You are using an out of date browser. It may not display this or other websites correctly.
You should upgrade or use an alternative browser.
You should upgrade or use an alternative browser.
Machine designed strategies. Do they work?
- Thread starter v75z52
- Start date
· Through intensive data analysis and creation and use of various models, the Ops Strats team are responsible for:
· Quantifying risks and costs in operations, including failed settlements, collateral disputes, unconfirmed trades, client confidentiality breaches, brokerage fees, money laundering and trader fraud.
· Ops Strats comes into focus post the booking of a trade. The team analyzes the business data, to understand the causes of post-trade problems
· Build models to predict what to expect in the future
· Expand on the usual data collected by the division in order to map costs throughout the lifecycle of a broken trade or operational risk event.
=> From a Goldman post of your link. I say: Go and have a stab, come back and let us all know how your machine learning exercise went in relation to booking trades and analyzing "post trade problems"
LOL
=> Not everything that shines is diamonds. Tags are used for a reason, to draw attention, tags are not job descriptions, I hope this makes it a little clearer
=> By the way, don't take it wrong, but pretty much every serious trader is running for the exit doors re Goldman and other iBanks. Or do you want to develop highly profitable algos and be paid a cash bonus cap of USD 125k? This is the cap at Deutsche right now and others are slightly higher but not far away. This is about the same that you pay to have a piss at a public toilet in Manhattan. Seriously though, nobody serious about trading is aiming for iBanks anymore. That aside, my claim is that there is not much merit in Machine Learning algos nor is it a technique that is paid much attention to on the financial quant front but hey, take my word for what it is, another opinion. I mentioned I have the experience behind it to believe I speak with some level of authority but I won't email you my resume. Take it or leave it.
· Quantifying risks and costs in operations, including failed settlements, collateral disputes, unconfirmed trades, client confidentiality breaches, brokerage fees, money laundering and trader fraud.
· Ops Strats comes into focus post the booking of a trade. The team analyzes the business data, to understand the causes of post-trade problems
· Build models to predict what to expect in the future
· Expand on the usual data collected by the division in order to map costs throughout the lifecycle of a broken trade or operational risk event.
=> From a Goldman post of your link. I say: Go and have a stab, come back and let us all know how your machine learning exercise went in relation to booking trades and analyzing "post trade problems"
LOL
=> Not everything that shines is diamonds. Tags are used for a reason, to draw attention, tags are not job descriptions, I hope this makes it a little clearer
=> By the way, don't take it wrong, but pretty much every serious trader is running for the exit doors re Goldman and other iBanks. Or do you want to develop highly profitable algos and be paid a cash bonus cap of USD 125k? This is the cap at Deutsche right now and others are slightly higher but not far away. This is about the same that you pay to have a piss at a public toilet in Manhattan. Seriously though, nobody serious about trading is aiming for iBanks anymore. That aside, my claim is that there is not much merit in Machine Learning algos nor is it a technique that is paid much attention to on the financial quant front but hey, take my word for what it is, another opinion. I mentioned I have the experience behind it to believe I speak with some level of authority but I won't email you my resume. Take it or leave it.
Quote from braincell:
Having a quick look at some job postings from the past 10 days here:
http://jobs.phds.org/machine-learning-jobs/quantitative-finance
it seems even Goldman Sachs is looking for machine learning experts. Why else would all of them put "Machine Learning Jobs" as one of the tags in the job opening?
The GS ad also has this:
see?
How was I wrong? I said that Mr. Bloomberg worked for ML. Is that wrong? When ML finances a project you work for them. But the serious issue is that you get obsessed literally with such details in a psychotic way and you ignore the main issue which is that your general statement is false because of Mr. Bloomberg.Quote from amazingIndustry:
P.S. You were utterly wrong M. Bloomberg and his association with ML and now you pretend you did not say what you said? Get outa here...
You have no right to ask anyone to get outa here. You sound like a very frustrated individual. I urge you to behave yourself. Acting like an idiot only exposes your weakness in working with other people, which was possibly the reason you lost your job and you are frustrated and take it on other people for insignificant details, especially people who try to educate you so you can get a second chance.
Quote from amazingIndustry:
That aside, my claim is that there is not much merit in Machine Learning algos nor is it a technique that is paid much attention to on the financial quant front but hey, take my word for what it is, another opinion. I mentioned I have the experience behind it to believe I speak with some level of authority but I won't email you my resume. Take it or leave it.
