry and see how well predictions held up. Looks like the the asymptote for execution speed is being reached at most firms...not a lot of money to be made in HFT at this point, at least not for a new shop. ML and data analysis is where it's at...that's the skillset you want to have now. And...it's highly extensible and highly flexible. That's a skillset you can use anywhere now. Going to be in demand for the foreseeable future.

"When you stop striving for perfection, you might as well be dead."
 

In some ways HFT is dying, every industry has a boom/bust cycle and only the most adaptable survive. But I think that has more to do with the new rules and the number of venues available to execute than with any technology frontier. And the fact that money has been flowing out of equities for a long, long time. There are fewer tourists walking into the casino these days.

I have heard more then one director or head trader complain that the fed is sucking up the risk that they would love to take on themselves. Once the fed allows rates to start moving again it could be a different story.

 

In the early 90s Richard Dennis gave up. More recently, Bruce Kovner. As markets evolve, so do those who can consistently win in them. Dennis' Turtle Trading was good enough in the 80s but not afterwards; I think much of this will be true to the less sophisticated algo traders as more and more money is ploughed into the space (after all there can only be one fastest company).

These guys should focus on applying machine learning concepts more broadly to markets, systems with more AI that can literally adapt to changing conditions, unsupervised/deep learning that can select new strategies automatically, and so on. These strategies are not popular atm (afaik) because of the latency penalty; a simple moving average based strategy can execute much faster than anything that has to apply a set of weights to feeds and compute an output decision. But considering the explosion in popularity (and ease of use, thanks to libraries like scikit-learn that allow even your average high school graduate to do ML) of the field at the moment I wouldn't be surprised to see it grow quite a bit, as the talent moves on from failed startups and corporates who don't know what to do with them, to markets who are willing to pay for the technology.

I'll head into that space in 2-3 years I think. Learning curve is kinda steep :P

Here's a Haskell open source algo trading platform which I thought was pretty cool: http://hyperq.github.io/index.html

HN post and discussion: https://news.ycombinator.com/item?id=5550930

 

Is there a pattern here? The quant areas of finance seem to start out very successful, but as everyone piles in the business becomes saturated with everyone running the same strategies - see AQR and GS Global Alpha and their equity factor models, the Getco's/HRT's of today, etc. Now even macro seems to be getting hit, with the arrival of the quant macro crowd. Eventually it reaches an equilibrium where only the established players can afford to stay in, but the growth has leveled off. There are always going to be younger guys coming out of school with sharper quant skills, and market experience doesn't seem to count as much in quant-land. At least, not relative to areas like equity/credit investing, where your experience/judgment factors hugely into your valuation of companies/credits, and the interpretation of the crunched numbers.

If you compare people of equal competence in the fundamental vs. quant side, would you all agree that fundamental is a safer and probably more lucrative place to be? (Obviously the same person can't choose whether he wants to be good at fundamentals vs. quant - this is just a thought experiment.)

 

Any space, whether macro, quant, long-short, etc., could get overcrowded and certain strategies perform better under certain conditions. There's no single "superior" strategy per se. For instance, macro funds got killed in 2011-2012 with the eurozone debt crisis and QE leading to a bizarre "risk-on, risk-off" environment. This year a lot of the big macro funds such as tudor, moore, de shaw oculus, caxton, have been crushing it, the main catalyst being Abenomics and the subsequent divergence in countries' monetary policies.

Also, my critique is limited to the HFT shops that dominate Chicago trading, not quant/algo strategies overall.

 
EURCHF parity:

In the early 90s Richard Dennis gave up. More recently, Bruce Kovner. As markets evolve, so do those who can consistently win in them. Dennis' Turtle Trading was good enough in the 80s but not afterwards; I think much of this will be true to the less sophisticated algo traders as more and more money is ploughed into the space (after all there can only be one fastest company).

These guys should focus on applying machine learning concepts more broadly to markets, systems with more AI that can literally adapt to changing conditions, unsupervised/deep learning that can select new strategies automatically, and so on. These strategies are not popular atm (afaik) because of the latency penalty; a simple moving average based strategy can execute much faster than anything that has to apply a set of weights to feeds and compute an output decision. But considering the explosion in popularity (and ease of use, thanks to libraries like scikit-learn that allow even your average high school graduate to do ML) of the field at the moment I wouldn't be surprised to see it grow quite a bit, as the talent moves on from failed startups and corporates who don't know what to do with them, to markets who are willing to pay for the technology.

I'll head into that space in 2-3 years I think. Learning curve is kinda steep :P

Here's a Haskell open source algo trading platform which I thought was pretty cool: http://hyperq.github.io/index.html

HN post and discussion: https://news.ycombinator.com/item?id=5550930

Agree with everything with exception of the first part.

Kovner retired just as he was about to reach his 70s, obviously that's not an optimal age to deal with the immense stress and focus required to trade macro.

Dennis is a kinda iffy guy - he was a discretionary trader who despite having systems that seemingly exhibited a long-run edge, selected which signals to act on, where to set stops, how to size positions, when to liquidate if stops are hit etc. Clearly he was far from disciplined. On the other hand, ignoring the past 3 years of mediocre performance that has plagued most HFs, the turtles are performing quite well.

I have been exploring ML methods applied to trading, particularly unsupervised learning (which isn't the same thing as deep learning- had to point that out before Kurzweil comes in here), and find that even with tools of cross-validation, it is nearly impossible to avoid curve fitting the data to death, only to see it break down when exposed to real out of sample data. Not saying it is impossible to apply ML, but obviously it is best left to other realms that are more stable.

I'm not sure if I'm mistaken, but you used to be a physical trader - what made you want to transition to the opposite spectrum of technical complexity?

 

Re: Kovner, it was one retirement that really shocked me at the time. There's plenty of old guys rocking it, from Julian Robertson and George Soros to Warren Buffett. His was the last in a string of high profile retirements spun on Bloomberg as "I can't find an edge in current market conditions". I see him as one of the giants of the last 40 years so it stuck even after I left markets.

I was on the "finance" desk so quite far from physical. More FX vol, some exotics. Macro. I saw a massive gap between the much older rainmakers on the desk and my own puny skill. Couldn't see a reliable, provable edge in my thinking. So dropped out to build it. As a former engineer I regretted not studying CS instead, although most undergrad CS courses really suck (even MIT and Stanford dropped Scheme in favor of Python, not as bad as Java but it does damage a young mind...)

