Skip to main content

Order flow risk measures in Python

Project description

Order Flow Risk Measures

Has been updated to support Python 3.8. Currently, the packages only has VPIN.


The default way is to open a console and execute

pip install flowrisk

One may also download from here and manually install

git clone
cd flowrisk
python install


To implement VPIN, we made

1.  an EWMA estimator of volatility (RecursiveEWMAVol)
2.  a numpy.ndarray based buckets with bulk classification of volumes in the MA style (RecursiveBulkClassMABuckets)
3.  a numpy.ndarray based buckets with bulk classification of volumes in the EWMA style 
4.  a recursive VPIN estimator (RecursiveVPIN)
5.  a recursive VPIN estimator with VPIN confidence intervals (RecursiveConfVPIN)
6.  a recursive model using an EWMA estimator of means and RecursiveEWMAVol, for modeling and log 
    innovations of VPINs and for calculating VPINs' confidence intervals (RecursiveEWMABand)
7.  a one-shoot VPIN estimator for a series of prices (BulkVPIN)
8.  a one-shoot VPIN estimator for a series of prices with VPIN confidence intervals (BulkConfVPIN)
9.  various configuration classes (RecursiveVPINConfig, RecursiveConfVPINConfig, BulkVPINConfig, 

For illustration, we also put the 1-min data of five small caps (CBIO, FBNC, GNC NDLS, QES) and five large caps (V, AAPL, NVDA, GS, INTC) from the US stock market. The data covers Nov 12 to Nov 21, 2018. The data can used by, for example,

import flowrisk as fr

class Config(fr.BulkConfVPINConfig):    

config = Config()

example = fr.examples.USStocks(config)
symbols = example.list_symbols('small')
result = example.estimate_vpin_and_conf_interval(symbols[0])


The piece of the code will automatically calculate VPINs and associated confidence intervals of GNC. We also put prices and volumes together with them into a nice picture, which is saved to ./pics/gnc.png by default. Note that the calculation of VPINs is fast, but making nice pictures is slow. One may also find out more in

Note that there are several differences between this implementation and the original paper:

Easley, D., L贸pez de Prado, M. M., & O'Hara, M. (2012). Flow toxicity and liquidity in a high-frequency world. 
The Review of Financial Studies, 25(5), 1457-1493.

For example,

1.  we use an EWMA estimator for the volatility of PnLs, instead of using all samples for estimating the PnL 
    volatility; and
2.  VPINs are calculated from the very beginning, instead of after a certain number of buckets have been filled.

We made the differences because the core of our package is a recursive estimator of VPIN.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

flowrisk-0.3.0.tar.gz (2.8 MB view hashes)

Uploaded source

Built Distribution

flowrisk-0.3.0-py3-none-any.whl (276.3 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page