Category Archives: financial markets

New release of Bloomberg-to-Mathematica code

A new release of my code to move data from Bloomberg to Mathematica. The big changes here include

  • rolling all four functions (current data, historical data, intraday data, and intraday bar data) into a single executable. This reduces the number of connections needed by Mathematica among other advantages,
  • standardization of function names,
  • addition of DPDF code to intraday functions, allowing the user to specify whether or not data should be adjusted for dividends, and
  • it has been compiled for the first time for x64 as well as x86.

This interface, its source code, and compiled binaries are freely available at
https://github.com/MichaelSternNYC/

Example:

x1 = msBBGhistory["IBM US Equity", "Px_Close_1D", {2015, 1, 1}, "", "Day", "use DPDF" -> False];
x2 = msBBGhistory["IBM US Equity", "Px_Close_1D", {2015, 1, 1}, "", "Day", "use DPDF" -> True];
DateListPlot[{Callout[x1/x1["FirstValue"], "without dividends"], Callout[x2/x2["FirstValue"], "with dividends"]}, PlotStyle -> {Red, Blue}, PlotLabel -> "Historical IBM Returns", Frame -> {True, True, False, False}, ImageSize -> Large]

Graph of normalized IBM prices demonstrating dividend adjustment and Callouts.

Bloomberg-to-Mathematica interface now supports overrides

Reference data requests made via the Bloomberg API can be simple (“current price of IBM Equity”) or complex (“trailing 12 month beta between AAPL US Equity and the Nasdaq index”). To send those more complex requests via external software requires overriding Blomberg’s default fields, something the Bloomberg-to-Mathematica interface did not allow. Until now.

moStarts = DateRange[{1990, 1, 1}, {2016, 4, 1}, "Month"];
moToInclude = 12;
corr1 = Table[{moStarts[[i + moToInclude - 1]], msGetBBG["SPX Index", "CORR_COEF", "noisy" -> False, "overrides" -> <|"BETA_OVERRIDE_REL_INDEX" -> "NDX Index", "BETA_OVERRIDE_START_DT" -> bbgifyDate[moStarts[[i]]], "BETA_OVERRIDE_END_DT" -> bbgifyDate[moStarts[[i + moToInclude - 1]]]|>]}, {i, 1, Length[moStarts] - moToInclude}];
corr2 = Table[{moStarts[[i + moToInclude - 1]], msGetBBG["SPX Index", "CORR_COEF", "noisy" -> False, "overrides" -> <|"BETA_OVERRIDE_REL_INDEX" -> "USGG10YR Index", "BETA_OVERRIDE_START_DT" -> bbgifyDate[moStarts[[i]]], "BETA_OVERRIDE_END_DT" -> bbgifyDate[moStarts[[i + moToInclude - 1]]]|>]}, {i, 1, Length[moStarts] - moToInclude}];

DateListPlot[{corr1, corr2}, PlotLabel -> "Trailing 12-Month Correlations", Frame -> {True, True, False, False}, PlotStyle -> {Blue, Purple}, Epilog -> {(Text[Style["SPX vs. NDX", FontSize -> 10, Blue], {{2010, 9, 2}, 1.01}]), (Text[Style["SPX vs. U.S. 10 Yr Rates", FontSize -> 10, Purple], {{2007, 3, 2}, .65}])}, ImageSize -> Large, Axes -> True]

Correlation of SPX vs. various things over time

The interface and source code can be found at GitHub

More revisions to the Bloomberg-to-Mathematica code

It can now take multiple simultaneous securities requests for current data, for a speedup of about 20x. You can download the WSTP conduits and source code at Github. https://github.com/MichaelSternNYC Sample below with some of The Economist‘s Big Mac Indices.


indices = {"BIGMUS", "BIGMBZ", "BIGMMX"};

data = Map[
msBBGgetHistory[# <> " Index", "Px_Last", {1997, 1, 1}, "", "Monthly"] &, indices];

DateListPlot[data, PlotStyle -> {Red, Blue, Purple}, PlotRange -> {1.1, 6.4}, Frame -> {True, True, False, False}, PlotLabel -> "Global Big Mac Prices (USD)", Epilog -> {(Text[Style["U.S.", FontFamily -> "Calibri", FontSize -> 10, Red], {{2007, 1, 10, 16}, 3.4}]), (Text[Style["Brazil", FontFamily -> "Calibri", FontSize -> 10, Blue], {{2007, 5, 10, 16}, 4}]), (Text[Style["Mexico", FontFamily -> "Calibri", FontSize -> 10, Purple], {{2007, 1, 10, 16}, 2.3}])}, ImageSize -> Large];

big macs

New release of the Bloomberg-to-Mathematica link

Those who use Bloomberg and Mathematica should find this useful. This interface allows the direct loading of Bloomberg data (historical prices, bulk data like dividends, current data like stock prices, etc.) into Mathematica. I’ve made the interface freely available since 2008, but this release represents the largest changes in over six years. Most importantly, it can now handle bulk data like current holders of a security and historical dividends, and it’s much more robust about errors.

Historically I’ve thrown source code in the .zip file with the binaries but starting with this release, the source code and the binaries will be in separate GitHub repositories. See https://github.com/MichaelSternNYC for everything. If you just want to install the links, check out https://github.com/MichaelSternNYC/bloomberg-to-mathematica in particular.


histLink = msBBGgetHistoryMakeLinkObject[];

x1 = msBBGgetHistory["IBM US Equity", "Px_Last", {2014, 1, 1}, "",
"Day", "use DPDF" -> False, LinkObject -> histLink];
x2 = msBBGgetHistory["IBM US Equity", "Px_Last", {2014, 1, 1}, "",
"Day", "use DPDF" -> True, LinkObject -> histLink];
DateListPlot[{x1, x2}, PlotStyle -> {Orange, Purple},
PlotLabel -> "IBM Stock Price via the new interface",
ImageSize -> Large,
Epilog -> {(Text[
Style["not dividend-adjusted", FontFamily -> "Times New Roman",
FontSize -> 10, Orange], {{2015, 1, 22}, 169}]),
(Text[Style["dividend-adjusted", FontFamily -> "Times New Roman",
FontSize -> 10, Purple], {{2015, 1, 22}, 140}])
}]

demo of the new interface

Significant update to the Bloomberg to Mathematica link

I’ve been making significant changes to the Bloomberg-to-Mathematica link recently, including

  1. the ability to specify whether historical data has DPDF adjustments (in other words, whether or not prices have been adjusted for dividends, splits, and other capital events),
  2. the ability to pull bulk data like historical dividend records and current holders, and
  3. the ability to keep the link live between queries, significantly reducing latency.

Other improvements include more sophisticated handling and reporting of errors and more widespread use of modern Mathematica data types (Associations and TimeSeries). I will provide all the source code, perhaps in the form of a GitHub depository. The new and improved functions are based on Wolfram’s current WSTP libraries and the current version of the Bloomberg BLP API. I’ve built to 32-bit architecture but I believe that 64-bit versions should be possible as well.

I’m waiting for Bloomberg to help with one source of crashes (when a company announces a distribution but cancels it before record date). When that’s taken care of I will upload the improved binaries and interfaces, etc. Those who want to play with the source code should visit https://github.com/MichaelSternNYC

[update March 7, 2016: the new executables are also available on Github, at https://github.com/MichaelSternNYC/bloomberg-to-mathematica]