Growth-Trend Timing with housing starts and income

The housing and income version of Growth-Trend Timing is a stock market timing rule from Jesse Livermore of the Philosophical Economics blog, published in January 2016. It watches two parts of the economy: housing starts, which turn early but jump around, and real personal income, which moves slowly. If either looks recessionary, the S&P 500 is held only while it is above its 10-month average, and Treasury bills otherwise.

The idea#

Growth-Trend Timing follows a price trend only when the economy looks like it is entering a recession, and stays invested otherwise. The author tested several measures of the economy for that switch.

Housing starts are a leading indicator: construction slows well before most recessions. But they are erratic, so he paired them with real personal income growth, which is steadier. He set the breakpoints at a 10% fall in housing starts, measured against the size of the labor force, and 3% growth in real income.

How it works#

At the close of the last trading day of each month:

1. Divide housing starts (FRED series HOUST) by the civilian labor force (CLF16OV), and compare the result with a year earlier.
2. Compare real personal income (RPI) with a year earlier.
3. If housing starts per worker are down no more than 10% and real income is up at least 3%, hold 100% S&P 500 (SPY).
4. Otherwise, hold SPY while it is above its 10-month average (a 210-day average here), and Treasury bills (BIL) while it is below.

Each figure counts from the day it was first published, as first reported. Personal income comes out about a month after the month it describes.

What the backtest shows#

The ETF-era test starts in mid-2007, when the Treasury bill fund begins. Since then it returned about 11% a year, against 10.7% for the S&P 500 and 8% for a 60/40 stock and bond mix, with a better risk-adjusted return than either. It gained 2% in 2008, when the S&P 500 lost 37%.

Its worst drawdown, 34%, was the COVID crash of early 2020, recovered by September. It lagged in 2016 (5% against 12%) and lost 21% in 2022 against 18%, when the signals moved it in and out of the market.

Since publication at the start of 2016 it has returned about 12.5% a year.

With simulated fund history the test starts in 1960 and returns about 11% a year, against 10.7% for the S&P 500, with a better risk-adjusted return. It gained 8% in 1974, when stocks lost 26%, and 24% in 1970 against 4%. It was whipsawed in 1990, losing 15% while the market lost 3%.

When it struggles#

• False alarms from housing. Housing starts can fall sharply without a recession, switching the trend rule on in a market that then recovers.
• A high income bar. Real income can grow less than 3% a year in slow expansions, so the trend rule can be on more often than in other versions.
• Sudden crashes. Monthly data cannot react to a crash as fast as 2020's.

Using it on Tactfolio#

The live strategy above uses first-release FRED data for housing starts, the labor force, and real personal income, with the source's breakpoints. Copy it to change the breakpoints, or compare it with the job growth version and the original Growth-Trend Timing.

Year by year

* Partial year.

The rules as implemented

This is the exact tree Tactfolio runs, rebalanced monthly with signals and trades at the close. Open it to inspect or copy it.

Sources and caveats

• Jesse Livermore (Philosophical Economics), Growth and Trend: A Simple, Powerful Technique for Timing the Stock Market (2016)

• The growth test uses the source's breakpoints: housing starts (FRED HOUST) divided by the civilian labor force (CLF16OV) down more than 10% from a year earlier, or real personal income (RPI) up less than 3%; either turns the trend rule on.
• The source compares the month-end close with its 10-month average; this version uses the daily 210-session average, the closest daily equivalent.
• Each economic figure counts only from the day it was first published, and as first reported: at a month end the latest figure is the previous month's, the one-month lag the source describes.
• SPY and BIL stand in for the source's S&P 500 total return index and 3-month Treasury bills.
• Signals and trades use the close of the last trading day of each month, as in the source.

Common questions#

What is the housing and income version of Growth-Trend Timing?#

It is a variant of the Philosophical Economics timing rule that turns its trend rule on when housing starts per worker fall more than 10% in a year or real personal income grows less than 3%.

Why combine housing starts with income?#

Housing starts give early warnings but are noisy, and real income is steady but slow. Using both lets the strategy react early without following every swing in construction.

Why are the breakpoints 10% and 3%?#

They are the recession breakpoints the author chose. Housing starts swing widely, so their bar is a large fall; real income is steady, so its bar is growth below 3%.

Does it still work?#

Since publication in 2016 it has returned about 12.5% a year. The live record on Tactfolio shows how it is doing now.

Related strategies

• Growth-Trend Timing (Philosophical Economics’ GTT)
• Growth-Trend Timing with unemployment (GTT-UE)
• Growth-Trend Timing with job growth
• Lethargic Asset Allocation (Keller’s LAA)