Nuclear Phynance
http://www.nuclearphynance.com/
The underground quantitative finance forum.en-us5NuclearPhynance Logo14047http://www.nuclearphynance.com/RSS/images/Nuclear%20Phynance%20RSS%20Feed%20Logo.png
http://www.nuclearphynance.com
Nuclear Phynance Reactor CoreTCS PhD student looking for advice on internships(not sure whether this belongs here or in the University forum - please move if necessary)<br><br>I'm a first-year theoretical CS PhD student at Berkeley working on complexity (lower bounds) and high-performance computing (mostly parallel linear algebra, cache-oblivious algorithms, and the like) looking for advice on internships.<br><br>I'd like to spend at least one summer of my PhD doing an internship in quant finance to get an idea of what working in the field is like. I'm interested in more math and research heavy roles, and much less so in implementation and infrastructure building (although I am decent at programming and have implemented some nontrivial multithreaded systems in the past), but I have no formal economics or finance education or experience.<br><br>I'm wondering:<br><br>- whether the coming summer is a good time to do one, or whether I should wait until a later summer to optimize my chances at getting an internship that reflects the field accurately<br><br>- what firms would be interested in someone with my background (Renaissance, 2Sigma, Jane Street, and DE Shaw are places that people in my research community tend to mention, but I would also would be interested in smaller firms, and Renaissance doesn't seem to offer internships).<br><br>- what reading would help prepare myself for such a role (<a href="https://www.quantstart.com/articles/Quantitative-Finance-Reading-List">Quantstart's reading list</a> has a lot of recommendations and I'm not sure what to start with, outside of the general questions list - I don't think I can go through more than four or five textbooks in any amount of depth before the summer).<br><br>Thanks!
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178244
NPCareersMon, 05 Oct 2015 08:21:38 GMTRecommendation on books with source codes for calibrationHi to all,<br>Can someone point to me beginner or intermediate books with working source codes examples for calibration of short rates? Preferably books that actually delicate to providing the advantages or disadvantages of the different algorithms (eg. Levenberg-Marquardt vs other algorithms).<br><br>Thanks!
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178177
xluikBooks & PapersThu, 01 Oct 2015 07:04:20 GMTMean Variance Portfolio Optimization This might be too basic a question for this forum and for that I apologize in advance. <br><br>I am keen to build a good mean variance optimizer in R and am struggling to figure out the best way to get started. Are there any good books you can recommend which cover not just the theory of MPT & Mean Variance Optimization but the practical aspects as well. I have bought Portfolio Optimization with RMetrics book, which is a good source for following the steps to build the tool but I don’t think will give a good foundation.<br><br>I have looked through a few books such as Robust Portfolio Optimisation & Management by Fabozzi and Advances in portfolio construction by Satchell but there seems to be many books out there. I would greatly appreciate if somebody could guide me on what be the best book to commit to?<br><br>Also, are there good open source optimizers, I could use to test the workings of anything I build myself?<br>
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178175
rohanc81BasicsThu, 01 Oct 2015 04:51:18 GMTHelp with advertising a position our company Hello, <br><br>We recently tried to advertise <a href="https://docs.google.com/document/d/1ACbjkC8bP8PPUbXNDelCsL2QDfGEhc03TZQpEvRv-W4/edit">this</a> position we need filling at our company however I was banned as a user from the site. <br><br>The users of this community seem like perfect candidates for such a position so I am wondering how I can post here without getting banned ? <br><br>Regards <br><br>Dave <br><br>dave<image SRC="/skins/default/images/emoticons/@ symbol.gif" ALIGN="middle">sheffieldcrypto<image SRC="/skins/default/images/emoticons/Period symbol.gif" ALIGN="middle">com<br>www.sheffieldcrypto.com
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178162
Sheffield_CryptoOff-TopicWed, 30 Sep 2015 10:15:20 GMTGolf balls in a school bus...I've just got back from a recruiting trip to Penn. First, Philly has got much nicer. This is mainly like how a toilet gets nicer when you flush it, but still at least it hasn't got worse...<br><br>But more importantly I was talking to a recruiter from Google and he said they no longer use the brainteaser type questions that they were famous for. I think i had heard this but I wasn't aware that they had done a study and proved that the questions had no correlation at all with subsequent ability to do the job well.<br><br>But as far as I know, most trading firms still use these. Does anyone think they are indicative of trading ability?<br><br>I'm going to say know. They might indicate intelligence but I think the correlation between this type of intellect and trading skill is fairly negligible.<br><br>Do people here agree?
