Estimating Informed Trading with PIN and VPIN

# Can we spot informed trading from market prices?


# Can we spot informed trading from market prices?












Is there any consensus on what is the price behavior in presence of informed trading? Can we observe in retrospect any anomaly in the time series of prices of realized transactions, or transformations thereof? What about considering volumes?

## Answer by Pleb (score 1)

https://quant.stackexchange.com/a/71683

### An extended comment:

As written in Mark Leeds comment, De Prado, et al. (2011) wrote an article that proposed a new method to estimate the presence of informed traders called the Volume-synchronized Probability of Informed Traders (VPIN). This is an extension of the Probability of Informed Traders (PIN) estimator found in Easley, D., et al. (1996)$^\star$. In a nutshell, if more informed traders enter the market, market makers faces adverse selection risk which can lead to higher expected losses, and in turn, they will liquidate their positions and leave the market. Thus liquidity diminishes, the increased participation of informed traders can imbalance the order book and as such, the price of the underlying might see significant change.

Abad, D., & Yagüe, J. (2012) gives a nice summary of the PIN model and the transition to VPIN. In essence, the PIN formula is given by:

$$ PIN = \frac{\alpha \mu}{\alpha \mu + \varepsilon_b+\varepsilon_s}, $$ where denominator denotes the arrival rate of all orders, and the numerator is the arrival rate of informed orders. The respective parameters are estimated using Maximum Likelihood Estimation on a microstructure model proposed in the original article. The authors above also gives an example on how to derive the VPIN estimator.

### Criticism of the estimator:

As remarked in Nbbo2's comment, the VPIN estimator has come under criticism in Andersen, T. G., & Bondarenko, O. (2014) (AB), where they document poor predictive performance of the estimator. This then lead to a response from the original authors in Easley, D., de Prado, M. M. L., & O'Hara, M. (2014), where they try to refute the claims of AB. And of course, the saga continues with a final response from AB commenting on the assertions made in the rejoinder. I have provided links to the papers:



- AB first answer to the original article: Andersen, T. G., & Bondarenko, O. (2014). VPIN and the flash crash.

- De Prado et al. rejoinder to AB: Easley, D., de Prado, M. M. L., & O'Hara, M. (2014). VPIN and the flash crash: A rejoinder.

- AB's responds to de Prado et al. rejoinder: Andersen, T. G., & Bondarenko, O. (2014). Reflecting on the VPIN dispute.

$^\star$ I could not find the reason why "N" is suddenly included in future articles.