Prometheus Markov Chain

The Prometheus Markov Chain Indicator is a custom-built tool designed to predict potential future price movements using a Markov Chain approach. A Markov Chain is a statistical model that assumes the probability of moving to a future state depends solely on the current state. In this indicator, states represent price movement classifications—bullish, bearish, or neutral—and are determined based on historical price changes (percentage returns). The indicator builds a transition matrix to calculate probabilities of transitioning from one state to another, enabling traders to identify patterns and forecast likely price actions.

Core Functionality and Transition Matrix
The transition matrix is the backbone of the Markov Chain. It captures the frequency of transitions between states in the historical price data and normalizes these counts into probabilities. For example, if the price was in a bearish state and transitioned to a bullish state 3 out of 10 times, the probability of transitioning from bearish to bullish would be 0.3. The matrix is created dynamically using the stateFunc function to classify states, which can use either dynamic thresholds (highest and lowest returns over a lookback period) or a user-defined percent return threshold. Below is the snippet that updates the transition matrix:

transitionMatrix = matrix.new<float>(dimension, dimension, 0.0)for i = 0 to array.size(vec) - 2    fromState = array.get(vec, i)    toState = array.get(vec, i + 1)    transitionMatrix.set(fromState, toState, transitionMatrix.get(fromState, toState) + 1)for i = 0 to dimension - 1    rowSum = 0.0    for j = 0 to dimension - 1        rowSum += transitionMatrix.get(i, j)    for j = 0 to dimension - 1        prob = transitionMatrix.get(i, j) / rowSum        transitionMatrix.set(i, j, prob)

This snippet iterates through historical price movements, counts state transitions, and then normalizes each row of the matrix so that the sum of probabilities for all possible transitions from a given state equals 1.

How the Indicator Predicts Future States
After constructing the transition matrix, the indicator calculates the current state of the price based on the latest percentage return and then uses the matrix to compute probabilities for transitioning to other states. The state with the highest probability is predicted as the next state, which is displayed on the chart using color-coded labels: green for bullish and red for bearish. The following snippet demonstrates how the current state and predictions are calculated:

current_chng = (close - close[1]) / close[1]var int current_state = naif not use_custom_thresh    highest_chng = ta.highest(current_chng, int(size) * 2)    lowest_chng = ta.lowest(current_chng, int(size) * 2)    current_state := stateFunc(current_chng, highest_chng, lowest_chng)else    current_state := stateFunc(current_chng, custom_thresh)predicted_probs = array.new<float>(dimension, 0.0)for j = 0 to dimension - 1    array.set(predicted_probs, j, transitionMatrix.get(current_state, j))

The indicator evaluates which state has the highest transition probability (highest_prob) and places corresponding labels on the chart. For example, if the next state is predicted to be bullish, a green "Bullish" label is placed below the current bar. This predictive functionality helps traders anticipate potential reversals or continuations in price trends based on historical behavior patterns.

Usage:



Here we see the indicator at work on $PLTR. The states predicted are bullish then bearish. In this example we then see price move in a way that verifies those predictions.



On this 4 Hour AMZN chart we see predictions play out in a short trade style. States quickly move from one to another but not without giving traders a way to take advantage.

This is the perspective we aim to provide. We encourage traders to not follow indicators blindly. No indicator is 100% accurate. This one can give you a different perspective market state. We encourage any comments about desired updates or criticism!