Ok, thanks for a approachingly civil response, but i'm still expecting some reasoning that could rebuke my claim in my earlier post to you regarding the theory i mentioned. If you missed the post, go back and read all of it, you'll see i actually agree with you, but there's a "but" in it, and i'd like to hear your opinion on that too.
he did not work for ML nor did ML finance a "project", ML had no say whatsoever in what BBG did or created, nothing whatsoever beyond client feedback that every user infrequently provides. Here you go, I provided an exit door for you, back to Machine Learning.
And you did not read my initial point carefully at all: I said that no successful trader would develop trading software. And I also said that those with talents such as Zuckerberg should pursue what they were destined to create. Did I not say that? Now, what is your point really other than sneaking around and retracting things you said that are simply plain wrong???
And you did not read my initial point carefully at all: I said that no successful trader would develop trading software. And I also said that those with talents such as Zuckerberg should pursue what they were destined to create. Did I not say that? Now, what is your point really other than sneaking around and retracting things you said that are simply plain wrong???
Quote from alexandermerwe:
How was I wrong? I said that Mr. Bloomberg worked for ML. Is that wrong? When ML finances a project you work for them. But the serious issue is that you get obsessed literally with such details in a psychotic way and you ignore the main issue which is that your general statement is false because of Mr. Bloomberg.
You have no right to ask anyone to get outa here. You sound like a very frustrated individual. I urge you to behave yourself. Acting like an idiot only exposes your weakness in working with other people, which was possibly the reason you lost your job and you are frustrated and take it on other people for insignificant details, especially people who try to educate you so you can get a second chance.
Quote from amazingIndustry:
they mentioned the 180 degree panoramic view of the skyline and then talked about machine learning? Interesting combination. But maybe after the machine learning thing did not work out the only thing that was left to attract new grads to their firm was the mentioning of the spectacular views.
I dont know which hedge fund you talk about as there are several billion+ AUM funds in the Houston greater area. I only spoke of things I claim I know and shut up about other stuff. I am a quant trader and claim to know a bit about the industry and machine learning is definitely not a tool widely used at all nor do I or most other quants see merits in considering it. The famous example of curve fitting (and thats exactly the stuff machine learning algos spit out) is the correlation between changes in butter prices in Bangladesh and s&p 500 index returns. Try to pitch the idea to a sell-side ibank MD and report back.
Edit: Why dont you do a search on NuclearPhynance maybe you find the gems there, at least its (correction: was) a rather professional forum (before questions such as "what is the cheapest broker to trade nasdaq futures" were allowed...recession is a bitch...)
yeah now I remember it was RGM Advisors. It does seem that you have worked in the industry so I am sure you know that RGM Advisors is not one of the billion+ hfs. Its right there in the league of DEShaw, RenTec etc.
Admittedly, I know little about machine learning, but it is a fact that I found RGM looking specifically for Machine learning skills. So, I asked you how sure that RenTec doesn't do Machine learning stuff.
Agree with your point that random correlations and random datamining won't stand a chance. But guys who run RGM or who work there are very smart. Even if they are using genetic programming and other things, they would be using them with utmost caution. The analogy I can offer is when I first learnt about regression in college - it was very superficial. Only in grad school, did I pick up the various assumptions that underline proper regression e.g. multicollenearity et all.
Quote from braincell:
Having a quick look at some job postings from the past 10 days here:
http://jobs.phds.org/machine-learning-jobs/quantitative-finance
it seems even Goldman Sachs is looking for machine learning experts. Why else would all of them put "Machine Learning Jobs" as one of the tags in the job opening?
The GS ad also has this:
see?
The "medical imaging" bit is interesting. That makes sense.
in case you refer to this post (I found no others addressed to me with *buts* I can only say you are a day dreamer if you believe that even 0.5% of the systems such algos spit out are primed to make any money. Dude, I seriously challenge you because I spent way too much time in this area, I can give you an algorithm that generated 100% in sample annualized returns and 30% out of sample and I can promise you that you will not end the the year above the water if you traded it in the market with actual exposure to the order books, liquidity taking and providing dynamics and micro market structure at work. Show me a SINGLE software application that does an excellent job at simulating a limit order book and I pay you 500,000 USD easily (because I could instantaneously turn around and sell it for a million to any given hedge fund shop). There are none. Literally none. Or do you implement Machine learning algos for long-term trading strategies? If thats the case then good luck dealing with ever changing correlation dynamics and market structure changes which no machine learning algo that was parameterized to historical data can capture nor model.
