Understanding Signals
Signals are the building blocks of every strategy in MangroveAI. A signal is a function that evaluates market data and returns true or false. When you combine signals into a strategy, they tell the system when to enter a trade and under what conditions.
This guide covers what signals are, the two signal types, how they compose into strategies, and how to tune parameters for better results.
What is a signal?
A signal takes a window of historical price data (OHLCV bars) and a set of parameters, then returns a boolean:
true — the condition described by the signal is currently metfalse — the condition is not met
For example, the rsi_oversold signal checks whether the Relative Strength Index is below a given threshold. If RSI is at 25 and your threshold is 30, the signal returns true.
MangroveAI provides 136 active signals across five categories:
| Category | Count | Examples |
|---|
| Momentum | 26 | rsi_oversold, stoch_oversold, williams_r_oversold |
| Trend | 38 | sma_cross_up, ema_cross_up, macd_bullish_cross, adx_strong_trend |
| Volume | 22 | obv_bullish, mfi_oversold, vwap_above |
| Volatility | 10 | bb_upper_breakout, atr_high_volatility, kc_upper_breakout |
| Patterns | 40 | bullish_engulfing, hammer, doji, morning_star |
TRIGGER vs FILTER
Every signal is classified as exactly one of two types. This classification determines how the signal is used in a strategy.
TRIGGER signals (66 active)
A TRIGGER detects a discrete event — something that just happened at this specific bar.
Think of TRIGGERs as answering the question: “Did something just happen?”
Common patterns:
- Crossovers — a fast moving average just crossed above a slow one (
sma_cross_up)
- Breakouts — price just broke above the upper Bollinger Band (
bb_upper_breakout)
- Reversals — MACD just crossed its signal line (
macd_bullish_cross)
TRIGGERs are typically true for only one or a few bars, then flip back to false. They fire at a specific moment.
FILTER signals (70 active)
A FILTER checks an ongoing condition — whether the market is currently in a certain state.
Think of FILTERs as answering the question: “Is the market in the right state?”
Common patterns:
- Price position — price is currently above the 50-period SMA (
is_above_sma)
- Overbought/oversold — RSI is currently below 30 (
rsi_oversold)
- Trend strength — ADX is above 25, indicating a strong trend (
adx_strong_trend)
- Volume confirmation — On-Balance Volume is trending upward (
obv_bullish)
FILTERs can remain true for many consecutive bars. They describe a persistent market state.
How to tell them apart
| TRIGGER | FILTER |
|---|
| Question answered | ”Did X just happen?" | "Is X currently true?” |
| Duration | Fires briefly (1-2 bars) | Can persist for many bars |
| Use | Timing — when to act | Confirmation — is the market ready |
| Naming patterns | *_cross_up, *_breakout, *_reversal | is_above_*, *_oversold, *_bullish, *_strong_trend |
The word “bullish” does not determine signal type. A macd_bullish_cross is a TRIGGER (a crossover event), while obv_bullish is a FILTER (an ongoing state). Check the signal’s signal_type field in the API response to be sure.
How signals compose into strategies
MangroveAI enforces a strict composition rule called the 1+1 constraint.
Entry rules
Every strategy entry requires exactly 1 TRIGGER + 1 FILTER:
- The TRIGGER fires when a specific event occurs (the timing signal)
- The FILTER confirms the market is in a favorable state (the confirmation signal)
- Both must be
true on the same bar for an entry to occur
This means an entry happens only when the right event occurs at the right time in the right market conditions.
Exit rules
Exit rules require:
- Exactly 1 TRIGGER (the exit event)
- Optionally 0 or 1 FILTER (exit confirmation)
Exit rules are optional. If you do not define exit signals, the strategy relies on the built-in stop-loss, take-profit, and time-based exits from the execution_config.
Why 1+1?
This constraint exists for good reasons:
- Simplicity — each strategy is easy to understand and explain
- Reduced overfitting — fewer degrees of freedom means less chance of curve-fitting to noise
- Clear intent — the TRIGGER says “when to act” and the FILTER says “is the market ready”
- Better debugging — when a backtest underperforms, you know exactly which signal to examine
Submitting more than 1 TRIGGER or more than 1 FILTER in entry rules will be rejected by the API. The system enforces the 1+1 constraint at validation time.
Example: a simple momentum strategy
Here is a concrete example. You want to buy ETH when the MACD crosses bullish and the price is above its 50-period SMA:
- Entry TRIGGER:
macd_bullish_cross — fires when the MACD line crosses above the signal line
- Entry FILTER:
is_above_sma with window: 50 — confirms the price is in an uptrend
{
"entry": [
{
"name": "macd_bullish_cross",
"signal_type": "TRIGGER",
"timeframe": "1h",
"params": {"window_fast": 12, "window_slow": 26, "window_sign": 9}
},
{
"name": "is_above_sma",
"signal_type": "FILTER",
"timeframe": "1h",
"params": {"window": 50}
}
]
}
macd_bullish_cross fires and is_above_sma is true on the same bar.
Parameter tuning basics
Every signal has configurable parameters with documented ranges and defaults. You can view these via the API:
curl http://localhost:5001/api/v1/signals/rsi_oversold
- Shorter windows (e.g., SMA window of 10) react faster but produce more false signals
- Longer windows (e.g., SMA window of 200) are slower but filter out noise
- Lower thresholds for oversold signals (e.g., RSI threshold of 20 instead of 30) trigger less often but at more extreme conditions
- Always backtest after changing parameters to see the effect on performance
Common signal combinations
These pairings are popular starting points. Results depend on the asset, timeframe, and market regime.
| Strategy Type | TRIGGER | FILTER | Idea |
|---|
| Trend following | sma_cross_up | adx_strong_trend | Enter on moving average crossover, but only when the trend is strong |
| Mean reversion | rsi_cross_up | bb_lower_breakout | Enter when RSI starts recovering from oversold, confirmed by price near lower Bollinger Band |
| Momentum | macd_bullish_cross | is_above_sma | Enter on MACD cross, confirmed by price above long-term average |
| Volume breakout | bb_upper_breakout | obv_bullish | Enter on Bollinger Band breakout, confirmed by rising volume |
Use the semantic match endpoint (POST /api/v1/signals/match) to describe your trading idea in plain English. The system will suggest signals that match your intent.
Next steps
