# Tag Info

4

Both free and paid access to data sets conatianing company financial statement items is available from Quandl. The free data sets are sourced from the SEC based on compnay electronic filings and go back about five years. For example, you could obtain five years of MSFT's quarterly net income using the R call Quandl("RAYMOND/MSFT_NET_INCOME_Q") Lists of ...

4

Start with http://www1.nyse.com/pdfs/closings.pdf which covers all closings through 2011 then use the following information from official exchange sources to get dates up to present day. 2012/2013: http://www1.nyse.com/press/1294398514465.html Weather related closures happened on Monday, Oct. 29, 2012 and Tuesday, Oct. 30, 2012: ...

4

Surely, there is; search for aggregational gaussianity in Google Scholar or ScienceDirect. In fact, 5 minutes returns are leptokurtic and fat-tailed; then as you increase timeframe, returns become more and more normal. Yearly data is almost normal, if you have enough points.

3

If you're looking for all transactions against any or a given set of securities on whatever exchange, you can get that from a data provider like IQFeed or eSignal. Most of them will have tick level data going back for at least several weeks. Some people have been collecting tick and market data for quite sometime against a variety of securities, and as ...

3

I was just like you when I started out: I had learned a lot about machine learning (mainly neural networks and genetic algorithms/programming) and used it heavily. I also had learned about classic statistics but not nearly as much as about ML. The problem with ML is - as I see it today - that you are often taking a sledgehammer to crack a nut, meaning: ...

3

SEC tends to keep CUSIPS handy: http://www.sec.gov/divisions/investment/13flists.htm

3

My main reference will be "Dan Xu, Christian Beck - Transition from lognormal to chi-square superstatistics for financial time series" Non-equilibrium statistical mechanics (more specifically, superstatistics) gives some ideas of explaining the relation between time frame and its distribution: "...to regard the time series as a superposition of local ...

3

In the long term you will underperform buy & hold because you need an accuracy of at least 65%. See these papers for more: Bauer, R.; Dahlquist, J.: „Market Timing and Roulette Wheels Revisited“, CFA Institute, 2012. http://www.cfapubs.org/doi/pdf/10.2469/irpn.v2012.n1.10 Sharpe, W.: “Likely Gains from Market Timing”, Financial Analysts Journal, ...

3

It's not bad but you have to backtest the method out-of-sample. Say you have discovered an indicator that works 100% in history, you still cannot be sure if it works next time. Another advise is you might want to investigate the distribution of loss when your system fails to work. If your system delivers 1% every time you trade, and loses 10% each time it ...

3

The round-trip latency from point A to a matching engine at point B can be thought of being comprised of two components: $RTT_{total,A \rightarrow B} = RTT_{network\_transit,A \rightarrow B} + MPL_{matching\_engine,B}$ Where $RTT$ is the round-trip time and $MPL$ is the message processing latency (how long it takes to receive a message and produce an ...

3

Most literature focus on comparing fund returns using a model alpha. A good overview is: Cahart (1997) and Berk and Binsbergen (2015). Basically you regress the fund returns on most common used factors (market return, HML, SMB, Liquidity and Momentum factors) and compare alphas after fees.

3

You can use refined methodologies but if you just need a rough estimation of liquidity, you can simply use an average of daily volume over N days. In practice, for equities, people tend to use N = 20 or 30. Once you have the average daily volume (say 100,000 shares), you compare it to your holding (say 50,000 shares) to determine the the size of your ...

3

I think there are a few conflating ideas here. With respect to the sum of logs idea, I think you're thinking about infinitely divisible distributions (https://en.wikipedia.org/wiki/Infinite_divisibility_(probability)). These ideas are indeed used to build more complicated models (i.e. Levy processes) for asset returns. With regards to the Efficient ...

3

Generally Kurtosis measures the degree to which a distribution is more or less peaked than a normal distribution. Positive kurtosis indicates a relatively peaked distribution. Negative kurtosis indicates a relatively flat distribution. In time series we can encounter high kurtosis which is caused by "fat tails" (higher frequencies of outcomes) at the ...

