Signal methodology

Throughout this page, c[t] is the close of bar t, h[t] is the high, l[t] is the low, and v[t] is the volume. n is the lookback parameter. All formulas operate on the user's chosen resolution (1m / 5m / 1h / 1D / etc.).

Simple Moving Average (SMA): SMA_n[t] = (c[t-n+1] + … + c[t]) / n. Used for trend identification and as an input to MACD-derivatives.

Exponential Moving Average (EMA): EMA_n[t] = α·c[t] + (1−α)·EMA_n[t−1] with α = 2/(n+1) and seed EMA_n[n−1] = SMA_n[n−1]. EMAs respond faster to recent price than SMAs of the same length and are used as the basis for MACD.

Drogo uses Wilder's RSI with the canonical 14-bar lookback: gain[t] = max(c[t]−c[t−1], 0), loss[t] = max(c[t−1]−c[t], 0). The first 14-bar averages are simple averages of gain and loss; subsequent values use Wilder smoothing: avgGain[t] = (avgGain[t−1]·13 + gain[t]) / 14. Then RSI = 100 − 100/(1 + avgGain/avgLoss).

RSI is bounded in [0, 100]. We classify ≤ 30 as oversold, ≥ 70 as overbought, ≤ 20 / ≥ 80 as extremely oversold/overbought. These thresholds are conventional defaults; they are not buy/sell signals on their own.

When avgLoss = 0 over the lookback (rare on liquid instruments, common on illiquid crypto), we report RSI = 100. This matches Wilder's original definition and avoids a division-by-zero NaN.

MACD = EMA_12(c) − EMA_26(c), signal = EMA_9(MACD), histogram = MACD − signal. We report all three. A bullish cross fires when MACD[t−1] ≤ signal[t−1] and MACD[t] > signal[t].

True Range at bar t: TR[t] = max(h[t]−l[t], |h[t]−c[t−1]|, |l[t]−c[t−1]|). ATR(14) is Wilder-smoothed TR. We expose ATR both as an absolute price unit and as a percentage of the closing price (ATR% = ATR / c · 100), the latter being more useful for cross-symbol comparison.

Realized volatility is the annualised standard deviation of log returns over the lookback: vol = √(252) · stdev(ln(c[t]/c[t−1])) for daily bars. Crypto uses 365-day annualisation.

N-day high breakout: fires on bar t when c[t] > max(c[t−N], …, c[t−1]) and v[t] ≥ k·avg(v[t−N:t]) (default k = 1.5). The volume gate filters out the typical illiquid grind-up that gets caught by pure price breakouts.

Volume spike: fires when v[t] ≥ k·avg(v[t−N:t]) with N = 20, k = 3 by default. Used as a screener filter and as a heads-up signal in the AI commentary.

Gap up / down: (o[t] − c[t−1]) / c[t−1] ≥ X% (gap up) or ≤ −X% (gap down). Default thresholds: ±2% on daily, ±0.3% on intraday.

Signals are deterministic functions of price/volume history. They do not embed any forward-looking opinion, and they are not the output of a trained model. They cannot account for fundamental catalysts (earnings, M&A, regulatory action) that occur between bars.

A signal firing is not a buy or sell recommendation. Drogo's screener and AI surfaces use signals as one of several inputs; the user is responsible for the trading decision.