Дополнительные рекомендации по использованию утилиты:
1. Количество свечей, взятых как эталонный участок, оптимально не более 30, а если вы хотите получить прогноз на более длительную перспективу, то лучше переключиться на старший таймфрейм.
2. Чем меньшее количество свечей в эталонном участке, тем более узким можно настраивать канал сверки (минимальное значение для входной переменной Channel равно 0.5), и тем более точным будет получаться прогноз.
3. Если процесс поиска синхронного экземпляра продолжается слишком долго (порядка 30 мин и более), то необходимо немного расширить канал сверки, то есть увеличить значение входной переменной Channel на 0.1.
4. Если на графике явно восходящая тенденция, то желательно, чтобы начальная свеча эталонного участка была восходящая и совпадала с каким-либо ближайшим нижним экстремумом (по цене открытия).
5. Если на графике явно низходящая тенденция, то желательно, чтобы начальная свеча эталонного участка была низходящая и совпадала с каким-либо ближайшим верхним экстремумом (по цене открытия).
6. Прогноз, полученный из явно флетового эталонного участка, как правило непригоден для использования.
7. Прогноз, отображающий флетовое состояние, также непригоден для использования.
Additional recommendations for using the utility:
1. The number of candles, taken as a reference area, is optimal not more than 30, and if you want to get a forecast for a longer term, it is better to switch to a higher timeframe.
2. The smaller the number of candles in the reference area, the narrower the reconciliation channel can be set (the minimum value for the input variable Channel is 0.5), and the more accurate the forecast will be.
3. If the process of searching for a synchronous instance lasts too long (about 30 minutes or more), it is necessary to slightly expand the reconciliation channel, that is, to increase the value of the input variable Channel by 0.1.
4. If on the chart is clearly an upward trend, it is desirable that the initial candle of the reference area is upward and coincides with any nearest lower extremum (at the opening price).
5. If on the chart is clearly a downward trend, it is desirable that the initial candle of the reference area is downward and coincides with any nearest upper extremum (at the opening price).
6. The forecast, obtained from explicitly to the flat reference area, is usually unusable.
7. The forecast, showing the flat state, is also unusable.
I'm using MC Predictor, Box Profile and DSD, thanks for those greats indicators.
I would like to suggest that you use threads or even GPU to improve MC Predictor performance, I've an workstation with a Xeon E5 1680 V3 processor with 64GB RAM but I've to set MC to 16 candles and improve channels to 0.2 to run MC predictor.
I'm using MC Predictor, Box Profile and DSD, thanks for those greats indicators.
I would like to suggest that you use threads or even GPU to improve MC Predictor performance, I've an workstation with a Xeon E5 1680 V3 processor with 64GB RAM but I've to set MC to 16 candles and improve channels to 0.2 to run MC predictor.
Best regards,
Roger.
Hello.
Thank you for your interest in my products.
Indeed, the indicator MCP does not work fast, but I put a lot of effort into optimizing the code to make the indicator work as fast as possible, and I reached the limit.
In addition, not all programs can be accelerated by increasing the power of the computer, because there are algorithms in which it is impossible to start performing the next action until the result of the previous action is known.
The simplest example is - in the expression (a+b)*c multiplication cannot be performed before the result of the parenthesized expression is known.
Therefore, the number of processor cores and the amount of RAM do not always contribute to the proportional acceleration of programs. ))
Perhaps, for some algorithms, it is better to create hardware (a highly specialized microchip), and then these algorithms will be executed at lightning speed, but this is a topic of a separate conversation that is not related to programming for a PC. ))
Hi Evgeny.
I'm using MC Predictor, Box Profile and DSD, thanks for those greats indicators.
I would like to suggest that you use threads or even GPU to improve MC Predictor performance, I've an workstation with a Xeon E5 1680 V3 processor with 64GB RAM but I've to set MC to 16 candles and improve channels to 0.2 to run MC predictor.
Best regards,
Roger.
Hi Evgeny.
I'm using MC Predictor, Box Profile and DSD, thanks for those greats indicators.
I would like to suggest that you use threads or even GPU to improve MC Predictor performance, I've an workstation with a Xeon E5 1680 V3 processor with 64GB RAM but I've to set MC to 16 candles and improve channels to 0.2 to run MC predictor.
Best regards,
Roger.
Hello.
Thank you for your interest in my products.
Indeed, the indicator MCP does not work fast, but I put a lot of effort into optimizing the code to make the indicator work as fast as possible, and I reached the limit.
In addition, not all programs can be accelerated by increasing the power of the computer, because there are algorithms in which it is impossible to start performing the next action until the result of the previous action is known.
The simplest example is - in the expression (a+b)*c multiplication cannot be performed before the result of the parenthesized expression is known.
Therefore, the number of processor cores and the amount of RAM do not always contribute to the proportional acceleration of programs. ))
Perhaps, for some algorithms, it is better to create hardware (a highly specialized microchip), and then these algorithms will be executed at lightning speed, but this is a topic of a separate conversation that is not related to programming for a PC. ))