A generation of quants has spent careers building indicators that beat the 200-day simple moving average on backtests, and most of them do, in the narrow sense that backtests measure. And yet the 200-day is still the line every desk draws, every algo respects, every CNBC chyron points at when SPY reclaims it. This is not because traders are slow. It's because the 200-day was never really an indicator in the first place. It's a coordination device.
An indicator predicts. A focal point organizes.#
An indicator is a claim about the future: when this number does X, the market tends to do Y. You can test it, optimize it, walk it forward, and decide whether to trust it. The 200-day, judged on those terms, is mediocre. The slope is laggy. The crossover signals are late. A 12-week donchian channel will smoke it on most regime transitions.
A Schelling point is a different animal. It's a place two strangers agree to meet without being able to talk first — a focal point that's salient because it's salient. The 200-day is the trader's Grand Central: not the optimal place to meet, just the one everyone knows the others are heading toward. Every fund's risk dashboard renders it. Every retail platform draws it by default. Every algo with a "long-term trend" filter checks it. When SPY taps the 200-day from below and reclaims it, you are not watching the line do anything. You are watching the population of people who pre-committed to bid that reclaim — long-only allocators rotating in, dealer hedges flipping, momentum funds adding — show up at the agreed-upon corner.
The math is the excuse. The agreement is the mechanism.
Why the 50-day and 20-day are noisier#
The 50-day is the second-most-watched line on the chart, and it works for the same reason as the 200, just with a thinner population pre-committing to it. The 20-day is thinner still. Walk down the ladder and you reach lines nobody coordinates around — a 14-day, a 33-day — and those are nearly indistinguishable from any other piece of fitted curve. Their predictive power is almost entirely a function of how many people pre-promised to act when price touched them. The math is identical; the social weight is not.
This is also why the 200-day "improvement" trap is so seductive and so wrong. You discover, in a backtest, that a 217-day SMA edges out the 200 by a few basis points on Sharpe. You congratulate yourself and switch. What you've actually done is optimize away the one property that made the original line work — the fact that millions of other people are watching it. A 217-day is a private signal. A 200-day is a public ritual. They are not the same instrument, and the backtest can't tell you that, because the backtest doesn't model the population.
The regime is the signal, not the line#
Here's the part that took me longest to accept. The 200-day is most useful not as a crossing event but as a regime descriptor. When SPY spends 80% of a year trading above its 200-day, you are in a market where dip-buyers are being rewarded and short-sellers are being slowly ground out. When QQQ spends 80% of a year below it, you are in a market where rallies fail and patient sellers eat well. The line itself is a piece of geometry. The regime around it tells you which side of the boat to stand on.
A reclaim in a downtrend regime and a reclaim in an uptrend regime are two different events with the same chart pattern. Pretending otherwise is how you get chopped up paying for the privilege of being early.
The line isn't predicting anything. The crowd around the line is. Make sure you're reading the right one.