Will reply to the ML stuff in more detail later. CV isn't all - what about regularization, PCA, etc? I also think unsupervised learning is extraordinarily expensive to do well today - you need hundreds of parallel servers and some very tight code to get anywhere close to what I'm implying (cf last year's Google paper on cat and human heads from YT videos; just look at the hardware and sheer layers and nodes in that NN, for a fairly simple task). Wish I had half an hour now to write the rest of this paragraph :P

 

Re: Dennis; if his turtles are doing well, it is not because they keep applying what he taught them but because they continuously learned and adapted. I thought his failure was a good example of how sometimes, an older guy just can't keep up with innovation and keep himself ahead of the curve. It can be a problem of the way someone is thinking.

For example: look in CS at someone like Peter Norvig. He continues to write the best books in the field. Compare this to your average C++ code farm employee/manager 10 years ago - where is he today? How you think and your willingness to keep learning and adapting (and moving up abstraction levels) is crucial to long term success regardless of the field.

One example I see today is the sheer number of people who insist on using Hive/Hadoop for big data tasks when Redshift abstracts away a lot of the crap that makes Hadoop so expensive. In fact in general I think AWS is much more groundbreaking than most people in the industry realize.

 

It depends on the specific role of the person at a HFT shop. At a broad level, they analyze volume, order book depth, bid-ask spread, market microstructure, etc., to design and implement strategies that aim to capitalize on very small market inefficiencies. They then rely on technology, whether it's great servers, co-location, ultra-fast code, etc., to "get" to the trade before their competitors.

If you're a hardcore computer science geek with pretty much no other interest, then this could be a great line of work for you. Solid pay, little office BS, hours coding. But for those with more broad interests, it's not a great gig because as I said there is no higher-level analysis of the financial markets at all.

 
mbavsmfin:

Nope. The poster in question is very interested in top business schools precisely because it will give him so many desirable career options. Many traders feel the same way.

Many people have unduly glorified trading for far too long, exaggerating how much money traders make or how "interesting" the work is.

Look, you don't think trading is interesting or lucrative. You're entitled to your opinion, but it would behoove you to recognize that you're in the minority, and that your "evidence" of one mediocre year in one fairly small space that was previously white-hot (HFT) does not generalize to your statement that trading is uninteresting or no longer lucrative.

Trading has ALWAYS been very difficult and competitive; there is no guarantee of success, but success in trading can pay in line with some of the best professional athletes in the world.

Before I end with "agree to disagree", let me just point out that, in my opinion, you massively overestimate how valuable a "top" MBA is, and it is comical how many people on this forum think, for instance, that HBS is a huge value-add or will help them get laid. The average HBS grad is 29 years old, and the median salary out of school is just over $100k, with a full 23% of the class of 2012 not having offers at graduation:

http://www.hbs.edu/about/facts-and-figures/Pages/mba-statistics.aspx

Underwhelming stats.

 
  1. I never said that trading is supposed to be easy or that one cannot make good money doing it. The upper extreme of most industries will do pretty well for themselves. My argument regarding this was twofold. First, due to various structural changes and increased costs, it's gotten very difficult for these prop shops to generate the profit margins they once did (note that i'm talking about prop shops only, not quant funds). Second, my point was in response to many on WSO who talked incessantly about how much money twentysomething traders at these prop shops were making and grossly exaggerating compensation, thus giving a very distorted view of what was going on.

  2. As for the value-add of HBS or any other top MBA, the amount of value ultimately depends on the person and his goals. If you're a CS monkey who wants to spend the rest of his life coding, then yes, MBA will not make sense for you. For many others, an MBA does make sense because they want to transition into a field/firm where an MBA is highly sought after.

  3. Regarding compensation, that number is just the median base salary, which does not include bonuses. Also there are a good chunk of people from HBS who are now doing startups or even nonprofit and thus are either not getting paid or getting a low base. As for % employed upon graduation that number was 77% last year and 89% about 3 months after graduation. The 77% figure seems "low" because buyside recruiting often happens late in the year, with students not finding jobs until after graduating. Moreover, keep in mind that HBS grads are picky and are gonna wait for the right opportunity rather than settling.

 
mbavsmfin:

Any space, whether macro, quant, long-short, etc., could get overcrowded and certain strategies perform better under certain conditions. There's no single "superior" strategy per se. For instance, macro funds got killed in 2011-2012 with the eurozone debt crisis and QE leading to a bizarre "risk-on, risk-off" environment. This year a lot of the big macro funds such as tudor, moore, de shaw oculus, caxton, have been crushing it, the main catalyst being Abenomics and the subsequent divergence in countries' monetary policies.

Also, my critique is limited to the HFT shops that dominate Chicago trading, not quant/algo strategies overall.

This is basically where my opinion lies...strategies do better or worse depending on market conditions and every time one lags for a few years it is briefly considered "dead". Most things described as "quant" or "alto" are by their nature short volatility strategies because they rely on the future looking like the past and previous relationships holding up. Macro, in all of it's forms, is generally a long vol strategy where you are paying someone to think imaginatively about how the future may differ from the past and find opportunities that cannot be found in the historical data. Hence, macro did better in the crisis and its immediate aftermath, "algo" did better in 2011-2012 during a short vol environment, and "macro" is doing better in 2013.

There are plenty of guys who employ regime-switching models that try to catch the short vol and the long, but those still inherently rely on historical data to decide which model to use which makes them not as great as they sound...for example this year USDJPY went from a mean-reverting market to a massive, rapid trending market in a heartbeat based on policies that have literally never been tried before in any market. Some model that turns on its "long vol" model based on something like VIX or currency vol, or something like that had very little chance against a seasoned human that could look at the newspaper and say "dude this is fukked up lets get short". Where I work we have a few teams doing quant macro-stuff and they didnt have much success catching the bulk of the moves this year I believe because it happened too rapidly....now of course if they reset everything to move faster and we go back to the 2011-2012 environemn t they will be screwed again...so point being its all an art not even close to inventing a machine that is perfect as of yet and Im not sure machine learning is helpful given the limitations of the data vs the outcomes that the real world sometimes produces.

HFT is an altogether different animal that to me is almost a replacement for the pit traders of old...ie when a normal buyside guy comes in to trade, they pay a commision, and then they also get robbed a bit by a HFT algo just like the floor trdaers used to do. That job can be lucrative, but as more players enter and perfect the craft it is never going to be as lucrative as it was in the early days of HFT.

