How Signals Work

The concepts behind Signals AI — read this once and the rest of the product makes sense.

The signal lifecycle

Draft (Canvas) → Saved → Deployed (Forward Test or Algo Trade)
                              │
                         Listening ⇄ Paused
                              │
                     Entry fires → position open → Exit fires → back to Listening

1. Draft — you build and preview the strategy on the Canvas.
2. Saved — the strategy lives in Saved Signals; nothing is running.
3. Deployed — an instance of the strategy runs in Forward Test or Algo Trade. One saved strategy can have multiple instances.

The Listening state

A deployed strategy in Listening is actively monitoring live market data for your entry conditions.

• It stays listening day after day — no daily re-activation needed. (For NSE/BSE brokers you still need to log in to your broker daily; see Algo Trade prerequisites.)
• Pause Listening suspends monitoring; Start Listening resumes it.
• When an entry fires, the strategy manages the position until an exit fires, then resumes watching for the next entry — subject to your signal limits.

When are conditions evaluated?

Evaluation depends on the Trade Monitoring mode in your execution settings:

• Candle Close — conditions are checked once per candle, on the completed candle's values. An "EMA crosses above EMA" is confirmed only when the candle closes. Slower to react, but immune to intra-candle fakeouts.
• LTP (Last Traded Price) — conditions are checked on live ticks, so signals can fire mid-candle. Faster, but a condition that's true mid-candle can be false again by the close.

This distinction is the single most common reason live behavior surprises people — a backtest evaluated on closed candles won't show entries that LTP monitoring would have taken mid-candle, and vice versa.

Why backtest, forward test, and live results differ

Differences between the three are normal. The main causes:

Cause	Effect
Slippage	Backtests assume a fill price; live market orders fill at the touch — usually slightly worse
Order type behavior	Limit orders can go unfilled live; backtests don't model missed fills
Evaluation timing	Candle-close vs LTP monitoring (above) changes which signals fire
Latency	Live signals take real time to reach the exchange
Data differences	Historical and live feeds can differ in ticks near the trigger price

Full write-up: Why live trades differ from forward test and backtest results

tip

Small, consistent differences are expected. Large or systematic divergence (e.g., live always enters a candle later) usually points to a monitoring-mode mismatch — compare your execution settings with your backtest assumptions, or debug the strategy.

Signals vs Signals AI (Classic vs AI)

AlgoTest has two signal products — don't mix up their docs:

	Signals (Classic)	Signals AI (these docs)
Strategy logic lives in	TradingView / Chartink / your code	AlgoTest (Canvas)
Needs a TradingView paid plan	Yes (Essential+)	No
Signal delivery	Webhook alerts into AlgoTest	Built-in — no webhooks
Backtesting	Via TradingView	Built into AlgoTest