3

Perhaps an answer coming from a different angle and giving you some perspective: The typical approach taken by statistics is top-down: Just looking at the data and finding patterns and stylized facts (like excess volatility, volatility clustering, fat tails, no autocorrelation in returns but significant autocorrelation in absolute returns etc.) The problem ...

3

The dummy function is always used to construct non-linear models. In your model, it is interpreted that the announcements have an non-linear effect on the return. So it is incorrect to say it is a linear regression problem, it should be called as a non-linear regression problem. In total, it means the announcements have asymmetric effects in explaining the ...

2

This site has monhtly sector weighthings data for S&P 500: http://marketcapitalizations.com/historical-data/sp-500-sector-weightings/

2

In Japan we get ISIN data with http://www.isin.org/isin-database they have free search tool.

2

Errors in some data can cause the calculation to go awry. For EPD, I have reported that they believe the stock had a 2:1 split on August 21, 2014 and on August 22, 2014. Only one of these splits occurred, so all the split adjusted data is off by a factor of 2 before the split that did not happen. I reported this error in August, but in November I noticed ...

2

The main reason to use traditional methods is interpretability. Specially when you are dealing with portfolios. Portfolios are nothing more than a linear combination of assets. Many Machine Learning methods are highly non-linear and therefore are hard to replicate with a real portfolio. For example if you want to minimize volatility of your emerging markets ...

2

Quantopian provides both the fundamental data (from Morningstar), as well as the backtest platform to reproduce results from the books you mentioned. Here's the introduction to our fundamentals offering: https://www.quantopian.com/posts/fundamental-data-from-morningstar-now-available-for-backtesting (disclosure: I'm the ceo of quantopian)

2

$\sigma S$ is in units of dollars per square root of a unit of time. $\sigma$ is usually quoted as an annual or daily percentage. $dX ^2$ is in units of time, as $E[(dX)^2] = dt$. Here is an online tutorial which you may find helpful. EDIT by kotozna: $\sigma$ has dimensions 1/(square root of time) and $dX$ has dimensions square root of time. ...

2

Some of the issues with this sort of request is: a) Today's S&P 500 components are not the same from 1 Jul 2013. By using today's components you are introducing pre-inclusion/survivorship bias. Are you going to be able to find data on the delisted stocks? eg. Since 1 Jul 2013, Sprint Corporation, BMC Software, NYSE Euronext, Molex, Life Technologies, ...

2

I faced the same issue some years ago and I solved by implementing the R script reported here; now, with new Yahoo disclaimer rules, it seems to be broken, but, anyway you should be able to replicate the data mining process using that script together with this. If you're pretty confident with R, you should be able to do that. Alternatively, you can visit ...

2

The January effect is easy to demonstrate. Always the same dates for multiple shares.

2

I'll assume the rest of the world doesn't have access to a similar oracle. Indeed if it did future returns would converge to the risk free rate instantly. In this case, I would prefer holding the AAA bond instead of the stock because the rest of the world would consider it to be much less risky. As a financial institution, reducing the risk of your ...

2

As background, Floating point precision is a way of storing numbers such that the precision is relative to the largest digit. For instance, the number $0.00123$ stored in fixed precision needs 6 digits of precision (3 zeros and the 3 non-zero numbers). However, this same number stored as floating point precision $1.23 \cdot 10^{-3}$ needs only 3 ...

2

The easy answer would be to look for exchanges that only have pit trading, ie people in a room that match up buyers and sellers. As far as I know no such exchange exists any more. In my opinion the best you are going to be able to do is to compare the NYSE now with the NYSE in 1998, which is to say you wont be able to do much of a comparison at all as ...

2

The TA_lib Technical Analysis library here has open source code for numerous indicators.

2

Both R and Python can do this very nicely. For Python you would need the pandas package and its dependencies. pandas has a lot of basic statistics, but for more advanced statistics like it looks like you want to do, you can use the statsmodels package, which can work directly with pandas data types. It can also download the csv files directly off the ...

Only top voted, non community-wiki answers of a minimum length are eligible