 

Bondarb, very well said as usual.

I noticed that a lot of traditional macro funds such as Tudor and even Bridgewater are now deploying quant strategies. About half of Tudor's AUM is now under the quant umbrella, and I hear Bridgewater uses a "secret" algorithm to determine which alpha signals are being generated (rumor is that only like Dalio and a few of his lieutenants are able to see the algo). Is this a trend you see happening across macro funds?

 
mbavsmfin:

Thanks for your pm. I agree with you that much of trading is VERY narrow; you develop expertise in a specific product/strategy, and it will become difficult to transition into something more interesting and with better exit opps. I highly recommend smart traders to target top b-schools.

what do you consider more interesting jobs with better exit opps? as someone with trading work experience and no MBA, I'm not interested in staying in trading for reasons stated in this thread but I also don't want to throw away my work experience after working hard to get into it

 

Well, this is my personal opinion but within finance, I think working at a large macro hedge fund, special situations, or even long-short is way more interesting than trading. I would also throw in mutual funds such as wellington/pimco/capital group/t rowe price/fidelity into the mix.

In terms of pure exit opps, if your main concern is breadth of opportunities, almost nothing beats MBB consulting since the skills you acquire is transferable across so many industries. I would also say that tech has pretty good exit opps in the sense that you could join a startup if you get bored.

 
mbavsmfin:

Bondarb, very well said as usual.

I noticed that a lot of traditional macro funds such as Tudor and even Bridgewater are now deploying quant strategies. About half of Tudor's AUM is now under the quant umbrella, and I hear Bridgewater uses a "secret" algorithm to determine which alpha signals are being generated (rumor is that only like Dalio and a few of his lieutenants are able to see the algo). Is this a trend you see happening across macro funds?

i have seen quant-macro come in and out of style a few times in the last 8-10 years. In 2006 you almost had to be quant to raise money, then in the blow-up of 2007-09 the discretionar guys did much better and the quants fell out of favor. Last couple of years I have seen the strategy gain fabvor again but not as much as pre-housing bust and this year has seen a mini-revival of old-school macro.. However quant macro is a very broad umbrella...for example I am a discretionary portfolio manager but I use enuff quant tools that if I was raising money I could definitely pass as "quant macro" if i had to do it to check a box for sone institutional investor. I am cynical but I think alot of the algo trading World in macro is taking simple styles and dressing them up for marketing purposes.

Bridgewater is supposedly almost 100% model driven but to me they are just a massive leveraged long bond position right now and have been for the past few years...which is why they dropped like 10% in may and june.

 

If bridgewater is 100% model driven, why do they have so many freaking employees? The investment associates there do a lot of macroeconomic research, so I'm guessing Dalio takes the best ideas and incorporates them into his model. I could be way off base on this though.

 
mbavsmfin:

If bridgewater is 100% model driven, why do they have so many freaking employees? The investment associates there do a lot of macroeconomic research, so I'm guessing Dalio takes the best ideas and incorporates them into his model. I could be way off base on this though.

i have no idea why he has so many employees...i have heard people theorize seriously that he enjoys his ridicculous sociological experiment with the rules, criticisms, etc and so he enjoys having all the employees to fukk with. And this is from a friend that has good connections with to the company...and he wasnt kidding. its a fukked up place generally i wouldnt want to work there.

 

I would not be surprised at all if that's true. It seems like only a handful of people at bridgewater actually have a say in strategy and everyone else is well-paid window dressing (especially the traders, who are in a separate campus from the research folks).

 
mbavsmfin:

1. I never said that trading is supposed to be easy or that one cannot make good money doing it. The upper extreme of most industries will do pretty well for themselves. My argument regarding this was twofold. First, due to various structural changes and increased costs, it's gotten very difficult for these prop shops to generate the profit margins they once did (note that i'm talking about prop shops only, not quant funds). Second, my point was in response to many on WSO who talked incessantly about how much money twentysomething traders at these prop shops were making and grossly exaggerating compensation, thus giving a very distorted view of what was going on.

2. As for the value-add of HBS or any other top MBA, the amount of value ultimately depends on the person and his goals. If you're a CS monkey who wants to spend the rest of his life coding, then yes, MBA will not make sense for you. For many others, an MBA does make sense because they want to transition into a field/firm where an MBA is highly sought after.

3. Regarding compensation, that number is just the median base salary, which does not include bonuses. Also there are a good chunk of people from HBS who are now doing startups or even nonprofit and thus are either not getting paid or getting a low base. As for % employed upon graduation that number was 77% last year and 89% about 3 months after graduation. The 77% figure seems "low" because buyside recruiting often happens late in the year, with students not finding jobs until after graduating. Moreover, keep in mind that HBS grads are picky and are gonna wait for the right opportunity rather than settling.

  1. Outside of finance, what industries are going to pay twenty somethings $500k+?
  2. I disagree completely; an MBA is probably most valuable for the CS monkey, and it would complement pre-existing hard skills.
  3. The extra compensation beyond salary for HBS is about $20k: hbs.edu/recruiting/mba/data-and-statistics/employment-statistics.html">http://www.hbs.edu/recruiting/mba/data-and-statistics/employment-statis…. The people getting extra comp (largely going to finance, lol) were probably already making a lot of money pre-MBA. Even 11% unemployed three months out is really bad; going to an MBA program and coming out without a job lined up is a travesty. Startups are not magical places; they're more risky than trading with at best the same expected long-run comp. Median non-profit/gov't is 3% of class, and at $90k so is not bringing down the numbers much.
 
  1. Outside of finance, what industries are going to pay twenty somethings $500k+?

Please, put the crack rock down. No one's getting 500k as a regular twenty-something in finance. Saying "well what about this guy" is pretty stupid once you realize that you can say "what about this guy" with the tech industry, with private equity, with small business entrepreneurship, etc. Finance is not a gold mine for fast money

 

Dude every single one of your posts is bashing how much people make in trading, and then you pimping small business/franchises as a way to actually make money. It's pathetic, just cause you're a failed trader at a tiny energy shop doesn't make you the authority on S&T compensation. Most VP's are in their late 30's and will pull in 350k-600k. at a bank. do us all a favor and please stop posting.

