S&P 500 SP5 Canary Strategy — monthly, 5 stocks, equal weight

Start with the current S&P 500 constituents.

Stock selection

Rank stocks by:

126-trading-day return + 0.5 × 252-trading-day return

Keep the top 80 (I stuck this in because googlesheets will not look up 500 stocks for you, but if you get a more limited list elsewhere (I use stockfetcher) and give it to googlesheets, then googlesheets will do the rest of the calculations for you)

The strategy without this step, using all 500 S&P500 stocks, was 27.1% cagr, i.e. still good

Among those 80, rank by:

SMA30 / SMA180

Keep the top 50.

Among those 50, rank by 200-trading-day average dollar volume.
Hold the top 5, equal weighted. Rank by trailing 200-trading-day average daily dollar volume (daily close × daily shares traded, averaged over 200 trading days). (i.e. I don't fully understand gtr's usage so this is making clear what I used in the test, adjusted close).

The historical signal date is the third-to-last trading day of the month, with the portfolio held for the following monthly period.

Canary timing

Among those final five stocks, calculate each stock's trailing 126-trading-day beta to SPY, using price data only through the sixth trading day before the signal. Select the stock with the highest beta. That one stock is the portfolio's “canary.”

Go to CASH for the coming month if either of these conditions is true:

The canary had any single-day gain of at least +9% during trading days −25 through −6 before the signal; or
The canary's price on day −6 is at least 10% below its price 60 trading days earlier.

Either of those timing signals work. One showed that a very large 1 day gain in the prior period in the stock in the strategy with the highest beta predicted bad performance of the stocks in the next month.

The other showed losing 10% or more in the prior 60 days (minus last week, since price reversal is so common based on last week's price change).

Otherwise, hold the five selected stocks.

The deliberate exclusion of the latest five trading days is part of the rule; those days showed different, due to 1 week reversal-like behavior in our tests of so many strategies.

Historical result

For the exact monthly canary history I calculated from start November 1993 through the March 2026 signal:

Statistic SP5 Canary
CAGR 30.87%
Maximum drawdown −24.15% took 13 months to recovery after that crash; 18 months from when the decline began until recovery
Sharpe 1.31
Sortino 2.89
Beta vs SPY 0.78
Ulcer Index 6.75
Worst full calendar year −10.84%
Median full calendar year 31.83%
Positive full calendar years 90.6%
Longest recovery 29 months from anydrawdown no matter how tiny to the largest
Average recovery 4.89 months
Worst rolling 5-year CAGR 8.38%
Worst rolling 10-year CAGR 13.88%
Worst rolling 15-year CAGR 20.69%

For comparison, the untimed SP5 had roughly a −54% maximum drawdown. So the notable result isn't merely getting about 30% CAGR from an S&P-500 universe; it's doing so historically with a maximum drawdown of about 24%.