Using Turnleaf Analytics inflation forecasts to trade foreign exchange (FX)
Turnleaf Analytics has devised a strategy that uses differences between its own inflation forecasts and consensus forecasts to trade FX. This approach is based on the insight that deviations in inflation expectations serve as a proxy for changes in monetary policy expectations, which influence FX rates.
For example, if inflation in a country rises more than anticipated, it suggests that the central bank may adopt a more hawkish stance, potentially leading to rising yields and a stronger currency. Conversely, a dip in inflation relative to other countries could signal a doveish approach, leading to currency depreciation.
Since 2018, Turnleaf Analytics' inflation strategy for trading FX has delivered risk-adjusted returns of 1.06 and annualized returns of 4.26 per cent, outperforming traditional trend and rates-based strategies, which have had risk adjusted returns of 0.0.02 and 0.53 respectively.
In Figure 1, we show the historical returns for trading FX pairs using three types of trading rules:
• In orange, using relative yields which can be a proxy for monetary policy expectations.
• In grey, using the underlying trends in spot.
• In yellow, using relative Turnleaf Analytics inflation forecasts compared to consensus.
In emerging markets, the relationship between inflation, monetary policy, and currency rates can be more complex than in developed markets.
There is sometimes an inverse relationship between yields and FX in emerging markets, when risk-averse investors can dump both the local currency and bonds at the same time if they do not see the central bank as credible. Hyperinflation can lead to currency depreciation if investors doubt the central bank’s ability to control inflation. Turnleaf tackles this by focusing on relative value pairs in more developed parts of emerging markets, such as Central and Eastern Europe.
The company constructs its inflation forecasts using macroeconomic, market, benchmark and alternative data. An example of the latter would be time series data on pollution, which can be used as a proxy for industrial activity. All data is collected, pre-processed and then fed into a machine learning model, which is relatively simple and easy to understand. The inflation model generates forecasts from one month to 12 months, and is updated for each country on a monthly basis shortly after the inflation release for that country. For over two thirds of the 33 countries covered, Turnleaf Analytics publish short term forecasts/nowcasts several days before each inflation print.
Turnleaf Analytics' pioneering approach to FX trading is an important advance in financial analytics and an example of what can be achieved by combining machine learning with macroeconomic insights.
The big news…
Turnleaf Analytics economic forecasting data is now available on Macrobond One.