 
Bondarb:
mbavsmfin:

If bridgewater is 100% model driven, why do they have so many freaking employees? The investment associates there do a lot of macroeconomic research, so I'm guessing Dalio takes the best ideas and incorporates them into his model. I could be way off base on this though.

i have no idea why he has so many employees...i have heard people theorize seriously that he enjoys his ridicculous sociological experiment with the rules, criticisms, etc and so he enjoys having all the employees to fukk with. And this is from a friend that has good connections with to the company...and he wasnt kidding. its a fukked up place generally i wouldnt want to work there.

Capacity and reward/risk tend to scale with number of models. I'm sure they have thousands.

 

The people you know that are looking into top MBA programs did they major in cs or engineering as an undergrad? If so I would say go for it. On the other hand if they were business majors as undergrads in my opinion an MBA is a waste of money.

 

I'm assuming you work at a quant fund. In that case it makes sense, sort of like how the geeks at star trek conventions would make fun of the popular guys. :)

In all seriousness, in most areas of finance, top MBAs are definitely not made fun of. You think HBS/Stanford/Wharton MBAs are being made fun of at say blackstone/kkr/bain capital/och-ziff/davidson kempner/baupost, etc.?

The bottom line is that in pretty much any finance role where programming is NOT the main job description, an MBA from a top school is highly coveted.

 
Bondarb:
mbavsmfin:

Any space, whether macro, quant, long-short, etc., could get overcrowded and certain strategies perform better under certain conditions. There's no single "superior" strategy per se. For instance, macro funds got killed in 2011-2012 with the eurozone debt crisis and QE leading to a bizarre "risk-on, risk-off" environment. This year a lot of the big macro funds such as tudor, moore, de shaw oculus, caxton, have been crushing it, the main catalyst being Abenomics and the subsequent divergence in countries' monetary policies.

Also, my critique is limited to the HFT shops that dominate Chicago trading, not quant/algo strategies overall.

This is basically where my opinion lies...strategies do better or worse depending on market conditions and every time one lags for a few years it is briefly considered "dead". Most things described as "quant" or "alto" are by their nature short volatility strategies because they rely on the future looking like the past and previous relationships holding up. Macro, in all of it's forms, is generally a long vol strategy where you are paying someone to think imaginatively about how the future may differ from the past and find opportunities that cannot be found in the historical data. Hence, macro did better in the crisis and its immediate aftermath, "algo" did better in 2011-2012 during a short vol environment, and "macro" is doing better in 2013.

There are plenty of guys who employ regime-switching models that try to catch the short vol and the long, but those still inherently rely on historical data to decide which model to use which makes them not as great as they sound...for example this year USDJPY went from a mean-reverting market to a massive, rapid trending market in a heartbeat based on policies that have literally never been tried before in any market. Some model that turns on its "long vol" model based on something like VIX or currency vol, or something like that had very little chance against a seasoned human that could look at the newspaper and say "dude this is fukked up lets get short". Where I work we have a few teams doing quant macro-stuff and they didnt have much success catching the bulk of the moves this year I believe because it happened too rapidly....now of course if they reset everything to move faster and we go back to the 2011-2012 environemn t they will be screwed again...so point being its all an art not even close to inventing a machine that is perfect as of yet and Im not sure machine learning is helpful given the limitations of the data vs the outcomes that the real world sometimes produces.

HFT is an altogether different animal that to me is almost a replacement for the pit traders of old...ie when a normal buyside guy comes in to trade, they pay a commision, and then they also get robbed a bit by a HFT algo just like the floor trdaers used to do. That job can be lucrative, but as more players enter and perfect the craft it is never going to be as lucrative as it was in the early days of HFT.

This post put a lot of things in perspective and confirmed the futility in a few things that I wasted time exploring, particularly various forms of regime detection models. I now believe that instead trying to detect regimes or get religious about any particular one, it would be far better to concurrently run the maximum number of uncorrelated strategies that will invariably be categorized as long/ short vol over various time-frames, with a biased weight towards long vol divergence strategies. The logic follows the notion that mathematically, if the number of uncorrelated assets or strategies approaches infinity, overall portfolio volatility should approach the mirage of zero. Whether these strategies should be developed automatically using methods such as genetic programming or through manual simulations is something that I'm naively unsure of, but I'm currently inclined towards the latter.

EURCHF Parity, I'm really curious to hear your opinion about this. If one were to take the evolutionary programming route, those EC2 clusters would sure as hell come in handy, and I totally agree that AWS is underrated.

Never found the need to use Hadoop or RedShift, even when I dealt with intra-day data north of 1TB. I mostly use end-of-day and oddly time-stamped intra-day data which makes me a snail.

 
Macro Arbitrage:
Bondarb:
mbavsmfin:

Any space, whether macro, quant, long-short, etc., could get overcrowded and certain strategies perform better under certain conditions. There's no single "superior" strategy per se. For instance, macro funds got killed in 2011-2012 with the eurozone debt crisis and QE leading to a bizarre "risk-on, risk-off" environment. This year a lot of the big macro funds such as tudor, moore, de shaw oculus, caxton, have been crushing it, the main catalyst being Abenomics and the subsequent divergence in countries' monetary policies.

Also, my critique is limited to the HFT shops that dominate Chicago trading, not quant/algo strategies overall.

This is basically where my opinion lies...strategies do better or worse depending on market conditions and every time one lags for a few years it is briefly considered "dead". Most things described as "quant" or "alto" are by their nature short volatility strategies because they rely on the future looking like the past and previous relationships holding up. Macro, in all of it's forms, is generally a long vol strategy where you are paying someone to think imaginatively about how the future may differ from the past and find opportunities that cannot be found in the historical data. Hence, macro did better in the crisis and its immediate aftermath, "algo" did better in 2011-2012 during a short vol environment, and "macro" is doing better in 2013.

There are plenty of guys who employ regime-switching models that try to catch the short vol and the long, but those still inherently rely on historical data to decide which model to use which makes them not as great as they sound...for example this year USDJPY went from a mean-reverting market to a massive, rapid trending market in a heartbeat based on policies that have literally never been tried before in any market. Some model that turns on its "long vol" model based on something like VIX or currency vol, or something like that had very little chance against a seasoned human that could look at the newspaper and say "dude this is fukked up lets get short". Where I work we have a few teams doing quant macro-stuff and they didnt have much success catching the bulk of the moves this year I believe because it happened too rapidly....now of course if they reset everything to move faster and we go back to the 2011-2012 environemn t they will be screwed again...so point being its all an art not even close to inventing a machine that is perfect as of yet and Im not sure machine learning is helpful given the limitations of the data vs the outcomes that the real world sometimes produces.

