By Pavel Chigirev, Deltix Quantitative Research Team
Introduction
This paper describes an implementation of an automated quantitative FX trading strategy based on country sentiment factors, and the results from back-testing this strategy. The country sentiment factors and strategy description are provided by RavenPack. We implemented the strategy in the Deltix QuantOffice research platform, a purpose-built C# development studio with embedded math, statistics and data libraries.
The strategy is described in the paper by Peter Hafez “Intraday Forex Trading Based on Sentiment” – http://ravenpack.com/research/index.htm
The strategy implementation called RavenPackESS MC Portfolio is based on the RavenPack Event Sentiment score (ESS) data. The strategy calculates the average ESS for a given economy over a given time window: for instance 1-day, 1-week, 1-month and 3-month. To detect sentiment inflection points, the strategy uses a cross-over between short term and long term averages. In this implementation, the strategy uses a 1 day versus 3 month cross-over.
If the 1-day average ESS moves above the 3-month ESS, the strategy opens a long position (where we consider ESS for base currency or a short position where we consider ESS as the quotation currency). For example, if the strategy trades EURUSD and uses the ESS calculated for the US economy, it will open a short position if the 1-day average ESS moves above the 3-month and a long position if the 1-day average ESS moves below 3-month ESS. The strategy holds the position for nine hours after opening and then closes it.
Results
The strategy “RavenPackESS MC Portfolio” has the following P&L curve over the period 1 January 2010 to 30 June 2013:
The period from 1 January 2010 to 31 December 2012 (highlighted orange) is the in-sample period; the period from 1 January 2013 to 30 June 2013 (highlighted green) is the out-of-sample testing period.
Strategy parameters:
Symbols: AUDUSD EURUSD GBPUSD NZDUSD USDCAD USDCHF USDJPY; 1-min extended bars;
ESS: US ESS
Initial capital: $10,000,000
Bet size: $10,000,000
Position holding interval: 9 hours
Detailed Results for In-Sample period: 1 January 2010 to 31 December 2012
Performance report:
Drawdown chart:
Daily returns distribution:
Red bars represent the lowest 5% of daily returns;
See Next Page for Out-of-Sample Results
Pages: 1 2
One Response to “Promising Results from Automated FX Trading Strategy using Country Sentiment Factors”
Read below or add a comment...