"The international energy market is dependent upon reliable transport. The blockage of a chokepoint, even temporarily, can lead to substantial increases in total energy costs."
~ U.S. Energy Information Administration, August 22, 2012.
Academic research shows that a small probability of an emotional event occurring - such as an oil supply shock or a military assault - are over-weighted in human probability assessments. As a result of this overweighting, humans overreact to even the mention of possible vivid, negative events. Overreaction refers to the tendency of asset prices, after violently moving in one direction, to drift back towards their origin. Fear is a measure of overreaction - an indicator that risk is being seen as greater than it actually is. So when fear is high, odds are that investors are overreacting to some vivid negative event, and prices are likely to revert to their pre-crisis level once fear begins to wane.
We ran a series of simple data mining experiments to sort out how Conflict, Violence, and Fear have predicted oil prices over the past 14 years. If you've got an aversion to data-mining, then you'll have lots of bones to pick with our analysis, but in our experience oil investors find these results illuminating. This is not a formal back-tested study, but rather we fit our data to the entire historical 14-year period to see if we could glean useful insights.
METHODS
We know from experience and research that sentiment and news are nonlinear - that is, important information hits the markets in bursts and is typically only significant at high levels.
Using our Thomson Reuters MarketPsych Indices (TRMI) as independent variables, and changes in the highest-volume futures contract of West Texas Intermediate Crude as our dependent variables, we investigated how different intensities of the TRMI's Fear, Conflict and Violence impacted oil prices. Using TRMI data up to the day's futures market close, we then forecasted the price from the following day's open price to the open price one month in the future. We tested 14 years of TRMI Crude Oil data. Our statistical software was custom-designed by our legendary former fund CTO Yury Shatz, and has been enhanced to analyze vast arrays of asset prices by our Chief Data Scientist, Aleksander Fafula, PhD.
In our sentiment data, Conflict refers to high levels of dispute and disagreement surrounding the oil markets. Violence refers to military threats and actions associated with the oil market. Fear is a measure of references to "worries," "concerns," and other symptoms of anxiety surrounding oil prices.
AS CONFLICT HEATS UP
After analyzing our data, we saw an interesting story emerge. First of all, when the news of a conflict or potential war hits the market, it does of course cause a spike in prices. Importantly for traders, that spike has momentum - oil prices continue higher for the following month.
For example a one week spikes in Violence (91 times in the past 14 years, 2.2% average following month return) and Conflict (94 times, 3.2% average future one-month return) lead to high returns. Amazingly, these "spikes" are only changes above average, and as a result cover half of all sampled months. We see a similar positive momentum effect for Fear, but only at high levels of change, probably because Fear is a relatively sparse variable. A rise in one week fear greater than 90% of all such spikes happens 19 times with average 3.2% monthly returns.
When we look at two variables together such as when Conflict rises AND ProductionVolume drops, the following month shows an average return of 8.2% (28 occurrences, 82% accuracy). This is an important result because it is predictive and it is common and it explains what drives prices higher - both conflict AND concerns about declining production. Oil traders have time to enter into these positions.
So we can see above that rising Conflict, Violence, and Fear stimulate further price rises, and these price increases are predictable. It appears that oil traders take their sweet time positioning themselves when tensions rise. This reinforces our personality test finding that speedy reaction time is one of the key traits of top traders.
WHEN CONFLICT IS BOILING
Despite seeing increases in Fear, Violence, and Conflict driving prices higher, static high levels of these sentiments actually precede price declines.
When the 3-month average of Fear is in the top 20% of its historical range, the Crude Oil price drops an average of -1.8% over the next month (37 samples). For Conflict in the top 20% of its historical range, the drop is -3.2% over the next month (37 samples).
When we again look at Conflict paired with another variable (an Association Rule), we find that high average levels of Conflict paired with a sudden one-week drop in Urgency precedes an average -5.5% (19 samples) drop in the crude price over the next month. When Urgency declines (the crisis ends), paired with high average 3-month levels of Conflict, there is a predictable opportunity to short oil and profit.