In this project, quant analysts Akos Nagy and Dragomir Petrov take a deep-dive into price spikes in the forex market. Our team analyzed millisecond-level price data for multiple currency pairs during the period of 2018-2019. The purpose of the analysis was to find non-randomness in the price movements, that is signalled by a sudden price change (or spike), which could further be exploited by a trading strategy.

Our research proposes a data model that tests the non-randomness by selected spike definitions and algorithmic trading strategies. By definition, a spike event starts when a price change is over a pre-set limit and lasts until another pre-set condition is met. Following this reasoning, a variety of spike definitions exist, of which our work introduces two and tests it on four different trading strategies.

The data model, including definitions and strategies, was developed entirely in python. Data exploration, condition setting and testing were performed using Pandas, while visualization, being an important aspect of technical analysis, was developed using interactive Plotly charts.

For the investigated period, several thousand spike phenomena was identified; however, the trades, that were done by the algorithm, resulted in losses in more than 50% of the cases. Although test results differ across strategies and currency pairs, moreover, they seem connected to which hour of the day they occur, the patterns are not robust across the entire dataset. This conclusion warrants for further improvements on spike definitions and trading strategies.

To read the full report, please see attached PDF below: