Show HN: Stock recommender system using hedge fund data and machine learning
navlio.com
navlio.com
The flaw is that Navlio is basing the recommendations on long equity positions (as well as other securities reported in HFs' 13F filings). But hedge funds hedge their bets -- hence the name -- by placing more complicated bets than just buying long, and that includes short sales. You don't have to report short sales in a 13F.
https://www.sec.gov/divisions/investment/13ffaq.htm
If you don't know both sides of HFs' bets, you actually have no idea how they feel about a company. They could have a big long position and a bigger short that's not reported.
It's central to a hedge fund's business to conceal the kind of bets they are making, and what they plan to do, until the precise moment that it benefits them for the rest of the market to pile in.
What's worse in this scenario is not the information itself, but incorrectly assuming the information is unbiased.
Imagine a coin tossing game where I know the result of a toss and you don’t, but where I can set the size of the bets. Every time you choose correctly I set the bet low and every time you choose incorrectly I set the bet high. It won’t take me long to completely fleece you despite the odds of you and me being right being 50/50 each time.
This is a simple analogy of betting against a hedge fund when you only have access to biased information of their strategy and they know this and can use it to induce you to bet high on a losing bet.
Hedge funds, particularly activist funds, do sometimes try to conceal the stocks they buy - usually by using derivatives which have different disclosure requirements. However, it is pretty rare for a hedge fund to be long and short the same stock unless they are doing it for tax reasons - this does not happen often. Going long a stock to create confusion about a short position (which is not disclosed) does not happen in the wild.
We believe that "knowing" in the aggregate just the stocks that managers are long has value. It is noteworthy that equity long/short strategies take a few quarters to materialize so using quarterly data is still valuable (hence we have excluded hedge funds that follow a short term statistical arbitrage strategy with many positions and very high turnover).
I don't understand how any of the suggestions it gave me works and that worries me. It's very discomforting when a site tells you what stocks to pick.
It's important that sites stay technological and give you information rather than actually promise alpha over the market. I don't think you'd have to be a genius to know that anyone telling you what to buy (rather than give you information you can process yourself) and can actually generate you alpha could just use their money instead of taking yours.
The downside of the idea "you can believe someone who back themselves enough to put their own money in" is that your recommendations are then, of course, not truly independent or unbiased.
Most big sell side broker dealers claim there is a chinese wall between their research analysts and their actual trading arm, but sometimes that chinese wall can be suprisingly thin. For example, when a merger is in the early offings, or other large deals about to happen, 'the street' can just go quiet on the subject. Only the (very few) truly independent research firms are still talking at that point.
Having previously been close to non-financial journalism it was a bit of a shock to me when I first realised how little truly independent (in the sense of not trading on their own opinion) news or research there was. But, I now understand why that is.
The optimum situation is probably to have a mix of approaches.
This is, of course, one position for one hedge fund. But, without supporting evidence for listing Micron as a "Hold" for Greenlight instead of "Buy", I'm not inclined to trust Navlio for other recommendations.
https://en.m.wikipedia.org/wiki/D._E._Shaw_%26_Co.
https://en.m.wikipedia.org/wiki/Renaissance_Technologies
Aren't we all reading HN to learn?
https://www.ted.com/talks/jim_simons_a_rare_interview_with_t...
http://www.forbes.com/sites/jaysomaney/2015/08/24/is-our-sto...
The math people (Merton, Scholes) saw the craziness, pulled enough of their money out in time, and didn't lose it. The finance people (first and foremost Merriwether) were those who believed they are smarted than everyone and went bankrupt.
They did not use math at the time they lost money. They were were essentially gambling (doing "risk arbitrage", which is all risk and no arbitrage).
The crazy thing about LTCM, and the thing that everybody overlooks is that LTCM didn't die by bankruptcy...they died because of the systemic risk of their portfolio size and the subsequent intervention from the federal reserve. They had $100B under management, but only lost $5B in 4 months. It was a major catastrophe, sure, but it was only major because of the implications from its absolute size. A 5% drop in 4 months is something that can be recovered from on a personal portfolio level or even a small hedge fund level...but at LTCM's size it was endangering the financial system as a whole because it was threatening to bankrupt overleveraged institutional investors.
Jane Street Capital, while very opaque about what they do, has given me enough room to believe that LTCM has inspired a lot of their strategy. This always makes others perceive me as a quack that doesn't like history, but if another LTCM came around that was using the same strategy but on a much smaller scale and with a well spelled-out liquidation strategy, I would invest. In fact, I would probably bet big on it (>50% of my portfolio).
Nobody is gonna touch it though, because everybody is so conservative in finance. "We're gonna follow LTCM's strategy" isn't a sales pitch, it's a punch line. But outside of finance, there is a ton of precedent for monumental successes with an idea that has failed before. And I think LTCM is one of those ideas.
These guys are trained professionals who've dedicated literally decades to graduate level pure & applied mathematics, and are simply of another a caliber. I can read JAMA all I want, quite fun indeed, but if I practice surgery, I'm certainly going to slice an artery. You have a better chance of going to Atlantic City and putting 100k on black than successfully making money in trading any set of securities, whether its Asian fixed-income, blue chip Wall Street equities, or CME pork belly futures. Just because you back-tested something successfully doesn't mean you should throw your hat into the ring, the sharks will bite till there's nothing left.