Quote from braincell:
That right there would be my answer to you. To not re-post my earlier posts i'll say i agree that TSL is being marketed to newbies and that some poor souls end up with a product they don't know how to use - because they lack the knowledge of statistics and a lot of other skills. However, with that knowledge, the fact that 80% of systems they find truly are curvefits (and i believe that), the other 20% that work out-of-sample on a statistically significnt portion of the data may just be an underlying market mechanism that was discovered. It all comes down to statistics, since you can claim there is still 0.05% chance that the system simply got lucky after 1000 good trades OOS, but that's a much lower risk compared to many others taken by traders on a regular basis. Of course, the statistical analysis is probably a major part of the game with such software(and should be undertaken very very seriously ), but the fact that the systems are composed of price derivations (indicators) and patterns which makes it an open white-box, allows you to see if it did in fact catch onto some mean-reversion logic or trend confirmation logic. These things are obvious to see, and a curve-fit system will often use an oscillator in the inverse way or do something that's obviously stupid and you'll know it's a curve-fit even before looking at OOS data.
I also agree that the way TSL is being sold is a minor crime, since the demos and the data presented on the site solely focus on the successul systems (the 20% or even 5% if you want to be a pessimist) that work OOS. Further, TSL has an erroneous way of counting filled limit/stop orders, no slippage, etc, further emphasising "good systems" that aren't really that good. On the newest EVORUN demos you can even see that the majority of systems in the OOS panel are squarely at around 0 net profit, going up and down around that value (if slippage and commission were used, they would mostly be negative). This is expected of curve-fitted and random systems and indicates a very low ratio of "valid" systems. When marketing TSL they don't tell you that. But what ultimately doesn't make TSL a complete fraud is the fact that with good analysis, in theory, you can find systems that have true underlying market dynamics in them. Considering it evolves 100s of systems per second, you're bound to find something after a few weeks of running the data. You can't deny this theory is valid.
As for the poster "milewski05" i got in touch with him via e-mail and checked him out. He's a real person and a real TSL user, i know that for sure.
I can say for sure that the stuff my alumni buddies worked on was utterly unrelated to Machine Learning. I cannot claim that Renaissance does not at all employ such techniques because of the Chinese Wall and high level of secrecy even between different groups according to the info I was given. So, can I prove or did I make a statement that there is not a single fund under the sun that looks at Machine Learning? No, I claimed it is a highly questionable approach to modeling financial algorithms in the same way than Neural Networks are even there are shops that advertise on their front page they employ NN algos. I have never seen such funds nor their sub-funds ever in the top decile, performance wise.
Quote from gmst:
yeah now I remember it was RGM Advisors. It does seem that you have worked in the industry so I am sure you know that RGM Advisors is not one of the billion+ hfs. Its right there in the league of DEShaw, RenTec etc.
Admittedly, I know little about machine learning, but it is a fact that I found RGM looking specifically for Machine learning skills. So, I asked you how sure that RenTec doesn't do Machine learning stuff.
Agree with your point that random correlations and random datamining won't stand a chance. But guys who run RGM or who work there are very smart. Even if they are using genetic programming and other things, they would be using them with utmost caution. The analogy I can offer is when I first learnt about regression in college - it was very superficial. Only in grad school, did I pick up the various assumptions that underline proper regression e.g. multicollenearity et all.
Quote from Rationalize:
The "medical imaging" bit is interesting. That makes sense.
Umm actually, I "think" but I'm not really sure, that certain types of engineers working in medical imaging require working with very large data sets and an ability to analyze them. They also require a high level of knowledge in math an physics (not the technicians that operate machines, the researchers). So there's probably some skill overlap they figure could be useful.
Anyway, any data, even interaction with clients and post-trade analysis can be digitized and correlations found. It's not practical, and that has probably nothing to do with machine learning (and it shouldn't) but i'm just saying - it could be done - machine learning can be applied to anything. lol