Applying a MACD to global temperature anomalies -as estimated by NASA's GISS in this case, which is monthly data- shows that since 2009 there's a steady decrease (MACD crossed over from above and both MACD and signal line are pointing downwards).
The fact that the MACD exactly captures the increase in temp. anomalies from 1910s to 1940s; captures the decrease from the 1940s to 1970s; and captures the increase from the mid 1970s to ~early 2000s (corresponding to the orange vertical lines that hit the 6th order polynomial where trend changes occur) strongly suggests that the MACD also correctly has identified a (the?) top in temp. anomalies (it remains however unclear if the top is temporary or not, with temporary meaning several years to a decade) in the mid 2000s. Unless global temp. anomalies start rising again, and rise quickly, the MACD will continue to go down. Note that the only time the MACD went down since the mid-70s uptrend was in the mid 1990s due to Mt Pinatubo's eruption (volcano) causing it's emissions in to the atmosphere to cool the globe temporarily. It's actually amazing to see the MACD is able to capture that event too! Hence, the current down turn is unique and rather unprecedented since there haven't been any major volcanic eruptions lately; besides those in Island which were minor compared to Mt. Pin. The reason for the current down turn is IMHO due to a shift in the ENSO cycle going from an el nino dominated multi-decade cycle to a la nina dominated cycle. The effects of the past 2 la nina's are already evident in the global temp. anomalies.
IMHO this is an elegant and simple tool to show the past and current trend in global temps. However, I don't know the MACD's predictive power in this case. But, there is NO reason why a MACD cannot be applied to this type of data, since it is simply based on EMAs. EMAs are indifferent to the underlying data set. In this case I used the monthly values as closing prices. Hence, the MACD clearly shows there has been no warming over the past few years (decade), which linear regression of the data set also shows.
An SSTO can be calculated from this data too, also showing a "sell signal". If interested I can post that plot too.
The fact that the MACD exactly captures the increase in temp. anomalies from 1910s to 1940s; captures the decrease from the 1940s to 1970s; and captures the increase from the mid 1970s to ~early 2000s (corresponding to the orange vertical lines that hit the 6th order polynomial where trend changes occur) strongly suggests that the MACD also correctly has identified a (the?) top in temp. anomalies (it remains however unclear if the top is temporary or not, with temporary meaning several years to a decade) in the mid 2000s. Unless global temp. anomalies start rising again, and rise quickly, the MACD will continue to go down. Note that the only time the MACD went down since the mid-70s uptrend was in the mid 1990s due to Mt Pinatubo's eruption (volcano) causing it's emissions in to the atmosphere to cool the globe temporarily. It's actually amazing to see the MACD is able to capture that event too! Hence, the current down turn is unique and rather unprecedented since there haven't been any major volcanic eruptions lately; besides those in Island which were minor compared to Mt. Pin. The reason for the current down turn is IMHO due to a shift in the ENSO cycle going from an el nino dominated multi-decade cycle to a la nina dominated cycle. The effects of the past 2 la nina's are already evident in the global temp. anomalies.
IMHO this is an elegant and simple tool to show the past and current trend in global temps. However, I don't know the MACD's predictive power in this case. But, there is NO reason why a MACD cannot be applied to this type of data, since it is simply based on EMAs. EMAs are indifferent to the underlying data set. In this case I used the monthly values as closing prices. Hence, the MACD clearly shows there has been no warming over the past few years (decade), which linear regression of the data set also shows.
An SSTO can be calculated from this data too, also showing a "sell signal". If interested I can post that plot too.