I also can't figure out why I get all kinds of "it's really hard" and "be very afraid" comments on HN.
With finance, high voltage electronics, and explosive chemistry, you can have very high cost consequences if you screw up. And people often delude themselves in finance.
There is a also a return aspect in finance. You need a large amount of starting capital until 80hrs/wk of work will outpace what you make 80hrs/wk in your job.
For example, lets say you make $250k/yr as a smart engineer. We assume best case scenarios and say you make %10/APR (very generous) over the index fund return. You would need $2.5million in starting capital before you can start matching your working self in returns. Most people don't have $2.5 million lying around. You would get better returns starting a business or investing in yourself and leaving your savings into index funds.
Maybe I don't want to invest but simply want to understand how math is used to make trades without human decision making involved once the algorithms are written.
I agree that it's an excellent choice though -- easy diversification, very low expenses, low turnover. VT (which includes foreign stocks) is also worth considering for extra diversification.
- Remember that the money that you make on this is taxed as income, which stinks for high income individuals
- The alternative is to set up a lendingclub IRA and put money in there, but now you're somewhat limited as to how much you can invest, as well as being ineligible if you've got an employer retirement vehicle like a 401k
- The return is very high at the beginning, and then tends to drop as more and more people default (or pay off early, which can be almost as bad from an investment perspective)
- It takes work - it's interesting if you have the time and inclination to pick out the notes you want, or they have an automated system - I did a bit of both
In the end, I'm making about 6.5% a year on it - not terrible, but at tax time every year it sucks when I'm getting charged state and federal income taxes on that 6.5%.
I think it makes a lot of sense for folks who are not concerned about reducing their taxable income and want a more active investment option than just vanguard ETFs, or who want an IRA option with higher returns. But it's not as awesome as it seems at first glance.
It's nice in theory, however this ETF strategy is extremely overcrowded, hence the 2008 meltdown. Hence, the August meltdown. Anything that works will eventually, come to an end... Let's not forget the people that make the guaranteed money in the ETF are the one's selling you them. (Might want to check your ETF fees as well)
In the meantime... I created this http://www.strategic-options.com/trade/
>> In the meantime... I created this http://www.strategic-options.com/trade/
Sorry if this sounds rude, but this comment comes across as disingenuous to me.
http://www.investmentnews.com/article/20150911/FREE/15091997...
You also have tulips to blame in 1637. https://en.wikipedia.org/wiki/Tulip_mania
(This is a genuine question, not a snarky one. I've always wondered would write a "how to get rich quick" book in my early teenage years, only to realize their methodology to get rich was to write that get rich book.)
ETFs are just a way of buying equity, not a market that can be overcrowded.
The 2008 meltdown happened because of subprime mortgages, the approximate opposite of conservative index fund investments.
ETFs are (in this context) just a convenient way of buying modest amounts of an index fund.
Index fund companies like Vanguard charge a very low, extremely reasonable commission for both direct purchases and ETFs.
Best of luck!
That's a hell of a thing. If you can make those kinds of returns why do you even need to play in the $50/mo market? Seems like the kind of thing you can sell for millions, or not sell and make billions.
Someone hears that Warren Buffet speak on CNBC, that you should invest in an index strategy. So let it be written, so let be bought. This keeps people warm at night, because Uncle Warren said something that you can understand.
Anybody that is critical of your buy an ETF strategy, is the enemy. But yet you reference a one line strategy and now you are genius! Where was your ETF strategy in 2008? Negative -36% return? How about early 2000? It would still be in the crapper hadn't Uncle Ben started printing money in 2009. But hey, you don't need to understand anything about interest rates, money supply, high frequency trading and or federal reserve, because at the end of the day you saw something on CNBC!
I'm more than happy to encourage buy more of your investment philosophy. Keep buying, buy the dip, buy when it goes up, buy when it goes down. Because the market is up ~250% since 2009 and it can only go up. Economic cycles those are a thing of the past, the market is only going to trend up, no inflation or deflation in sight. It's now clear you know what you doing! Your incredibly intelligent, successful, profitable, secret strategy. Remember not tell anyone, well because they might overcrowd the trade and reduce profitability.
Let us never forget "Ignorance is Strength"
If you're only in a US market ETF there's some non-neglible chance you'll be flat or down in 10 years esp. if you're unlucky enough to buy close to a bubble top. So long term in this context should read as > ~20 years.
EDIT: If you look at buying a 10 year bond right now you get ~2.2% interest per year guaranteed today. If you buy the total US stock index it's hard to tell with any degree of certainty given today's high P/E ratios that the stock market will beat that. It might. But it's more risky. Hence there's a premium you're paying for risk. It's true that yields are likely (but not certain) to go up and so over time your new investments + rebalancing will buy you into those higher yields. Otherwise you're just trying to time the bond and stock markets which isn't simple, you can play a little of timing game (e.g. saying you think stocks are cheap and bonds are expensive today) but you can't usually predict better than the market already does.
