COLUMN-Disentangling signals from noise in oil: John Kemp
(John Kemp is a Reuters market analyst. The views expressed are his own)
By John Kemp
LONDON Jan 27 (Reuters) - Intra-day price movements are becoming less informative about longer-term changes in oil prices as the ratio of genuine price signals to random noise deteriorates, at least over shorter-time horizons.
Even a cursory look at the chart for front-month Brent futures reveals that significant intra-day moves are increasingly being reversed before the market closes, more often than not leaving prices back where they started.
Futures prices never move in a straight line. Partial reversals are common. But the ratio of continued moves (trends) to reversals has been falling, implying short-term moves over timescales ranging from seconds to minutes and hours are becoming less useful as predictors of where the market will close or where prices might go tomorrow, or next week or next month.
Since summer 2011, both intra-day volatility (measured by the high-low trading range) and inter-day volatility (measured by changes in closing prices) have declined. But inter-day volatility has fallen faster. More intra-day moves are being reversed before the market closes.
Charts 1-3 show how intra-day moves (measured using the Parkinson high-low volatility estimator) compare with the inter-day volatility (conventionally measured using closing prices).
Because of the way they are defined, intra-day volatility measured by the Parkinson high-low estimator has been lower than conventional close-to-close volatility for most of the last two decades (Chart 1). But the gap has been narrowing, and for much of the last couple of years recorded intra-day volatility has actually exceeded the inter-day measure (Charts 2-3).
The rise in random noise relative to genuine signals is characteristic of range-bound markets lacking a strong trend. Front-month Brent prices have been trapped on the inside of a tight $105-115 range since July 2011, so it is unsurprising that the ratio of signals to noise has worsened (Charts 4-5).
More structural changes may also be at work, however. On a macro time scale lasting days, weeks or months, the futures market seems to have become deeper and more liquid as the commodity boom has drawn in a greater range of financial participants and market makers. As a result, close-to-close volatility has been constant or fallen slightly.
News events that would once have generated sharp price moves now produce a more subdued response because the range of traders and viewpoints has expanded.
In contrast, on a micro scale lasting from seconds to minutes and hours, in which the market is increasingly dominated by computer-driven trading programmes, a significant amount of noisiness and volatility remains and may be increasing.
In many cases, computers appear to trade mostly against each other, which can create price perturbations over short-time horizons.
Since computer-driven trading programmes generally have short holding periods, however, and limited or no mandate to hold positions overnight, most of those positions must be unwound before the close, helping explain why a higher-than-usual number of price moves are being partially or fully reversed. (editing by Jane Baird)