HFT is an altogether different animal that to me is almost a replacement for the pit traders of old...ie when a normal buyside guy comes in to trade, they pay a commision, and then they also get robbed a bit by a HFT algo just like the floor trdaers used to do. That job can be lucrative, but as more players enter and perfect the craft it is never going to be as lucrative as it was in the early days of HFT.

This post put a lot of things in perspective and confirmed the futility in a few things that I wasted time exploring, particularly various forms of regime detection models. I now believe that instead trying to detect regimes or get religious about any particular one, it would be far better to concurrently run the maximum number of uncorrelated strategies that will invariably be categorized as long/ short vol over various time-frames, with a biased weight towards long vol divergence strategies. The logic follows the notion that mathematically, if the number of uncorrelated assets or strategies approaches infinity, overall portfolio volatility should approach the mirage of zero. Whether these strategies should be developed automatically using methods such as genetic programming or through manual simulations is something that I'm naively unsure of, but I'm currently inclined towards the latter.

EURCHF Parity, I'm really curious to hear your opinion about this. If one were to take the evolutionary programming route, those EC2 clusters would sure as hell come in handy, and I totally agree that AWS is underrated.

Never found the need to use Hadoop or RedShift, even when I dealt with intra-day data north of 1TB. I mostly use end-of-day and oddly time-stamped intra-day data which makes me a snail.

I wouldnt toss out the concept or running regime-shifting models just because i said I have seen some people who havent done well with them...maybe your work will end up being better. I have not really seen a regime-shifting model that I am convinced is better then a very good human at picking regime changes, but that certainly doesnt mean one isnt out there or that you cant develop one. I generally believe in following your own instincts rather then being deterred and changing the second someone else says your concept wont work (this goes for just about everything in life not just trading)...

 
Best Response
mbavsmfin:

I'm assuming you work at a quant fund. In that case it makes sense, sort of like how the geeks at star trek conventions would make fun of the popular guys. :)

In all seriousness, in most areas of finance, top MBAs are definitely not made fun of. You think HBS/Stanford/Wharton MBAs are being made fun of at say blackstone/kkr/bain capital/och-ziff/davidson kempner/baupost, etc.?

The bottom line is that in pretty much any finance role where programming is NOT the main job description, an MBA from a top school is highly coveted.

i do not work at a quant fund and we rarely if ever hire MBAs. if you want to trade two years spent doing team building exercises and taking ski trips is not valuable.

 
Bondarb:

I wouldnt toss out the concept or running regime-shifting models just because i said I have seen some people who havent done well with them...maybe your work will end up being better. I have not really seen a regime-shifting model that I am convinced is better then a very good human at picking regime changes, but that certainly doesnt mean one isnt out there or that you cant develop one. I generally believe in following your own instincts rather then being deterred and changing the second someone else says your concept wont work (this goes for just about everything in life not just trading)...

Amen to that first sentence. It took 15 iterations for my first ML based product to be good. The field is relatively new and a lot of development (for me) was about exploring edge cases or just flawed logic in the algo, as opposed to collecting data and blindly putting it through a scikit-learn library (in fact I am a huge fan of implementing a NN from first principles, and ask anybody who wants to work on my team to do so in a functional or array language of their choice amongst other things). Cf Knuth and black boxes - I interpret what he said as meaning "know what every line of your code does in detail".

Same for not listening to others. If you have an idea, implement it to death before you give up. There are always a dozen paths to your output, picking the best takes a while and a lot of (thinking) work.

 
Macro <span class=keyword_link><a href=/resources/skills/trading-investing/arbitrage target=_blank>Arbitrage</a></span>:

EURCHF Parity, I'm really curious to hear your opinion about this. If one were to take the evolutionary programming route, those EC2 clusters would sure as hell come in handy, and I totally agree that AWS is underrated.

Never found the need to use Hadoop or RedShift, even when I dealt with intra-day data north of 1TB. I mostly use end-of-day and oddly time-stamped intra-day data which makes me a snail.

Re: big data tools, I'm surprised you didn't need them on 1TB datasets - I assume you were doing a lot of in-place C stuff? I guess I don't strictly speaking "need" Redshift, however my development cycle is very short (3-5 days, sometimes 1 day if product is simple enough) so I don't like waiting even 20 minutes for a data processing task to finish. I stick it on Redshift, do the preprocessing there (for example on server log files or raw web analytics data) and then get the training data in the right format back out to be fed to the ML product. (bear in mind I started learning CS only last year, so am missing out a lot of the picture)

Do you know how you can severely improve a ML algorithm by doing smart preprocessing? E.g. say you are trying to detect vibration patterns which are indicative that your part is about to break, if you do a Fourier transform of the vibration data feed and then give the frequencies you know are indicative of breakage to the algo as input, you'll save yourself a lot of pain vs an unsupervised or even supervised, but on the raw data ML algo.

So for me the opportunity - low hanging fruit - lies in looking into how successful traders/investors think, and trying to replicate their thought patterns, at least as a preprocessing tool, on a much less granular level than HFT is operating at (so 1-3 year holding periods, as for a retail investor). At the end of the day, someone like Soros or Buffett is just a very complex set of neural nets with a billion+ input vector. If you look at how the brain works it's a set of neural nets which lie in the subconscious and can take several MB of data per second; when one detects y = 1 it fires up the 4KB/s output towards the conscious level and that starts the process of thinking about what action to take. These guys see the same data as you. Assuming most L/S equity funds are not just insider trading (and by this I include taking execs from companies out, getting them drunk and piecing together the mosaic), the information in the 10Ks should be enough for your algorithm to pre-select promising companies for you. Let's say I wanted to pick 2 insurance companies to add to my home portfolio. I would probably stick every line of the 10K as an input, set up a fairly high level features SVM or NN (this is what your average investor selecting via price to book or liquidity ratios is doing anyway, but you are making it more systematic and thorough), take 30 or 50 years of performance data, feed the date/time period as a feature as well, and set some kind of optimization objective to be the outperformance of one stock vs the group across my estimated holding period. I'm sure there's a way to learn the specific conditions in which you happen to be currently based on macro inputs like interest rates. I'd stick a 6% outperformance level as being my positive Y as I'm retail and it costs me about that much to trade stocks (3% in 3% out). Going further, I'd look into famous books on value investing (like Graham but also Financial Shenanigans) and try and model their insights as another supervised learning algorithm that I could use to filter out particular stocks from my training data.