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178159
filthyCareersWed, 30 Sep 2015 01:06:05 GMTTrading System Architect Needed - #MachineLearning #CryptoCurrency Trading system architect needed - Work with advanced predictive models <br><br>Introduction <br>We are <a target="_new" href="http://sheffieldcrypto.com/">SheffieldCrypto</a>, we specialise in developing cutting edge predictive models. These ‘smart models’ utilize machine learning and deep neural networks, they are constantly learning and evolving given new market conditions. Currently we are applying them to various digital currency markets, the results have been very promising. We require a financial expert, specifically an expert in trading systems, to design and implement a system to utilise the signals produced by our models. <br><br>Job Description <br>Design and implementation of a semi-automated/automated trading system utilizing the signals produced by our models. <br>Custom indicator/feature design <br>Implementing the infrastructure connecting the trading system to crypto exchanges.<br><br>Person Specification should be experienced or knowledgeable in <br>Algorithmic trading <br>Machine learning <br>Natural Language processing<br>Market analytics <br>Hedge fund operation <br>System Architecture <br>Feature/custom indicator design<br>Agile software development (CI, TDD, Scrum)<br>Python<br><br>Additional Info<br>We need the position filled asap <br>We will be offering equity in the company for the filled position<br> We are not offering a salary<br><br><br>Model Information and Performance<br>On one simulation we tested our TA (technical analysis) algorithm with around 30 coins with shared initial parameters over a 3 month period, however many of these coins had little data (small market cap/volume), the results were as followed:<br><br>Start capital: $5,000<br>TA trader: $12,706<br>Buy and Hold: $-731.79<br>Random $-1,145<br>Buy and hold and random are included as comparisons ie how a portfolio would have performed if one was to 'buy and hold' the commodity<br><br>Another model being developed is based on NLP (natural language processing). This relies on input from social media or ‘text data’, such bitcointalk threads, reddit, twitter etc. The semantical content is analysed and patterns seen in the data are correlated to price movements. We are currently working on combining the output from both algos into a deep neural network.<br><br>Final words<br>We have 2 PhD’s on our team developing these models. The models collectively have had 6 years of development (3 years/ PhD) and we are now ready to implement them into a trading scenario.<br><br>If you feel you are suitable for the position please apply via the <a target="_new" href="http://sheffieldcrypto.com/recruitment ">application form</a><br><br><br>Yours sincerely <br>Dave - marketing and networking <br>dave<image SRC="/skins/default/images/emoticons/@ symbol.gif" ALIGN="middle">sheffieldcrypto<image SRC="/skins/default/images/emoticons/Period symbol.gif" ALIGN="middle">com<br>www.sheffieldcrypto.com<br> <br><br><br><br><br><br><br><br>
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178142
SheffieldCryptoGeneralTue, 29 Sep 2015 06:20:32 GMTForecasting Daily Volume ?Dear colleagues :<br><br>For portfolio construction doing some forecast of trading costs, and was wondering what are good methods or models to forecast daily trading volume ?<br><br>Seems a lot of attention is given to forecasting volatility, which is only a part of trade cost estimation, but what about the volume ?<br><br>Any suggestions/advice or papers will be much appreciated.<br><br>Thanks in advance,<br><br>Lexx
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178134
lexxTradingMon, 28 Sep 2015 19:36:06 GMToptions and borrow costHi,<br><br>I am trying to calibrate the borrow cost to some options (using call-put parity), but I end up having a borrow cost which will depend on the strike of the options (which is wrong obviously).<br><br>I have for example used NFLX or GOOGL options (in order to remove divs uncertaincy) and to use some good old BS formula.<br><br>I have taken care of using the Libor curve for the fwd and the OIS curve for the discount, but I still end up with this strike dependency.<br><br> Bid Ask Mid<br>250 0.92% -0.44% 0.24%<br>255 0.84% -0.03% 0.40%<br>260 0.93% 0.77% 0.85%<br>265 0.45% 0.77% 0.61%<br><br>The result is also dependent of the use of the Bid/Ask/Mid quotes.<br><br>Do I miss something or is there a good hidden secret of the industry for this?<br><br>Thanks<br><br>
