Best & Worst Month

What is Best & Worst Month?

Take the full series of monthly portfolio returns produced by a backtest. The Best Month is the single largest value in that series. The Worst Month is the smallest. Both are expressed as percentage returns, usually with the calendar month they occurred in. For a 20-year backtest of an equity strategy, the Best Month is often somewhere between plus 10 and plus 20 percent and the Worst Month somewhere between minus 12 and minus 25 percent, depending on the strategy and the specific window.

These two numbers are tail-extreme statistics, not summary statistics. They are not designed to describe the typical behavior of the strategy. They describe the rare moments at the edges of the distribution. That makes them complementary to averages and standard deviations, which describe the center of the distribution and the spread around it. Average return tells you what to expect on a typical month. Best and Worst tell you what an outlier month looked like.

As reference points, the S&P 500's Worst Month since 1950 was October 1987 at roughly minus 22 percent, with March 2020 a close second at minus 12.5 percent and October 2008 at minus 16.9 percent. The Best Month was October 1974 at plus 16.3 percent, with several other months in the 11 to 13 percent range. Single-stock and concentrated strategies routinely produce monthly extremes well outside that range in either direction. Reading Best and Worst against the index is the quickest way to know whether a strategy is offering a smoother ride or a wilder one.

Formula

The series of monthly returns is built from portfolio values sampled at month ends, net of any transaction costs already deducted in the simulation. Each return is the percentage change from one month-end value to the next: (V_i minus V_i-1) divided by V_i-1. The Best Month is the maximum of that series; the Worst Month is the minimum. Both are reported as percentages, with the calendar month they correspond to.

One convention worth being explicit about: these are non-overlapping calendar months, not rolling 30-day windows. A rolling 30-day window starting in mid-month could easily span a single sharp move (the worst week of October 2008, for example) and produce a different worst-period figure than a strict calendar reading. The calendar-month convention is the standard in mutual fund and hedge fund reporting, which is why it is the default here. If a different period is wanted, it is computed and labeled separately, never confused with the calendar reading.

Why Best & Worst Month Matter in Backtesting

Worst Month is a stomach check. It puts a number on the question every investor secretly asks: in the worst month I would have held this strategy, how much did I lose? It is a different question from Max Drawdown, which can compound across many bad months in a row. Worst Month asks how bad a single calendar month got, on its own. For investors who watch their portfolio monthly (which is most of them) and decide based on the most recent statement (which is most of them, again), Worst Month is the visceral version of risk that Max Drawdown only approximates.

Best Month is less famous but informative in a different way. A strategy with a very large Best Month is one whose returns are concentrated in a few big winners. That is a signature of trend-following, momentum, and many long-volatility approaches. A strategy with a very ordinary Best Month is producing returns through frequent small gains rather than rare large ones, the signature of carry and mean-reversion strategies. Reading the two numbers together gives a fast read on the underlying texture of the strategy without needing to look at every monthly observation.

The pair is also useful for asymmetry checks. A strategy with a Worst Month of minus 18 percent and a Best Month of plus 9 percent has negative skew at the extreme: the worst surprise is twice the size of the best one. That asymmetry is hidden from average return, volatility, and even Sharpe ratio, but it shows up directly in the gap between Best and Worst. When that gap is meaningfully wider on the downside than on the upside, the strategy is one that pays out steadily and then takes a sharp loss, a profile that volatility-based metrics chronically understate.

How SledgeKey Implements Best & Worst Month

Best Month and Worst Month appear on the backtest results page as a pair, each shown as a percentage with the calendar month in which it occurred. They are derived from the simulated monthly return series, which is built from end-of-month portfolio values net of any transaction costs already deducted by the simulation. The maximum of the series is the Best Month; the minimum is the Worst Month; the dates carry through directly from the underlying monthly observations.

The benchmark's Best and Worst Month are computed the same way over the identical window and the same calendar-month frequency, using the benchmark's monthly returns. The side-by-side reading is the most useful framing. A strategy whose Best Month is much larger than the benchmark's, while the Worst Months are similar, is one that captures upside more aggressively than the index without giving up downside. A strategy whose Worst Month is meaningfully deeper than the benchmark's, even with a similar average return, is one taking more idiosyncratic risk than the index in exchange for the same average outcome.

Because Best and Worst are reported with their dates, they double as a quick calibration tool. If a backtest's Worst Month does not occur in October 2008, March 2020, or another known stress window for an equity strategy, that is a flag worth investigating: either the strategy has structurally different exposures than the broad market, or the test window happens to miss a known stress, or there is a data issue. For most long-only equity strategies, the Worst Month landing in one of those well-known stress windows is the expected result, and the magnitude relative to SPY's monthly loss in that window is the more interesting question.

Common Pitfalls

The first pitfall is treating Best and Worst Month as estimates of what the next extreme will look like. Each is a single observation drawn from one realized history. A 20-year backtest produces exactly one Worst Month. That number says very little about how often a similar loss is likely to recur, and almost nothing about whether the next bad month will be larger or smaller. Tail magnitudes require many independent windows or a parametric assumption to be estimated honestly; a single backtest provides neither. Worst Month is a useful anchor for what has happened, not a forecast of what will.

The second pitfall is confusing Worst Month with Max Drawdown. They sound similar and they are related, but they are not the same thing. Worst Month captures the single calendar month with the most negative return. Max Drawdown captures the largest peak-to-trough decline in the portfolio's value, which can compound across many consecutive bad months. A strategy that loses 8 percent in each of four consecutive months has a Worst Month of minus 8 percent and a Max Drawdown of roughly minus 28 percent. Reading one and assuming the other has been answered is the common mistake.

The third pitfall is window dependence. A 5-year backtest that happens to miss October 2008 will show a Worst Month maybe a third the size of the same strategy tested across a 20-year window that includes it. Whenever a Worst Month is being compared across two strategies, the test windows must match, or the comparison is just a comparison of which window saw the bigger drawdown month. Whenever a Worst Month is being used to anticipate worst-case risk, the test window must cross at least one true bear market, ideally two, or the number is not load-bearing.

Related Concepts