Open trading account 
Economic Research on Forecasting High and Low Prices: Cointegration
Forecasting, even with all its ambiguity and uncertainty, always remains one of the most alluring questions in financial modeling. And while spot prices themselves are commonly considered unpredictable, there are many theories that explore deterministic trails in other asset parameters. In this research we take a closer look at the idea of forecasting high and low prices through their linear combination – or, in other words, by using the notion of cointegration.
Results
1. Cointegration is not equally pronounced in all currency pairs.
2. Cointegration analysis results might be connected with asset volatility.
3. The pairs from different groups are best described by different models.
4. ECM gives an apt, but not supreme forecast of the high and low prices.
Conclusion
In this research we have studied a method for forecasting time series that evolve in equilibrium relative to each other. We have confined the study to high and low prices, as they offer a channel forecast for future movements, but the same analysis can be carried out for ask and bid, or spot and futures prices.
It appeared that some currency pairs do indeed possess the property of keeping a stable range between highs and lows, but a great number of tests is required to find out if the results are reliable. We have also seen that asset volatility might have some influence on the intensity of the trait, so that the pairs with higher and less stable volatility are less likely to give positive results.
Apart from the purely theoretical characteristics, this type of analysis gives us an opportunity to build a special kind of forecasting model, ECM. The results of our study showed that ECM offers adequate forecasts, and in some cases performs better than the other tested method. Moreover, it is known to be possible to improve the forecast by employing an augmented, more complicated model - FVECM. This method was not covered in our research, but other econometric studies report various promising results. With that it can be said that this time consuming and laborious analysis is, after all, noteworthy and useful.
Forecasting, even with all its ambiguity and uncertainty, always remains one of the most alluring questions in financial modeling. And while spot prices themselves are commonly considered unpredictable, there are many theories that explore deterministic trails in other asset parameters. In this research we take a closer look at the idea of forecasting high and low prices through their linear combination – or, in other words, by using the notion of cointegration.
Results
1. Cointegration is not equally pronounced in all currency pairs.
2. Cointegration analysis results might be connected with asset volatility.
3. The pairs from different groups are best described by different models.
4. ECM gives an apt, but not supreme forecast of the high and low prices.
Conclusion
In this research we have studied a method for forecasting time series that evolve in equilibrium relative to each other. We have confined the study to high and low prices, as they offer a channel forecast for future movements, but the same analysis can be carried out for ask and bid, or spot and futures prices.
It appeared that some currency pairs do indeed possess the property of keeping a stable range between highs and lows, but a great number of tests is required to find out if the results are reliable. We have also seen that asset volatility might have some influence on the intensity of the trait, so that the pairs with higher and less stable volatility are less likely to give positive results.
Apart from the purely theoretical characteristics, this type of analysis gives us an opportunity to build a special kind of forecasting model, ECM. The results of our study showed that ECM offers adequate forecasts, and in some cases performs better than the other tested method. Moreover, it is known to be possible to improve the forecast by employing an augmented, more complicated model - FVECM. This method was not covered in our research, but other econometric studies report various promising results. With that it can be said that this time consuming and laborious analysis is, after all, noteworthy and useful.