Trend Following
Trend following is the oldest and most robust systematic trading strategy. Gordon is built entirely on this approach.
The Core Idea
Markets trend. Prices don't move randomly — they exhibit momentum. An asset going up is more likely to continue going up than to reverse, and vice versa. Trend following exploits this by:
- Detecting a trend (price breaks above resistance, moving averages cross, etc.)
- Entering in the direction of the trend
- Riding the trend until it reverses
- Exiting when the trend ends
You don't predict where the market will go. You react to where it is going.
Why It Works
Trend following profits from behavioral biases in market participants:
- Anchoring — traders fixate on recent prices and react slowly to new information
- Herding — once a move starts, others pile in, extending the trend
- Disposition effect — traders sell winners too early and hold losers too long
- Overconfidence — retail traders bet against trends, providing fuel for continuation
In crypto specifically, these effects are amplified:
- 24/7 markets with no circuit breakers
- Retail-dominated with high emotional reactivity
- Leverage cascades (liquidations force more selling, extending moves)
- Regulatory news creates sudden regime shifts
Why Crypto
Crypto is one of the last asset classes where simple trend-following strategies still work with meaningful Sharpe ratios. Equities and commodities have been heavily arbitraged by institutional trend-followers since the 1980s. Crypto is younger, less efficient, and more volatile.
However, this edge is compressing. BTC volatility has dropped from 7.58% (2013) to 2.24% (2025) as ETF institutionalization brings more sophisticated capital. Gordon's multi-strategy approach is designed to survive this compression.
The Bet
From Gordon's North Star:
Crypto trends. Retail traders are emotional. The combination of momentum signals, sentiment overlays, and strict risk management extracts value from this consistently — not every trade, but over hundreds of trades.
This is a long-term compounding bet (5+ year horizon), not a get-rich-quick scheme.
Time-Series vs Cross-Sectional Momentum
There are two types of momentum strategies:
| Type | Method | Gordon's approach |
|---|---|---|
| Time-series (TSMOM) | Compare an asset's current price to its own history | Primary — all 4 strategies use this |
| Cross-sectional (XSMOM) | Compare assets against each other, long winners, short losers | Not used — less effective in crypto |
Academic research (Han et al. 2024, Fieberg et al. 2025) confirms TSMOM dominates XSMOM in crypto markets.
Key Principles
- Simple beats complex. OLS outperforms neural networks for crypto prediction. Feature engineering matters more than model architecture.
- Long-only or asymmetric. Crypto has a structural long bias. Shorts get whipsawed.
- More assets = more Sharpe. Diversification across uncorrelated assets is the primary lever for improving risk-adjusted returns.
- Mean reversion is dead at daily+ timeframes. Only works at HFT frequencies in crypto. Skip entirely.