ONYX Research - Research & Analysis for business & investment clients Research & Analysis for Business
and Investment Clients
Oil Price Fluctuations - ONYX RESEARCH, INC.®



Percent change data are binned and fit to a Gaussian in figure 4. The fit generally matches the data, yet fails to capture counts found farther from the mean (+/-5% change). Depending on how the fit is performed, the fit may over-estimate or under-estimate highly improbable events.

Percent Change Gaussian Histogram Linear Figure 4. Percent change histogram with Gaussian fit. Red curve plots the Gaussian fit. Green curve plots the 95% upper confidence limit. Blue curve plots the 95% upper prediction limit.

Figure 4 plots three curves. The red curve is a Gaussian fit. The green and blue curves plot the 95% upper confidence limit (95% UCL) and 95% upper prediction limit (95% UPL), respectively. The UCL captures most data, while the more conservative UPL demonstrates the high statistical uncertainty when considering fluctuations far from the mean.


Percent change data are again binned and fit with a function better representing the data in figure 5. Figure 5 plots the same three curves. The red curve is the fit. The green and blue curves plot the 95% upper confidence limit (95% UCL) and 95% upper prediction limit (95% UPL), respectively.



Fits distill information to a few parameters. Parameters are then used to gauge risk and compare data sets. The parameters obtained from the fit in figure 5 provide a better data representation than the figure 4 fit parameters.

Percent Change Histogram Linear Figure 5. Percent change histogram with improved fit. Red curve plots the more complex fit. Green curve plots the 95% upper confidence limit. Blue curve plots the 95% upper prediction limit.

Financial models relying on symmetrical distributions (i.e., Gaussian/Normal, or other symmetric distributions) exhibit what is termed "skewness risk". Kurtosis risk ("fat tail" risk) is another area where underlying model assumptions influence risk calculations. Both skewness and kurtosis risk can have important implications for risk measurement (i.e., value at risk (VAR)).

Oil prices are positively biased (but only slightly). Obviously, based on figure 1, prices rise with time and must have net positive bias. Positive fluctuations are ~3% more likely than negative price fluctuations. This value is close to current world GDP of ~4% (International Monetary Fund (IMF) estimate).

Given this knowledge, long and short-term strategies accounting for random and systematic risk exist. Such strategies may net significant financial gain.

Raw data source: US Department of Energy

Looking for knowledge unique to your needs? Click on the Onyx Premier Black icon below.