Direct answer
Multi Indicator Confirmation is calculated by combining the outputs of several indicators into one decision variable. A common, general approach is: (1) compute each indicator’s reading from price data, (2) convert each reading into a standardized signal value (such as direction +1/0/−1 or a normalized score), (3) optionally apply weights, and (4) declare “confirmation” when the combined value meets an agreement rule (such as a threshold or a required number of agreeing indicators).
Because different platforms implement the wording “Multi Indicator Confirmation” differently, the verifiable part is the method you choose: the exact formula, the conversion rules from indicator output to signal value, the weights, the threshold, and the time window used for indicator calculations.
Mechanism or definition
1) Indicator outputs are not directly comparable
Most indicators produce outputs on different scales and units. For example, one indicator may output an oscillator centered near zero, while another outputs a moving-average difference measured in price units. A multi-indicator method therefore needs a conversion step that maps each indicator output to a shared representation.
Two widespread representation choices are:
- Direction encoding: each indicator becomes a discrete value such as +1 (bullish/positive), 0 (neutral), or −1 (bearish/negative).
- Scaled scoring: each indicator becomes a continuous score, such as a normalized number between −1 and +1, computed from where the indicator lies relative to its own typical range.
2) A lookback window and timing rule
Indicators usually require a lookback period (how many past candles/bars are used) and a timing rule (whether you use the most recent completed bar, whether you shift signals by one bar to avoid using future information, etc.). For independent verification, you must state:
- which timestamp you evaluate the combined result at,
- which bar you treat as “current” for each indicator,
- the lookback/period parameters for each indicator.
3) Agreement rule (the “confirmation” step)
Once every indicator is expressed as a standardized signal value, confirmation is calculated using an agreement rule. Common general forms are:
- Weighted sum + threshold: compute a combined score and compare it to a cutoff.
- Count of agreeing indicators: require at least K indicators to agree on the same direction.
Below is a general weighted-sum formula that matches many “confirmation” implementations in concept (you still must confirm the exact choices in your source system).
4) A general formula
Assume you combine N indicators.
- Convert each indicator output to a standardized signal value:
- (s_i \in [-1, +1]) (or (s_i \in {-1,0,+1})).
- Choose weights:
- (w_i \ge 0). Often weights are normalized so (\sum_i w_i = 1), but that is not required if you adjust the threshold accordingly.
- Compute a combined confirmation score:
- (C = \sum_{i=1}^{N} w_i, s_i).
- Apply an agreement threshold to classify confirmation:
- Example rule: confirm “positive” when (C \ge T), confirm “negative” when (C \le -T), otherwise classify as not confirmed.
If your method instead uses a count rule, you would compute:
- (A = \sum_{i=1}^{N} 1{\text{indicator } i \text{ is bullish}}) and then declare bullish confirmation when (A \ge K). A similar count can be made for bearish confirmation.
Evidence or example (how to verify the calculation yourself)
Because no live data is assumed here, the goal is to show how you can independently reproduce the math using your own historical data and chosen indicator parameters.
Example setup (fully specified assumptions)
Assume:
- You combine N = 3 indicators.
- Each indicator is converted to direction encoding (s_i \in {-1,0,+1}).
- Weights are equal: (w_1 = w_2 = w_3 = 1/3).
- Confirmation uses a threshold (T = 1/3).
- You evaluate at a specific bar time where each indicator has already completed its required lookback.
Suppose the standardized signals at that time are:
- Indicator 1: (s_1 = +1)
- Indicator 2: (s_2 = +1)
- Indicator 3: (s_3 = -1)
Compute the confirmation score:
- (C = (1/3)(+1) + (1/3)(+1) + (1/3)(-1) = 1/3).
Apply the rule:
- Since (C = 1/3) and (T = 1/3), the combined result qualifies as positive confirmation under this example.
Now change only one assumption to see the sensitivity:
- If the third indicator were neutral ((s_3 = 0)), then (C = 2/3), which remains positive confirmation.
- If instead you set a stricter threshold (T = 2/3), the earlier case would become “not confirmed.”
This illustrates what is actually being calculated: the mapping from indicator output to (s_i), then the weighted combination, then the threshold decision.
Where indicator settings enter the calculation
If one indicator uses a longer lookback, its output will lag or smooth more than a shorter one, changing (s_i) at each evaluation time. Therefore, even with the same formula for (C), the confirmation result can change purely because of the indicator period settings.
Limitations and risks (material failure modes)
1) Disagreement and neutral zones
If indicators frequently disagree (some bullish, some bearish), the combined score (C) may hover near zero and fail threshold tests. Direction encodings also create neutral states; depending on your definition of neutral, you can end up with many “no confirmation” outcomes.
2) Lag and repainting-style timing issues
Indicators based on moving averages, filters, or multi-step computations can respond with delay. If an implementation allows an indicator value to change after the bar closes (for example, due to how it sources data or updates), then the confirmation value you compute historically may differ from what would be produced in real time. Independent verification requires locking the rule: which bar you treat as final and which data is used at evaluation.
3) Noise, correlated indicators, and false agreement
Some indicators are built from similar price inputs (and may be statistically correlated). Combining correlated indicators does not guarantee independent confirmation; you may get high agreement because they all react to the same underlying move. In such cases, the confirmation metric can sound more decisive than the information content actually is.
4) Costs and execution effects are not part of the math
The calculation of (C) (based on indicator outputs) does not include costs, spreads, slippage, or execution constraints. As a result, even a method that classifies “confirmation” mathematically cannot account for trading frictions. This limits how well the confirmation concept maps to real outcomes.
5) Threshold choice is a modeling assumption
The threshold (T) (or required count (K)) is a subjective modeling choice. Different thresholds can dramatically change classification frequency.