"I calculate the motion of heavenly bodies, but not the madness of people."

-Sir Isaac Newton

ARMA and ARCH for Forecast

Train an ARMA(p,q) model to forecast returns and an ARCH(d) model to forecast volatility. Use these values to get a probability of positive return. Search over bounds b=0.5,...,1 such that one buys if and only if the predicted probability of positive return exceeds b.

[Note]



Random Utility with Experimentation

Many stochastic choice datasets have no random utility representation. I expand the maximal support of the rationalizing distribution from the set of all rational choice functions to include irrational choice functions that are a certain number of 'mistakes' away from the rational ones. Choice functions with more 'irrational weight' can be demed more consistent with experimentation.

[Note]



Measuring Risk Aversion using Sharpe Ratios

Can we determine how risk averse a portfolio manager is by looking at their portfolio alone? Step 1, use the Sharpe Ratio as a benchmark and project the portfolio onto the tangency line. Step 2, find the risk aversion paramater that makes this projected portfolio the arg max of the Markowitz utility function.

[Note]



Strategic Abstention from the Office Holder

Most political accountability models assume that the incumbent will run again. Yet, most governments prevent incumbents from running for more than two terms. This note proposes a model that allows the incumbent to abstain from re-election immediately in order to seek re-election later. The model involves private information about a shock that affects the cost of being an office holder in the future.

[Note]