1994issue C051-23
The late-bull intermarket clock and regression-channel gates
A late-cycle intermarket timetable was used to reweight style quadrants and short-term bond exposure first. Only after that clock was set did a regression channel, with a volume envelope as a preliminary filter, admit or evict a single holding.
- The intermarket timetable treats a rise in gold, then commodities, then a decline in bonds as an end-of-bull-market clock once that sequence appears.
- When the clock turns late, style quadrants shift toward value and away from growth, with a larger short-term bond sleeve and a geographic overlay for markets in a different cycle stage.
- The admission triangle keeps a name only when portfolio fit, fundamental support, and technical confirmation all agree, and it evicts the name if any gate fails.
- The regression channel is the last technical gate: a break below an established upward channel trails the exit, and a spike through the upper band flags a rise treated as typically not sustainable.
Read the clock first
A late-cycle intermarket timetable was defined as a rise in gold, then a rise in commodity prices, then a decline in bond prices. That sequence was not observed in the prior one to two years. Once it appeared, it was treated as an end-of-bull-market clock.
Intermarket transmission was described as a perturbation that can move from gold through other commodities into bonds and then equities. Transfer functions were proposed as a way to quantify those cross-market correlations.
Reweight the style quadrants
After metals and other commodities advanced, the book was shifted toward short-term bonds in anticipation of higher policy rates. Equities then declined on 4 February 1994 following the first rise in the federal-funds rate.
When traditional valuation measures labeled the domestic equity market historically expensive, allocation was described as skewed toward value names plus a 50 percent weight in short-term bonds, with a preference for dividend income where large capital gains looked less likely.
Style was changed toward the end of a bull market versus the beginning: more value and less growth. Late-cycle, non-dividend small-capitalization names were judged harder to hold with an acceptable risk-adjusted profile.
Holdings were kept in standing style quadrants of large versus small capitalization and growth versus value, plus a geographic overlay of markets closely tied to the United States versus markets such as Japan that sat in a different cycle stage.
Broad diversification was described as eight to 16 names spanning size, style, and geography, generally nearer 16 when those dimensions were all required.
Admit or evict one name last
No security entered the book unless portfolio construction, fundamental work, and technical confirmation all agreed. A holding was sold if it ceased to satisfy all three. That three-gate rule is the admission triangle.
A least-squares regression line with parallel bands that contain price was used as the technical gate. A breakdown of that channel in April 1992 was labeled bearish. A later buy required both an on-balance-volume cross of a 40-day average inside a two-standard-deviation, 40-day envelope and an upside break of a downward channel.
The volume envelope is that cumulative volume line with a 40-day average and a two-standard-deviation band. It was used as a preliminary timing filter.
Once price occupied an established upward regression channel, a break below the channel was used as trailing-stop logic. A spike through the upper band was treated as a warning that the rise was speculative and typically not sustainable.
All readings on this track · 21 readings
- 1991A commodity-bond regime check before an equity relative strength index
- 1993A case study in refusing an isolated entry
- 1994The late-bull intermarket clock and regression-channel gates
- 1995Seasonal and political cycle maps as tested abstention regimes
- 2001Year-end high and low as a four-state regime filter
- 2002A six-market moving-average regime map from 2002
- 2003Gold as a regime marker after equity leadership fades
- 2003A four-season regime map before the shorter chart
- 2003From downtrend to trading range: implied volatility and volume after a shock fades
- 2006Exit discipline after regime shocks
- 2006Liquidity, growth, and midcycle market regimes
- 2006Yield-curve spread as an equity regime filter
- 2008Commodity, credit and bond alignment is not a forecast
- 2013Two lookback windows locate the cycle across bonds, stocks, and commodities
- 2014When ranging ends, replace the relative strength index, not its lookback
- 2016A Treasury-high-yield ratio as an equity regime filter
- 2016One-off shocks versus seasonal and secular regimes
- 2018Yield-curve flattening as missing market-regime context
- 2019Classifying six market phases around a single trade
- 2020Evaluating a two-window seasonal equity-bond switch
- 2020Dating energy regime breaks before the trade