Another one is correlation (or the general task of clustering stocks for your balanced portfolio). Instead of just looking at correlation between two stocks, I would find the ratio of the days they moved together (defined as similar % move in the same direction X days) over the total days they move as a similarity index; I would then map this to a graph with each vertex a stock and edges the connection if the index passes a certain level. Then you could traverse that graph recursively to figure out the connection between stocks that are not necessarily connected on the first degree but are on second or third degree and get a better idea of whether you are truly diversified (I do something similar to figure out how products we sell are connected, as chances are product 1 was not bought with product 2 because product 2 was not seen). Or do this with lines of the CF statement or P&L or whatever. This assumes there are patterns in markets that repeat themselves over time; I make this assumption because books like Reminiscences of a Stock Operator are still applicable today and reading Galbraith's book on the 1929 crash was - with wording only slightly altered - like reading headlines and analyses of 2007-9.

The issue I have with those who try and build a magic NN that will somehow trade for them and replace the human is that they just don't realize how complex the problem is. Look at the Google paper last year on higher features: over 1,000 CPUs in parallel, with 100 million fairly high resolution images fed, for what? detecting a human or cat face. How on Earth are you hoping that your tiny little cluster or 50 EC2 XL instances is going to get anywhere near complex or good enough to abstract the idea of a risk adjusted P&L?

 

Re the Abenomics comment, that was ENTIRELY predictable. Both the senior guys on my desk 2 years ago (phew, so long now) were prepared for a yen breakdown, which they saw as inevitable based on their 30+ years combined experience trading JPY and yen-related exotics in both equities and FI. If they picked it up so would a sufficiently well trained, accurate global macro ML algo.

 
EURCHF parity:

Re the Abenomics comment, that was ENTIRELY predictable. Both the senior guys on my desk 2 years ago (phew, so long now) were prepared for a yen breakdown, which they saw as inevitable based on their 30+ years combined experience trading JPY and yen-related exotics in both equities and FI. If they picked it up so would a sufficiently well trained, accurate global macro ML algo.

I agree that short yen was a well-flagged story, but USDJPY had many many head-fakes over the last two years before it finally broke out and kept going....it had been a macro graveyard in 2011-2012. The rise of Abe was just the latest in a series of events that led many discretionary macro folks to want to be short JPY and therefore alot of people had too many scars from previous trades and didnt get in when it actually made its real move. The same thing happened in 2007...despite what yopu read the housing bubble was widely predicted and pretty easy to see, but most of the guys who saw it had gotten carried out of the trade in 2005-2006. I am not saying a model cannot catch a move like that and I am sure many did (although i would say discretionary traders fared better then models on this one), but it wasnt easy to see how it was different from 2011-2012 just by looking at prices especially since many other indicators that people use to gauge the environment were not flashing that a regime change was at hand.

 
mbavsmfin:

If bridgewater is 100% model driven, why do they have so many freaking employees? The investment associates there do a lot of macroeconomic research, so I'm guessing Dalio takes the best ideas and incorporates them into his model. I could be way off base on this though.

Bridgewater isn't 100% model driven. Far from it. Even the AllWeather part, which is what y'all are referring to, is based on some discretionary (and somewhat arbitrary) assumptions. PureAlpha is just as discretionary as everyone else.
 

Sure. It's definitely not a quant model driven fund such as rentech,de shaw, or two sigma. But they definitely use models to generate their alpha signals (more than 100 of these signals), which are then relayed to the traders for execution. However, Bridgewater doesn't have quants per se, so my guess is that the investment associates and portfolio strategists conduct extensive research on various macroeconomic topics (ie, history of bond yields in Italy going back 50 years or impact of oil shocks on U.S. stocks) and then incorporate those insights into some sort of "master" algorithm that spits out alpha signals when certain conditions are triggered. I recall talking to someone whose friend was actually hired away from microsoft to re-write bridgewater's "master" algo, the one that only dalio and a few others have access to. They offered him like $1 million/year for the job.

 
Bondarb:

I agree that short yen was a well-flagged story, but USDJPY had many many head-fakes over the last two years before it finally broke out and kept going....it had been a macro graveyard in 2011-2012. The rise of Abe was just the latest in a series of events that led many discretionary macro folks to want to be short JPY and therefore alot of people had too many scars from previous trades and didnt get in when it actually made its real move. The same thing happened in 2007...despite what yopu read the housing bubble was widely predicted and pretty easy to see, but most of the guys who saw it had gotten carried out of the trade in 2005-2006. I am not saying a model cannot catch a move like that and I am sure many did (although i would say discretionary traders fared better then models on this one), but it wasnt easy to see how it was different from 2011-2012 just by looking at prices especially since many other indicators that people use to gauge the environment were not flashing that a regime change was at hand.

But that's not what models are for! Or technology, generally. Technology (including whatever you build in Excel) only enables you to better understand and act on thoughts you have yourself had (a brain extension if you will). For example, options are routinely quoted in vol points instead of prices. That assumes BSM (according to Emanuel Berman, mostly because once BSM based pricers were in place nobody could be bothered to move on to models pricing in leptokurtosis) and of course you don't always assume that this reflects reality but vol points and greeks are much easier to reason about ("this is what the market is pricing") than prices in option world. Same with VaR. In my experience people used VaR as a proxy for capital allocation, because it's much better than any other measure in that respect, rather than as a true measure of risk. In fact at my old firm, when VaR exploded management was understanding that it was usually because the events the desk wanted to happen were happening and a VaR overuse was approved fairly painlessly. (the parallel with the ML world is PCA, which rationalizes features into more complex features incorporating several similar features).

I was just responding to the suggestion that Abenomics was completely unpredictable and caught a lot of people unprepared. I think it was as irresponsible to have an implicit yen position with that looming above, as to have been implicitly long RE in the late 00s. I also think a well crafted ML model that serves as your interface to "what is going on in the world" would pick that up.

