sell in may backtest

What the numbers mean

There is no indicator here. The entry test is a month number, and so is the exit — hold SPY through the winter half of the calendar, move to cash through the summer half, one position at full notional, nothing else. The equity curve is a staircase: half a year of ordinary index behaviour, then half a year of a perfectly flat line, repeated. Because one round trip closes per calendar year, the number of trades is essentially the number of years tested.

That structure fixes most of the statistics before any market data arrives. A six-month hold in a broad index finishes higher more often than not, so a high win rate here is a property of the holding period rather than evidence the calendar found anything. Profit factor over so few closed trades is fragile — one bad winter moves it a long way. The figure worth staring at is the drawdown, which is far deeper than the flat summers imply. There is no stop, no take-profit and no maximum holding period, so whatever happens between November and April is taken in full, and the sharpest declines in the tested window landed inside the held months, not the avoided ones.

Cash earns nothing in this model, so the summers contribute exactly zero and the entire case rests on the winters. Read the risk-adjusted number against the headline return — being invested about half the time while still carrying a full-market decline is what holds that ratio down — and treat the same money left invested all year as the real comparison.