EDIT: And while we're talking about this it helps to recall that there's a continuum that goes beyond 100% stocks. If you're so sure interest rates are going up and stocks are going up then you can borrow money at today's rates and invest it in stocks. I.e. use leverage. If you're right your returns will be much higher...
How so?
Will I make more or less than I would under your strategy? I don't know, but averaged over all possible universes it's unlikely to be significantly less, and this way I at least get the fun of being in control, and deciding where to put each new chunk of money.
Speaking of your portfolio, if you're investing in "a few" companies, your volatility and risk are going to be enormous. Especially if you're not balancing your portfolio between industries, company sizes, and less risky investments. Automatic rebalancing is a fantastic way to obey "buy low, sell high", as you intrinsically buy more of cheaper instruments and less of expensive ones to maintain that balancing.
By "a few" I mean more like a dozen.
https://en.wikipedia.org/wiki/The_Superinvestors_of_Graham-a...
Whenever it comes to investing advice, I understand how my family must feel when their computers break.
Don't worry, it's perfectly fine to not know what these things are.
Maybe you haven't been working for long enough (or at the right companies) but I think people usually pick up an interest for this kind of stuff when they want to know what to do with RSU (Restricted Stock Units) or Options that they've gotten from a job. Some also get interested by the idea of having your money work for you (to make more money), instead of just spending money on things that don't bring return.
Generally a mutual fund's value (called it's Net Asset Value or NAV) is calculated once at the end of each trading day. When you put money into a mutual fund, you buy at that price. An ETF[3] is a special type of mutual fund that is itself listed on a stock exchange. When you buy an ETF you're still buying a share of the underlying mutual fund, but the price changes throughout the trading day. The difference between the price you pay and the NAV of the underlying fund is called the premium (if you pay more than NAV) or discount (if you pay less).
You don't have to call Vanguard if you want to open an individual, joint, or retirement account in the United States. You just go to their Open An Account page[4] and select which account you want to start.
[1]: a company organized to aggregate many people's money and invest it in a particular way
[2]: a list of companies that share some quality. For example, the "S&P 500" is a list of the 500 biggest publicly-traded companies in the United States.
[3]: Exchange Traded Fund
There is always an aspect of market timing, even with ETFs. Looking at the massive bull run in the last 5 years, it certainly makes me nervous to enter the market right now, especially with the recent volatility.
Tell navlio the companies you like and it will apply machine learning to match your views with like minded fund managers and predict what else you might like and dislike.
They are not trying to pick stocks for you that they claim have high alpha, just use machine learning to match your "style" with money managers (presumably using data from mandatory reporting by money managers).
Is this going to raise the returns of the average site user? Probably not. Does it help people to do invest their money according to their own style or beliefs (wise or otherwise)? Yes, at least according to their claims. Is it a net gain for society or something that deserves attention? In my opinion, no, people should stick to the basics (i.e. index funds).
I would suggest trying to backtest with quantopian.com to start. If you are able to generate alpha (which it will tell you), you could try entering their monthly contests where they give you $100,000 to play with. Or you could just go talk to investors, up to you.
We ourselves used to track the number of "smart money" investors in biotech stocks like BLUE (bluebird bio) in Excel spreadsheets. The motivation for Navlio was to make this easier and more functional. Using BLUE as an example, the number of "smart money" investors increased from 2 to 10 before its stock price increased over 6x. We used Navlio to find other stocks like BLUE that hedge funds were buying and invested in some of those as well. If you want to actively buy stocks, this is a way to complement your research.
Any suggestion would be researched before being adopted, so I don't think a failure would be disastrous.
-Enter few symbols (including ETFs), for which you are interested
-show actual data you have for them (like who bough & sold)
-Show your ML predictions
-Show related stocks/ETF to balance out portfolio
But this is such a horrible idea I think it needs to be shot down and shut down. And I'm obviously not talking about anything related to the technology behind the site.
If you think for a half second that millions of dollars of Amazon instances aren't doing this, for real, and for better... then you shouldn't be in the market at all.
Taking this repeatedly-failed concept from consumer nonsense and applying it to financial products is just idiotic.
Moreover, it's OLD data. It's annoying enough when I buy a four pack of 9V batteries for my smoke detectors and Amazon tells me that other people bought C cells and Tickle Me Elmo. Gee thanks!
But this site is the equivalent of buying a winter jacket and having Amazon tell me that other people are buying swimsuits.
Bad site, bad idea -- and as, usual -- pandering, platitude financial advice at the top of a HN comments thread in a thread about real issues that affect real people.
Yahoo Finance had "related companies" 15 years ago and it's equally useful. Why does this site even exist? To get karma points? To get bought by some C-list online investment company to help them sell their unethical products?
Real hedge funds (and even casual nerd investors) have stuff that makes this look like the child's toy that it is. And every financial site has moved into "managed portfolios" for a reason. Fidelity, Schwab -- even Vanguard.
Or you could use this random blue site and hope their Ruby scripts didn't choke scraping the SEC website. Your money, your choice I guess.