Loading basis…
Spot minus the class front-month FFA, in $/day. Formula basis_v1_spot_minus_class_front_mon.
Last updated
All history / 1y / 3y: every trading day in that sample, not the same calendar dates. Percentile is where the last print (or the selected-window average) sits in that pile of days. Z is (value − sample mean) ÷ sample stdev of levels, not daily moves. All-history mean mixes old regimes, so 1y and 3y are the more honest centres when the spread has drifted.
This block ignores seasonality. It asks: vs the last few months, is the market running, stretched, or noisy?
Momentum: how many points the price has moved vs 1, 3 and 6 months ago. Positive = higher than then.
IQR robust bands: the middle 50% of prices over the last year, plus a wide “normal” fence. If price is inside the fence it is not an outlier vs the last year. Robust means a few crazy days do not blow the bands out.
Vol-scaled distance: how far price is from its recent average, measured in typical daily moves. −5 means about five quiet days of selling below that average.
Half-life (days): if the spread is stretched, about how many days until half of that stretch typically fades. Short = snaps back; long = wanders.
Jump risk (σ tail): the rare big day (99th percentile) vs a typical day. ~2.3 would look like a “normal” bell curve; higher means fatter crashes/spikes.
IQR median / band: last year’s middle price, and the wide “still normal vs last year” range.
Jump days / year: how often you get a 3-sigma daily move, scaled to a year.
Excess kurtosis: extra fat tails vs a bell curve (0 = normal). Positive = more extreme days than you’d expect.