Technical analysis II — indicators & method

A moving average answers one question: what has price averaged over the last N periods? Plotted over time it smooths the noise and makes trend visible. The SMA weighs all N periods equally; the EMA weighs recent ones more, so it turns faster.

Every period choice is the same tradeoff. Short period: hugs price, reacts fast, whipsaws constantly in chop. Long period: smooth and reliable in trends, but hands back a large chunk of every reversal before it responds. There is no setting that is both fast and smooth — anyone selling you one is selling you a backtest overfit.

Momentum oscillators (RSI is the standard) measure the speed of price change, scaled to a range. Above ~70 is called overbought, below ~30 oversold — but in a strong trend, 'overbought' can stay pinned for weeks. The naive fade-the-extreme trade is a money loser in trends.

The higher-quality signal is divergence: price makes a new high, but momentum makes a *lower* high. The move is stretching on weakening force — like a ball still rising as it decelerates. Bullish divergence is the mirror: new price low, higher momentum low.

No single signal deserves a trade. The professional habit is confluence: act when independent lines of evidence agree — structure (a level), momentum (divergence), volume, and context (trend). Two aligned signals beat five redundant ones; five moving averages are one signal wearing costumes.

Any rule you trust should survive a backtest — but backtests lie by default. The classic sins: overfitting (tuning parameters until history looks perfect), look-ahead bias (using data not available at the time), survivorship bias (testing on coins that are still alive), and ignoring costs (fees + slippage turn many 'edges' negative).