Interestingly, if you look at how Paulson/Pellegrini played the mortgage trade, it was very much using technology in the form of attempting to find an instrument that was mispriced. Turned out CDS were such an instrument. I would say 80% of being a good macro trader is about picking the right instrument for the trade. E.g. short China by shorting the KOSPI and AUD, not Shanghai stocks. Similarly, those who were generally short real estate in 2005-06 and subsequently got killed, should have played it safer. The wise old guys like Jeremy Grantham prefer the "stay on the sidelines" approach. The sophisticated value guys like Klarman just waited and stayed big time in cash. The most extreme version was Beal who just closed his bank waiting for the blow up, racing cars and playing poker instead.

Regarding indicators, when I studied for CFA (failed level II and gave up) I thought a lot of them were flawed or gamed. That's why I am so keen to get back and apply ML because that's precisely where I think the crude, still being developed, low level methods we can afford to use today will shine, and are underused.

 

i think ur point about products is a good one but as someone who launched my career by doing very well in the crash of 2008 I can say that they for me was not being stubborn...ie in 2006-2007 whern i got stopped out of short risk trades, i just took them off, took the loss, but never lost enuff that it threatened my seat or scarred me emotionally. a lack of intelectual arrogance is really important in these situations...u r never smarter then the market so the goal when your not doing well is always to preserve your capital (mental as much as physical) so that you can be there when the market comes around to your view.

 
Bondarb:

Bridgewater is supposedly almost 100% model driven but to me they are just a massive leveraged long bond position right now and have been for the past few years...which is why they dropped like 10% in may and june.

If they are so over-leveraged do you think they might blow up in a changing yield environment LTCM style?

BTW, this thread has become really interesting and contains some of the best discussions on global macro trading products and strategies I have seen. Especially the back and forth between EURCHF and Bonardb. As someone who is not in that industry, these have been invaluable for me to gain some insights on how things work in that space. I felt like sitting in the front row of a panel discussion with top notch professionals in the field.

Too late for second-guessing Too late to go back to sleep.
 
Martinghoul:

AllWeather is not "overleveraged", just leveraged (there are other, smaller risk parity funds are quite a bit more spicy). They're not likely to blow up, more like experience some not so good times. Given the good times they've had previously, it's only fair.

I see. As I took another look at bonarb's post I see that he said "massive leveraged" (large AUM) not "massively" leveraged. I should have read more carefully, a couple letters made a big difference.

Too late for second-guessing Too late to go back to sleep.
 

To be sure, like I said, Bridgewater is two main big bits: AllWeather and PureAlpha. The bit that bondarb is referring to is AllWeather (arnd $70bn AUM), which isn't really a "macro hedge fund" per se, as it's mostly a passive strategy (that happens to use leverage). That's the fund which, along with a bunch of other risk parity funds, had a hard time recently during the big bond mkt selloff (I suspect, actually, that they caused quite a bit of it themselves).

 
Martinghoul:

To be sure, like I said, Bridgewater is two main big bits: AllWeather and PureAlpha. The bit that bondarb is referring to is AllWeather (arnd $70bn AUM), which isn't really a "macro hedge fund" per se, as it's mostly a passive strategy (that happens to use leverage). That's the fund which, along with a bunch of other risk parity funds, had a hard time recently during the big bond mkt selloff (I suspect, actually, that they caused quite a bit of it themselves).

BTW what do you think are some of the other trades and instruments to short China in light of the unraveling of that country's massive and largely insolvent shadow banking system? EURCHF mentioned shorting the KOSPI and AUD. I know that a number of other HFs have been shorting mid-sized Chinese lenders (e.g. Mingsheng, Shenzhen etc.) in the hope that the government would sacrifice them, other firms are shorting the Chinese CDS credit indices--an imperfect proxy but no one sells CDS protections on individual Chinese firms as far as I know.

I guess my fundamental question is would the unraveling of China's shadow banking system derail the country's attempt at soft landing? At this point I think some of the bigger players are not really concerned with the shadow banking risk factor. I was at a conference this week where Doug Hodge, the COO of PIMCO was the keynote speaker. He went over PIMCO's assessment of each major economy. On China he was confident that the glass was "3/4 full" and that the country should be able to engineer a soft landing of 7.5% GDP growth just fine while keeping its RMB peg intact. I know some traders that are betting on the opposite, that the shadow banking system (as the Chinese lenders are holding more assets off balance sheets than on their books, despite the illusion of low NPL ratio) would cause the country's financial system to implode and result in, among other things, insolvency of many lenders and massive devaluation of the RMB.

Too late for second-guessing Too late to go back to sleep.
 

It's a very tough one, China...

In general, I would have to say that, based on the work that we have done at my shop, we're cautiously optimistic. Not quite "glass 3/4 full", but our overall view is that the central government a) has more than enough resources to deal with a modest adjustment; and b) generally has a handle on things and is full of intelligent and capable people who know what they're doing. As a result, we don't believe that there's some sort of a massive explosion waiting to happen.

This view has two implications, trade-wise. Firstly, to trade the sort of a "softish landing w/a culling of the weak" scenario you need to do some serious work on the specific names. I know people who are doing that pretty well, but it's not really my cup of tea. Other than that sort of approach, I can't really think of any obvious broad macro trade for this outcome, although it could be as simple as being long Aussie rates a little further out (they're pricing in hikes, after cuts this year).

If you want to bet on a big China blowup, the implications of that are pretty immense and will be felt everywhere. If you want to bet on that, there's all sort of cheap(ish) optionality out there. I would say that the most obvious one for me is USD, which will probably go ballistic.

Just my Z$2c, so take w/pinch of salt, esp in light of the fact that I am mostly not a macro guy...

 

Bashing how much money people make in trading because the general consensus here is that you can just run in there and make tons of money when in reality that's not the case. "Most VP's are in their late 30's and will pull in 350k-600k. at a bank." This alone shows me you don't know what you're talking about. Being a VP at a bank is not necessarily more than VP at a company in other industries like tech. You try to say "well what about x guy" which makes no sense. You can just as easily reference a higher-up guy in another industry at the same role that makes the same. I'm not saying don't do trading at all or it's dead or you definitely won't make money, but people need to stop posting bogus crap making it seem like every other guy makes that much. They don't. This is not a growing industry, that's a fact and not debatable.

