Strategies · intermediate · 10 min

Why Most Strategies Fail the Gate

Across this track we have taken six of the most famous, most heavily marketed strategies in retail trading, written each as a precise mechanical rule, and run it on real SPY data through Visor's robustness gates. This lesson puts all six verdicts in one place — because the pattern across them is the real lesson, worth more than any single result.

Here is the pattern, stated up front: almost all of them made money, and every single one failed. That is not a contradiction. It is the most important thing a new trader can understand, and it is the thing every course selling these strategies is structurally unable to show you.

The six verdicts

Each card below is the honest robustness report, computed once at build time by the same engine the app runs on every backtest. Read How to read a robustness report if the four gates are unfamiliar.

The golden cross — the SMA(50)/SMA(200) crossover. Made about +60% on SPY over five years:

A moving-average trend follower — the faster SMA(20)/SMA(50) version:

Order blocks — buying the impulse away from the last opposing candle. Made about +54%, and even survived out-of-sample:

Fair value gaps — trading the three-bar imbalance. Made about +45%:

Liquidity sweeps — fading the false breakdown. The one that fails hardest, in two ways at once:

Mean reversion — buying RSI-oversold dips:

What the pattern is telling you

Line the six up and three failure modes account for all of them:

1. They were long a market that went up. SPY roughly doubled over the window. Almost any rule that spends time long makes money — the return is mostly participation, not skill. Four of the six made a positive return and still lost to simply buying and holding. "It made money" was never the question.

2. The timing didn't beat its own scramble. The random control re-runs each rule 500 times with its entry timing shuffled, keeping the trade count, exit, and market identical. Every one of the six was beaten by a meaningful share of its randomly-timed twins. When a coin-flip on when to enter does about as well as your signal, the signal's timing — its entire claim to skill — was not doing measurable work.

3. The ones that looked best in-sample broke out-of-sample. The liquidity-sweep rule made +17% on the first 70% of the timeline and lost on the held-out 30% — the textbook overfitting fingerprint. A rule that works on the data you can see and stops on the data you held back has fit the past, not found the future.

The honest conclusion — and the honest caveat

This is not a claim that technical analysis is worthless, that these patterns are fake, or that no one can trade. It is a narrower, harder, more useful claim: the simple, mechanical, first-taught version of every famous strategy, on this instrument and window, was indistinguishable from luck once tested properly — and the reason nobody selling these strategies tells you that is that the test is easy to hide. Show the in-sample line going up; don't show the held-out half or the scrambled twins.

Two honest caveats keep this fair. First, these are specific mechanical versions on one instrument over one five-year window — a discretionary trader reading the same pattern in context is doing something these rules don't capture, and a different instrument or period could look different. Second, and more important: none of this tells you what will happen. It tells you what these rules did, and how to check whether a result is signal or luck. That check — the random control and the out-of-sample split — is the single most valuable habit in trading, and it is the one thing Visor insists on that a £997 course never will.

What to read next

• The Random Control — the scrambled-twin test every one of these failed.
• The Overfitting Trap — the in-sample/out-of-sample failure, in depth.
• Building Your Own — put your own idea through the same gates.