Imagine a videogame that, halfway through the match, changes the rules without warning. Suddenly gravity is different, enemies move twice as fast, and the jump button now fires. What made you win five minutes ago now makes you lose. Frustrating, right?
The market is exactly that cheating videogame. Or, if you prefer, think about the weather: there are sunny days and stormy days, and the perfect beach outfit is a disaster in a blizzard. The market has seasons too, and it almost never warns you that they're about to change.
Non-stationarity: the market is not a fixed machine
Scientists have a fancy word for this: non-stationarity. It sounds complicated, but it means something very simple: the rules of the game don't stay still.
A fixed machine always behaves the same: you press a button and get the same result. The market isn't like that. It's made of millions of people whose moods, fears and plans change. What moved prices one way yesterday may move them another way tomorrow. That's why a strategy that worked beautifully last year can stop working this year, without anyone having "broken" anything.
The past is not a promise. It's just one of the many seasons the market has already lived through.
Regimes: the market's different climates
We call those different "climates" regimes. You don't need to be an expert to recognize them. The most typical ones are opposite pairs:
- Trending vs. ranging: sometimes price climbs (or falls) in a fairly clear line for weeks; other times it just wanders around in the same spot, as if trapped in a box.
- Calm vs. volatile: there are quiet stretches with gentle moves, and nervous stretches with sharp jumps that make you sweat.
- Risk-on vs. risk-off: sometimes investors feel confident and buy risky things; other times they're scared and run toward safety. The collective mood shifts.
The key detail is that a strategy is usually designed for one climate. One that wins by following trends can get shredded in a ranging market. One that works in calm can sink in a panic. It's not that it's bad: it's that the season has changed.
Why edges wear out
This is also the deep reason why edges (those small advantages) wear out. Not only because others discover them, but because the market itself changes regime and the advantage stops making sense in the new climate. An honest strategy has to prove it survives several climates, not just the one it happened to live through.
How AlphaLab handles this
AlphaLab starts by accepting this uncomfortable truth: the market is not stable. That's why it doesn't settle for testing a strategy on a single pretty slice of history.
- It tests across different regimes. With techniques like Walk-Forward, it checks whether the strategy keeps working when the market changes season, not just in its favorite period.
- It simulates many possible futures. With Monte Carlo, it shuffles the ups and downs to see whether the advantage holds across different climates or only survived by luck.
- It watches health over time. A strategy has a lifecycle, like a living thing: healthy, then on warning if it starts to weaken, then degraded if its advantage fades, and finally retired when it no longer helps. Watching it prevents you from trusting something a regime change already killed.
This is not magic or a crystal ball. No test predicts the future, and trading can always cause losses. What AlphaLab does, as a lab that runs on your own computer, is help you keep only the strategies robust enough to survive more than one climate, and let them go in time when the wind changes.
Key takeaways
- Non-stationarity means the market's rules change: it's not a fixed machine.
- Regimes are different climates: trending/ranging, calm/volatile, risk-on/risk-off.
- A robust strategy must survive several climates, not just the one it happened to live through.
- AlphaLab tests across regimes and watches the lifecycle: healthy, warning, degraded, retired.
If you want strategies that don't collapse the moment the weather turns, try AlphaLab free for 14 days at this link and see which ones hold up across more than one season.