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178124
LapinPricing & ModellingMon, 28 Sep 2015 15:11:59 GMTLaTeXI have a long equation that I am trying to wrap across a number of line in LaTeX but I can't seem get the \\ to break the equation into multiple lines.<br><br>\fbox{<br> \addtolength{\linewidth}{-2\fboxsep}%<br> \addtolength{\linewidth}{-2\fboxrule}%<br> \begin{minipage}{\linewidth}<br> \begin{equation}<br> P=\left(1+\frac{Y}{F}\right)^{-DF} \cdot<br> \left\{<br> \frac{C(1-TX)}{F} \cdot FCF+\frac{C(1-TX)}{Y}<br> \Bigg[ 1-\frac{1}{ \Big( 1 + \frac{Y}{F} \Big)^n} \Bigg] + \\<br> \Big[ RV-TXG(RV1-P0)+\frac{C(1-TX)}{Y} \Big] \nonumber \\<br> \cdot \left( 1+\frac{Y}{F} \right)^{-n-LDF}<br> \right\}<br> -\frac{C(1-TX)}{F} \cdot ATF<br> \end{equation}<br> \end{minipage}<br>}<br><br>Can anyone see the issue?<br>
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178103
dgn2SoftwareSun, 27 Sep 2015 19:35:44 GMTMathematical Expression Parsing and treessome advice needed.<br><br>There are lots of math expression parsers out there (for C#) but I have a fairly specific need. Also, seems like a lot of the available parsers end up compiling the expression, which I guess speeds things up but for now I want to try to avoid that approach.<br><br>I built a generic computational tree class (where the nodes are arbitrary functions that output objects of type T). In the specific case where T=double, I have a special type of tree in which the nodes (and specifically the leaves) are Indicators that automatically update given the state of the market (it's part of an algo trading system i've been working on). Problem is, I can set up config files to load the trees but it's very messy and unnatural; for example one could do xml files whose tree structure matches that of the tree, but what a lame way to go.<br><br>I'd rather be able to do the following (for sake of simplicity I'm making the function up randomly):<br><br>x0="MidPriceIndicator"<br>x1="BidAskIndicator"<br>string bla="Log(x0)+Exp(x1)"<br><br>I have code for grabbing the MidPriceIndicator and the BidAskIndicator from some container that has a dictionary of all the indicators.<br><br>Tree would have either a constructor or parse function that parses bla to have "SUM" as a parent node, two child nodes corresponding to Log and Exp, and respectively two children of those corresponding respectively to the MidPriceIndicator an the BidAskIndicator.<br><br>So, I'm looking for a parser that will effectively return the treenodes. <br><br>if anyone knows a good parser that can do the above, much appreciated for the link.<br><br>
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178094
NoniusSoftwareSat, 26 Sep 2015 20:13:05 GMTSystematic Trading BlogPlease check out my new blog site<br>http://mintegration.eu/<br>
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178090
nkirkBooks & PapersSat, 26 Sep 2015 15:19:33 GMTWriting to investorsI think most investment managers agree (including many famous ones) that writing quarterly/whatever letters is one of the most important things they do. Buffett and Munger are strong examples of having mastered a certain style (beginning with the letter "f"). But I also see M El-Erian as a master of the form.<br><br>John Macaskill <a target="_new" href="http://www.euromoney.com/Article/3488597/Macaskill-on-markets-Reasons-to-be-fearfulthe-big-three.html">described</a> Bill Gross' style (which I think matches El-Erian's on this dimension) as having "a typically meandering prelude". The commonalities I'm noticing include a certain "college-y-ness" — without being actually intellectually intimidating or sophisticated (compare New Yorker or LRB, whose authors may run shoulders with the money men at art auctions); absolute conformity to standard good style (strunk & white) — may he interpretable as heavy proof reading—; and a human touch that may actually be sincere, briefly kissing up against the literary author's stock in trade—emotionally resonant honesty.<br><br>Notably absent : red meat / finbro jock / models & bottles / sexist crap, objectivism, …… the way to score "easy points" appears to me to be at least one level of remove of obviousness, but not too many meta levels up. A bit deep, although certainly not too deep (I wouldn't want Susan Sontag to push herself if I hired her as my speechwriter).<br><br>These are just some thoughts I've passively accumulated. Hope they're interesting to someone here, or spur others who have smarter things to day to chime in.
