Experimental comparison of methods that turn weekly index-futures positioning into
price-axis context. Three stacked panels share one time axis over the last six months.
The spine leaves a forward window for a one-month scenario fan.
- SPX spine — spot, regime shading, extreme markers, cost-basis line, and anchored levels. Colored dashed lines are constant-drift SPX tapes (+1σ / flat / −1σ / −2σ) that leave the last SPX print. Dotted rays mark the ST / MT / LT trend pivots — the 20 / 60 / 200-day moving averages where each horizon flips sign.
- CTA panel — historical trend-follower estimate. Each horizon compares price to its moving average, divides the gap by the noise a random walk makes over that window, and passes the result through tanh. The three horizons are averaged and then vol-targeted: exposure shrinks when realized vol runs above 15% and levers up to 1.5× when it runs below. Colored dashed lines are implied CTA exposure under each SPX tape, re-scored every session, with the value at the horizon end labeled.
- COT index — 0–100 normalization over a 156-week rolling window; 90/10 bands.
Positioning is aligned to its public release date (Friday, three days after the Tuesday snapshot)
so the chart does not front-run the weekly report. The latest report is therefore up to a week
behind the tape, and says nothing about flow since the snapshot.
The badge reads asset-manager flow — the four-week change in that cohort's net
book, with the size in the subtitle. Asset managers are the directional cohort here: over the
156-week window they were net long in every week, while leveraged funds were net short in every
week. The leveraged short is largely a cash-futures basis position, not a view, so its level and
its entry price carry no sentiment. The entry line is labelled by side — for a net-short book an
entry below spot is a loss, not support.
The CTA panel carries a dollar axis only when the model earns it. Unit exposure is fit by least
squares to CFTC leveraged-funds net notional over the full 156-week window; the subtitle shows
the correlation. Below r = 0.40 the panel keeps the [−1, +1] signal axis instead of publishing a
dollar number the data does not support. Leveraged funds are a proxy cohort, not CTAs, so treat
the dollar scale as an order of magnitude.
Experimental comparison harness — not investment advice.