Algorithm Review: How Trading Strategies Work
A complete review of the major algorithm categories, the logic behind them, and how to understand why they work.
Algorithm Review: How Trading Strategies Work
Before you build a strategy, review the algorithms themselves. This lesson breaks down the major categories and shows the logic behind each one so you can understand why they work — and when they fail.
The Common Algorithm Blueprint
- Universe: Which symbols are eligible?
- Signal: What triggers entry and exit?
- Sizing: How big is each position?
- Risk: How much can you lose?
- Execution: How and when do orders get sent?
Every strategy is built from these pieces. The edge lives in the signal, but the algorithm only survives through sizing, risk, and execution.
Algorithm Categories
| Category | How It Works | When It Works Best | Typical Risk |
|---|---|---|---|
| Trend-Following | Buy strength, sell weakness | Trending markets | Whipsaw risk |
| Mean-Reversion | Buy extremes, sell back to average | Range-bound markets | Trend continuation risk |
| Momentum | Buy recent winners, sell recent losers | Persistent return continuation | Crash/reversal risk |
| Volatility/Options | Trade implied vs realized volatility | Volatile markets with clear risk premium | Volatility spikes |
| Statistical Arbitrage | Exploit small pricing discrepancies | Highly liquid, correlated assets | Execution risk |
How Each Algorithm Generates Signals
- Trend-Following: Uses breakout or moving average cross signals to capture direction.
- Mean-Reversion: Identifies overbought/oversold conditions and waits for a bounce.
- Momentum: Ranks assets by recent performance and follows the winners.
- Volatility/Options: Compares implied volatility to realized volatility or monitors Greeks.
- Statistical Arbitrage: Uses pairs, cointegration, or correlation to trade relative pricing.
Signal vs. Noise
A good algorithm distinguishes signal from noise. That means filtering entries so you only trade when the market behavior matches your edge, not every small move.
- Identify the edge: What market behavior does the algorithm exploit?
- Define the signal precisely: The exact condition that triggers a trade.
- Confirm with filters: Use trend, volatility, or regime filters as guardrails.
- Exit clearly: Profit targets, stops, or time-based exits are essential.
Example: Trend-Following vs Mean-Reversion
# Trend-following entry
if close_price > sma_50 and close_price > previous_high:
signal = 'buy'
# Mean-reversion entry
if rsi < 30 and close_price < lower_bollinger:
signal = 'buy'
# Same output, different logic.
# Trend-following expects continuation.
# Mean-reversion expects a bounce.
When Algorithms Stop Working
- Trend-following in a choppy market
- Mean-reversion in a strong trend
- Momentum during sudden reversals
- Volatility strategies before sharp spikes
- Arbitrage when liquidity dries up
Understand the Failure Mode
Every algorithm has a market regime where it performs poorly. Know that regime before you trade the strategy with real money.
The Role of Execution
A signal is only an idea until it becomes an order. The same strategy can win or lose depending on order type, timing, slippage, and execution speed.
- Market orders: fast, but can blow through price levels.
- Limit orders: priced control, but may never fill.
- Time-of-day execution: open, close, or random intraday timing.
- Slippage assumptions: build realistic costs into your models.
The Complete View
A strong algorithm is edge + risk management + execution. Review all three before calling it “ready.”
- All profitable algorithms share the same basic structure: universe, signal, sizing, risk, and execution
- Trend-following, mean-reversion, momentum, and volatility strategies each have distinct logic and failure modes
- Understanding the why behind a signal prevents blindly chasing indicators
- A strong algorithm is edge plus risk management and execution discipline