Temperature records are inferred from climate proxies. Ice cores are the primary proxy used. Scientists manually count the number of air bubbles in a sample of

Temperature records are inferred from climate proxies. Ice cores are the primary proxy used. Scientists manually count the number of air bubbles in a sample of ice to estimate the level of CO2. The amount of oxygen isotopes present is used to estimate temperature. The accuracy of these methods cannot be validated using current ice cores because of the time it takes for the layers to form. In other words, data used in climate science is extrapolated and estimated. As you can imagine, this introduces a great degree of uncertainty to the data. Nevertheless, this data is used as input to climate models. Climate models are pieces of software written in Fortran that contain 500,000 lines of code. Models are run backwards to measure how closely they match the "observation" above. Stock market algorithms use a similar method of backtesting data. The problem with stock market models is they are only accurate with the historic data they are based on. Future price movement can simply not be predicted by using a few of the millions of variables that exist in the market.