Bull-Market Genius

intermediate6 min read

A strategy that only ever saw a rising market has not really been tested. Cover the bad years.

Regime bias is testing a strategy only over a period dominated by one market environment — usually a bull market — and mistaking the environment’s tailwind for the strategy’s skill. “In a bull market, everyone’s a genius.”

The trap: *a rising tide lifts all boats, so a backtest run only during a bull market can’t tell skill from luck. If your test period is 2009–2021 (a historic bull run), almost any long-biased strategy — even “buy random stocks and hold” — looks brilliant. You haven’t learned whether your edge works; you’ve learned that being long during a bull market works (which everyone already knew). The real test of a strategy is the bad* years: did it survive 2008, the 2020 crash, the 2022 drawdown? A strategy that’s never met a bear market, a sideways grind, or a volatility spike is untested, not proven. So your backtest must deliberately span multiple regimes — bull, bear, sideways, high- and low-volatility — or you’re simply measuring the weather, not the navigator. Always ask: what environment did this strategy actually face, and what happens when that environment flips?

The bias — testing only in one (usually bullish) regime credits the environment’s tailwind to the strategy.
The honest test — performance in bad years: crashes, bear markets, sideways chop, volatility spikes.
Untested ≠ proven — a strategy that never met a bear market hasn’t been validated, however good its bull-market numbers.
The fix — span multiple regimes; explicitly stress-test against historic crashes (a later lesson).

ExampleA momentum strategy backtested over 2013–2021 shows a glorious 28% CAGR. But that window was almost all uptrend. Extend it to include 2008 and 2022 and the picture changes — momentum can crash hard in sharp reversals. The original test wasn’t wrong, it was incomplete: it never showed the strategy facing the conditions that hurt it most.

Key takeawayRegime bias mistakes a bull market’s tailwind for strategy skill — in a rising market, everyone looks like a genius. A strategy that never met a bear market, sideways chop or volatility spike is untested. Always span multiple regimes and stress-test the bad years before trusting a backtest.

FAQs

What time period should my backtest cover?

As long and *varied* as quality data allows — ideally spanning at least one full market cycle including major bull and bear phases (e.g. 2008, 2020, 2022 for Indian/global equities). Coverage of diverse regimes matters more than sheer length; a short period that includes a crash is more informative than a long one that’s all bull market.