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178076
RashomonGeneralFri, 25 Sep 2015 21:54:56 GMT"linear" transaction costsNeed a bit of advice or reference articles.<br><br>When I had developed an equity stat arb model, I was lucky to have all the transaction costs except for market impact fixed (actually there was an impact cost formula as well...). that was to say, the algo continuously streamed order requests to another execution algo (which actually aggregated all the firm's hi to med freq liquidity taking algos), and there was an agreement that the execution algo would execute at a fixed price (mid + X) where X should handle exchange fees and bid ask. in this context, this paper was fairly useful (<a target="_new" href="http://arxiv.org/pdf/1203.5957.pdf">potter et al</a>)<br><br>I'm now testing a similar algo but on a different asset class and I don't have the luxury of that fixed cost. The algo is producing decent PL, but the vol and drawdown are too high. It's a fairly strong mean reverting signal. what I'm realising is that the bid-ask is pretty damn volatile, such that if one sets a threshold (let's make it simple and say current Bid Ask +exchange fees), then it could fire off a trade when the signal is above that threshold and then the order direction would flip to the other side when the signal flips signs. (I'm simplifying because I'm using methods to dribble out the orders to reduce impact) I think you see what I'm getting at. But, when the signal flips signs, the threshold could be greater than it was before. Thus, if there is a temporary widening of bid ask, the PL could drawdown. In the long run the bid ask doesn't go to infinity, so it's ok, it may mean revert then jump to a new quantum state, but it seems to fair ok over 5 years of tick data backtesting. but too high vol, too high drawdown given the fact that this particular model is HFT.<br><br>So, what to do? I would think the idea would be to embed the stochastic nature of the bid ask into the prediction signal itself, so, if my predictor is P and my net reward is:<br><br>P- threshold if P>0<br>P+ threshold if P<0<br><br>was thinking of defininng <br><br>Q= P-bidask if P>0<br>Q= P+bidask if P<0<br><br>Then, play the same game but Q being predictor and one has a fixed threshold cost= exhange fees. Then apply the same principles in the cited paper above.<br><br>But before I go down that road, which will be a bit complicated modelling bidask dynamics (which will be very noisy and subject to sharp jumps), was wondering if people had some recommendations on reading or general ideas.<br><br>Thanks.<br><br>
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178064
NoniusTradingFri, 25 Sep 2015 12:09:08 GMTAdvanced derivatives valuation course/workshop anyone?Hi,<br><br>I have been looking for an advanced course or workshop on derivatives valuation that is somewhat applied (maybe delivered by a practitioner) but leaves out the math/stats primers and intros to Black Scholes. I'd love to see it use MATLAB, Python or similar (but not a necessity) and delve into fixed income derivatives valuation/trading (SABR and/or market models) or volatility modelling (local(,) stochastic volatility).<br><br>I have considered CQF but deemed it a little too broad and covering too much basics with respect to math/statistics. I have already searched for this on and off for two weeks, so any pointers would be appreciated!<br><br>Maybe you know that certain experts offer courses/workshops or know a site on which I can find announcements for such?
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178063
jokkCareersFri, 25 Sep 2015 09:10:05 GMTLooking at Europe: Germany, Switzerland and the UKHi,<br><br>I am interested in getting a handle on the job situation in quant finance (whatever that means) in Germany and Switzerland, possibly the UK. <br><br>What is it like there? Is it like here in the tri-state area?<br><br>Can anyone recommend a few good recruiters please? PM me if you want to please.<br><br>Your help would be much appreciated.<br><br>Thanks,<br>M
http://www.nuclearphynance.com/Show%20Post.aspx?PostIDKey=178050
mtsmCareersThu, 24 Sep 2015 13:50:28 GMT