1990issue C011-4
Constructing fund cash and manager-preference filters
Two monthly filters can be built from one fund-manager ledger. Cash-percent scales unused cash as potential-demand. Manager-preference scales, smooths, and standardizes net allocations so an equity lean is kept separate from cash on the sidelines.
- Cash-percent is cash and short-term holdings divided by the sum of stock-fund assets and bond-and-income-fund assets, and a high reading is treated as potential-demand.
- Cash-percent is labeled constructive above a plus-one standard-deviation line and adverse below the matching lower band.
- Manager-preference compares asset-scaled net-stock-purchases with the matching other-investment net after each series is tamed with an exponential-average or a simple-moving-average and converted to a standardized-deviation.
- The cash-position construction is treated as the more trustworthy fund-manager filter. The purchase-and-sale construction is treated as needing more caution.
Two filters from one ledger
The construction draws on seven monthly series: common-stock purchases, common-stock sales, other purchases, other sales, cash and short-term holdings, stock-fund assets, and bond-and-income-fund assets.
From that single fund-manager ledger, two filters are built. One measures unused cash. The other measures how net allocations lean after they have been scaled, smoothed, and standardized.
How cash-percent is built
Cash-percent is cash and short-term holdings expressed as a share of combined stock-fund and bond-and-income-fund assets. It is built by dividing cash and short-term holdings by the sum of stock-fund assets and bond-and-income-fund assets.
A high cash-percent reading is treated as a large stock of unused buying power that can later support equity demand. That unused cash is potential-demand.
Cash-percent levels are labeled constructive above a plus-one standard-deviation line and adverse below the matching lower band.
Why raw purchases are not a clean reading
Raw stock purchases and sales are not treated as a clean manager-sentiment reading because shareholder subscriptions and redemptions also move those totals.
How manager-preference is built
Manager-preference is estimated by comparing net common-stock purchases with net purchases of other investments after each net is scaled to total fund assets.
Net-stock-purchases are monthly common-stock purchases minus monthly common-stock sales. The other side of the comparison is net purchases of other investments, built from other purchases and other sales.
Total assets are defined as stock-fund assets plus bond-and-income-fund assets. Each net is expressed as an asset-scaled-net so growth in the asset base does not dominate the reading.
The asset-scaled-net is then tamed with an exponential-average or a simple-moving-average. The exponential-average is a 33 percent exponential smoother applied to the erratic asset-scaled-net series. The simple-moving-average is a five-month arithmetic alternative offered for the same series.
After smoothing, each value is converted to a standardized-deviation by subtracting the series mean and dividing by the series standard deviation.
The two standardized curves are compared so a lean toward common stocks can be distinguished from a lean toward other investments. That comparison is manager-preference.
Which construction is treated as more trustworthy
The cash-position construction is treated as the more trustworthy of the two fund-manager filters, while the purchase-and-sale construction is treated as needing more caution.
All readings on this track · 21 readings
- 1988Classifying primary futures for regime-aware trades
- 1989Odd-lot short interest as a constructed breadth series
- 1989Volume-scaled member versus public short sales as a two-class classroom test
- 1989Evaluating public versus member short-sale breadth
- 1989Lock smoothing inside each short-share before the breadth ratio
- 1989Headline intensity as a three-horizon contrarian filter
- 1989Evaluating technical-fund open interest and daily volatility
- 1990Constructing fund cash and manager-preference filters
- 1990Stock-style futures rules miss index-spread drivers
- 1991Constructing short-squeeze pain from interest and margin
- 1995Contrary sentiment and put-call premiums around large Dow swings
- 1995Act on a pattern change, not on formula signals
- 2002Single-stock futures: contract selection and the margin box
- 2003A crowd fade needs an insider-pattern divergence gate
- 2005A quiet insider tape as a seasonal regime case
- 2006Unwritten pit cues need testable rules
- 2010A gold corrective-phase case after the 1265 high
- 2014Ticket charges, margin carry, and bank liquidity limits
- 2014A liquidity filter for majors, crosses, and exotics
- 2017Test a high-low forecast as one procedure, then cut the book
- 2020Follow or fade as one trading procedure