STRATEGY GRAVEYARD

Where Strategies Come to Die.

Most trading projects show you the winners. We show you everything — especially the failures. Every strategy below was tested with real data, and killed when the evidence said it didn't work.

"If you can't prove it survives, don't trade it."

4

Strategies Killed

1

Survived Testing

-$14,167

Total Losses (Killed)

500+

Coins Backtested

KILLED Jan 2026

Momentum Breakout LONG

-$14,115

THE THESIS

When price breaks above the 20-candle high with 2x average volume and an uptrend, ride the momentum. Classic breakout strategy used across all markets.

WHAT HAPPENED

Backtested at +400%. Deployed live. Lost $14,115 in weeks. The strategy was a textbook case of overfitting — it looked perfect on historical data because it was using future information (look-ahead bias) in the backtest.

Once the code was fixed, the true performance was revealed: 37.5% win rate against a breakeven requirement of 71.4%. The strategy was structurally broken.

THE LESSON

A single line of code using the wrong candle index turned a losing strategy into a fantasy winner. After this, we established the rule: backtest code must be 100% identical to live trading code. No exceptions.

AUTOPSY

PnL-$14,115
Win Rate37.5%
Breakeven Needed71.4%
DirectionLONG
SL / TP5% / 10%
Time Active~3 weeks
Cause of Death

Look-ahead bias + overfitting

KILLED Feb 2026

BB Squeeze LONG

-$26

THE THESIS

Mirror of the BB Squeeze SHORT strategy, but in the opposite direction. If squeezes predict short entries, maybe they predict long entries too.

WHAT HAPPENED

Barely lost money, but that's not the point. The win rate was 51% — essentially a coin flip. After fees and slippage, the expected value was negative.

THE LESSON

Symmetry doesn't work in crypto. The market has a structural bias: LONG strategies in crypto altcoins face different dynamics than SHORT. We tested 18 different LONG configurations. All failed.

AUTOPSY

PnL-$26
Win Rate51.0%
Expected ValueNegative
DirectionLONG
Coins Tested500+
Data Range2+ years
Cause of Death

No statistical edge

ABANDONED Jan 2026

HV Squeeze (Historical Volatility)

Inconclusive

THE THESIS

Replace Bollinger Bands with Historical Volatility squeeze detection. HV might capture different compression patterns and provide uncorrelated signals.

WHAT HAPPENED

HV Squeeze generated similar signals to BB Squeeze but with no measurable improvement. Running both would increase complexity without adding alpha.

THE LESSON

More indicators doesn't mean more edge. If two systems generate correlated signals, adding the second one just adds maintenance cost without improving returns.

AUTOPSY

PnLN/A
Edge vs BBNone
CorrelationHigh
Live TestedNo
Cause of Death

Redundant with BB Squeeze

SHELVED Jan 2026

ATR Breakout

Inconclusive

THE THESIS

Use Average True Range to detect volatility expansion. When ATR exceeds its moving average by a threshold, enter in the direction of the breakout.

WHAT HAPPENED

Backtested extensively across 500+ coins. The strategy didn't outperform BB Squeeze on any meaningful metric. Shelved for potential future review if market conditions change.

THE LESSON

Not every strategy needs to die dramatically. Sometimes a strategy just isn't better than what you already have. The discipline is in not deploying it "just because."

AUTOPSY

PnLN/A
vs BB SqueezeUnderperformed
Live TestedNo
Cause of Death

Didn't outperform existing strategy

THE SURVIVOR

BB Squeeze SHORT

Out of every strategy tested, one survived: Bollinger Band Squeeze SHORT. Not because it's exciting. Because it's the only one with a statistically significant edge across 500+ coins and 2+ years of data.

It has evolved through 16 versions. Parameters have been optimized, filters have been added and removed based on evidence, and it continues to be validated against live performance data.

It's not perfect. It will have losing streaks. But it has one thing the graveyard strategies didn't: a proven edge.

CURRENT STATUS

StatusLIVE
DirectionSHORT only
Backtest PF2.55
Win Rate (BT)70.4%
Coins536
Data Range2+ years
Versions16+
Why it survived

Positive expected value across all test periods

Why Publish Our Failures?

Because anyone can show you a winning trade. The real test of trust is showing you everything that went wrong.

ACCOUNTABILITY

Publishing losses makes us accountable for every decision.

EDUCATION

Others can learn from our mistakes without paying the price.

TRUST

If we show you the bad, you know the good is real.

DON'T BELIEVE. VERIFY.