Independent formula timers kept as a testable combination

The job was to limit losses

The practitioner described the main job of timing as limiting losses in declining markets, not as maximizing gains while prices are rising.

Mechanical timing was said to keep capital out of the market as much as 40% to 50% of the time. That idle time was treated as a risk-reduction feature rather than a defect.

A single timer was hard to tolerate

The first procedure used was a simple price-based trend-following system with a long paper history. Live experience still led to treating one timer as hard for most people to tolerate day to day.

After starting with a single timer in 1983, the book was switched in 1984 to four formula-only systems with different in-or-out rules. Any procedure that needed artistic judgment or a subjective decision was rejected.

Four formulas, one shared exit stance

All four systems were required to share a stop-loss stance so that none could stay constructive if the market fell sharply.

The four equity rules were specified as price-only trend following; price and momentum with an interest-rate factor; advances versus declines plus up versus down volume; and ratios that mix volume, price, advance-decline, and new highs versus new lows.

Companion notes described one timer as putting daily up and down points and volume into a long-term moving-average view, another as requiring six price and breadth factors to agree, a third as trend-following price reversals with momentum and interest rates, and a fourth as a filtered price moving-average trend system. Each was described as averaging only a few trades a year.

Equal sleeves, not a committee blend

A buy from any one system was allowed to move only 25% of the portfolio. The four procedures were not blended by committee or consensus because each had been built to stand alone.

After a buy, the sleeve was filled with no-load, diversified rather than sector funds chosen by relative strength since the latest market low. That relative-strength selection was used on systems that typically traded about two to four times a year.

An override makes the rules untestable

Listeners who already used price-based moving averages, advance-decline, new highs and lows, and momentum still reported that they would not follow a signal without adding other variables. The interview treated that override as making the procedure untestable.