 

My mistake, meant to say late 20's. and comparing a VP at a BB to a tech company isn't exactly a fair comparison, considering VP is generally considered C-level at most tech/industry companies compared to the numerous VP's at a BB. You're right its not like the old days where everyone was killing it and making millions, but if you survive 8 years in S&T on a decent desk you'll be pulling around 500k, which puts it right around the ibd/pe compensation for someone at the same level in their career. Someone who's been working at microsoft/facebook/yahoo will be making 200-300 a year by the time their roughly 8 years into their career, which is completely fair considering it has a much better lifestyle and much less risk. I still don't see why you'd have any insight into BB S&T compensation considering you haven't worked there. (Source: older brother is a vp and uncle is an MD in S&T at a BB)

 

I haven't heard the same things, I know a guy who's a VP at a BB and even though I don't know the exact number he makes, I know it's not THAT high because we talked about business (he's a partner in a nightclub here in the city) and he said if he had the whole stake, he'd probably leave his job. He obviously feels stuck and we were talking about making big money, so that implied he wasn't making huge money. Another guy who was VP at one of the banks, can't remember which one, quit his job to go start a business. Sorry if it seems I'm saying business is lucrative here, just telling the story. Nowhere I look online in compensation reports and such do I find such high figures. I'm sure there's guys that make some good money but there's not that many of these roles, and in that case we're talking about exceptions. And if you talk about exceptions, then you can just say what about the guy who idk.....made cheez whiz and made millions. It's rare, is what I'm saying.

 

Trader at an Event-Driven Hedge fund here, mostly execution, but am getting to the stage of coming up and implementing some of my own trade ideas after 5 years here straight out of college and a CFA Charter under my belt. Long term the industry itself is not, in my view, a sensible way to pursue long term fulfilment/happiness. I earn more than the average dude my age but the job is generally boring and the execution is something I can do with my eyes closed. Nobody kid yourselves, it is gambling plain and simple. If you are a builder and you spend six months building a wall, you don't come in one morning and find that the wall is gone. In the hedge fund game an entire years' work (which is of course only measured in PnL) can disappear in a month, a week or even a day if things go really bad. It creates an odd mindset, in other industries you produce something THEN get paid for it....One thing I will never understand is why smart CS and Engineering grads get into this industry. Why not do something interesting like designing pil rigs, building infrastructure projects or complex computer systems for some real tangible purpose? That way you can point to a project and say...that was me, I designed/built that. If you work in an industry that revolves solely around money, all you will have to show for it in the end, if you are lucky, is a healthier bank balance.

 

Just wanted to bump this again, end of page 1 and all of page 2 is some of the best discussion I've seen on this site, and very helpful.

Also would love to see people chime in about what they think is going to happen in the prop scene over the next few years. Have heard generally negative things from many people, but some are still optimistic.

"When you stop striving for perfection, you might as well be dead."
 

Quia libero est molestiae vero dolores. Et magni et suscipit cumque. Est voluptatem distinctio repudiandae quasi sapiente.

Placeat aliquid et error sequi quaerat est possimus. Repellendus omnis nam perspiciatis ut nobis ipsa. Est tenetur voluptatem et. Quo eum id ut sit.

Amet autem sed deleniti quibusdam eaque. Velit voluptatum et libero eum rerum. Accusamus exercitationem ab cupiditate aperiam labore fugit nulla.

 

Ut et libero laborum quae sit id alias. Nam ex voluptas ratione consequatur ullam voluptates. Eligendi expedita nostrum et officiis quis nihil. Eaque nobis autem blanditiis qui molestias minus. Sequi error et consequatur enim vitae cum voluptas recusandae.

Adipisci beatae provident saepe ut. Quia sunt nesciunt asperiores esse aut.

Est cum ut voluptatibus omnis dolore. Occaecati quis consequatur quia modi possimus sed. Occaecati fuga aliquam nostrum dolore dolore doloremque.

 

Suscipit ab voluptate est voluptatem consequatur ut. Dolores odit rerum voluptatem veritatis omnis ut. Et quam deleniti rerum ut odio. Necessitatibus culpa dolor rerum earum temporibus iusto dolorum.

Nam molestias doloribus nam pariatur molestiae dolores optio sit. Et quos dolore aut in sit sit. Ut incidunt molestiae ea quia. Exercitationem omnis illum quaerat est architecto laudantium maiores. Sint aut eos aspernatur ullam. Repudiandae et eum est sit nulla. Corporis ullam non cum ut voluptas illo aut vel.

Career Advancement Opportunities

March 2024 Investment Banking

  • Jefferies & Company 02 99.4%
  • Goldman Sachs 19 98.8%
  • Harris Williams & Co. (++) 98.3%
  • Lazard Freres 02 97.7%
  • JPMorgan Chase 03 97.1%

Overall Employee Satisfaction

March 2024 Investment Banking

  • Harris Williams & Co. 18 99.4%
  • JPMorgan Chase 10 98.8%
  • Lazard Freres 05 98.3%
  • Morgan Stanley 07 97.7%
  • William Blair 03 97.1%

Professional Growth Opportunities

March 2024 Investment Banking

  • Lazard Freres 01 99.4%
  • Jefferies & Company 02 98.8%
  • Goldman Sachs 17 98.3%
  • Moelis & Company 07 97.7%
  • JPMorgan Chase 05 97.1%

Total Avg Compensation

March 2024 Investment Banking

  • Director/MD (5) $648
  • Vice President (19) $385
  • Associates (86) $261
  • 3rd+ Year Analyst (13) $181
  • Intern/Summer Associate (33) $170
  • 2nd Year Analyst (66) $168
  • 1st Year Analyst (202) $159
  • Intern/Summer Analyst (144) $101
notes
16 IB Interviews Notes

“... there’s no excuse to not take advantage of the resources out there available to you. Best value for your $ are the...”

Leaderboard

1
redever's picture
redever
99.2
2
Secyh62's picture
Secyh62
99.0
3
Betsy Massar's picture
Betsy Massar
99.0
4
BankonBanking's picture
BankonBanking
99.0
5
dosk17's picture
dosk17
98.9
6
DrApeman's picture
DrApeman
98.9
7
kanon's picture
kanon
98.9
8
CompBanker's picture
CompBanker
98.9
9
GameTheory's picture
GameTheory
98.9
10
Jamoldo's picture
Jamoldo
98.8
success
From 10 rejections to 1 dream investment banking internship

“... I believe it was the single biggest reason why I ended up with an offer...”