The Markov strategy is a backtesting trading strategy that uses a Markov regression model to predict market movements and make buy/sell decisions. Here's a breakdown of how the strategy works:
The strategy operates on a 30-minute timeframe. The populate_indicators function calculates various indicators based on the price and volume data of the assets being traded.
These indicators include: returns: The logarithmic returns of the asset's closing prices.
volumeGap: The ratio of the current volume to the previous volume.
fractHigh: The fraction of the difference between the high price and the opening price relative to the opening price. fractLow: The fraction of the difference between the opening price and the low price relative to the opening price. The strategy fits a 3-regime Markov regression model to the data, using the endog (dependent variable) as the logarithmic returns and the exog (independent variables) as the volume gap, fraction of high, and fraction of low. The model is trained and fitted to a rolling window of 2 days' worth of data. For each window, the smoothed marginal probabilities of the three regimes (up, side, down) are calculated. The populate_buy_trend function identifies buying opportunities based on the following conditions: The probability of an upward movement (prob_up) is greater than 0.8. The probability of a sideward movement (prob_side) is less than 0.5. The probability of a downward movement (prob_down) is less than 0.5. If these conditions are met, a buy signal is generated. The populate_sell_trend function identifies selling opportunities based on the following conditions: The probability of an upward movement (prob_up) is less than 0.5. The probability of a sideward movement (prob_side) is less than 0.5. The probability of a downward movement (prob_down) is greater than 0.8. If these conditions are met, a sell signal is generated. The strategy aims to capture potential market trends by using a Markov regression model to estimate the probabilities of different price movements.
These indicators include: returns: The logarithmic returns of the asset's closing prices.
volumeGap: The ratio of the current volume to the previous volume.
fractHigh: The fraction of the difference between the high price and the opening price relative to the opening price. fractLow: The fraction of the difference between the opening price and the low price relative to the opening price. The strategy fits a 3-regime Markov regression model to the data, using the endog (dependent variable) as the logarithmic returns and the exog (independent variables) as the volume gap, fraction of high, and fraction of low. The model is trained and fitted to a rolling window of 2 days' worth of data. For each window, the smoothed marginal probabilities of the three regimes (up, side, down) are calculated. The populate_buy_trend function identifies buying opportunities based on the following conditions: The probability of an upward movement (prob_up) is greater than 0.8. The probability of a sideward movement (prob_side) is less than 0.5. The probability of a downward movement (prob_down) is less than 0.5. If these conditions are met, a buy signal is generated. The populate_sell_trend function identifies selling opportunities based on the following conditions: The probability of an upward movement (prob_up) is less than 0.5. The probability of a sideward movement (prob_side) is less than 0.5. The probability of a downward movement (prob_down) is greater than 0.8. If these conditions are met, a sell signal is generated. The strategy aims to capture potential market trends by using a Markov regression model to estimate the probabilities of different price movements.
stoploss: -0.05 timeframe: 30m hash(sha256): 5b59524cebb9e4677d12c473c9f0205546acc3179ef8cb3b40dd582e6c7a4ded indicators: high volumeGap fractLow volume volumeGap fractHigh fractLow prob_down fractHigh close open low prob_side prob_up returns
Similar Strategies: (based on used indicators)
Strategy: Discord_MarkovV2, Similarity Score: 93.75%
Strategy: Discord_MarkovV4, Similarity Score: 93.75%
Strategy: MarkovV4, Similarity Score: 93.75%
last change: 2024-04-29 21:07:40