CRYPTOLEARNING: A NEW STAGE FOR OUR FORECASTING SERVICE
Forecasting market basing on seasonality is one of the great goals of cryptoforecast project. Our Pillar algorithm is working really well and it is going to be improved (see https://medium.com/@Forecasts/pillar-my-dear-pillar-great-upgrades-incoming-to-our-algorithm-cryptoforecast-team-has-a-dream-7140b793c18f).
But, in order to give more freedom to our users, we are going to introduce 12 new forecasting algorithms based on seasonality that will greatly help our users in order to diversify their analysis.
Together with those 12 algorithms we will introduce another seasonality model, the Autoregressive Integrated Moving Average (ARIMA). See here for other information (https://en.wikipedia.org/wiki/Autoregressive_integrated_moving_average)
This is cryptolearning.
These algorithms will be usable for generating charts and will also have some customization option. The most important for all will be:
1) A Benchmark selection option. “Benchmark” is a past time period that is used by algorithm for an experimental forecasting, using dataset information. The algorithm whose prevision for the benchmark period is more similar to effective price in that period is probably a good model, and so because of this it is very useful for forecasting. So Benchmark gives to users an effective information about the expectable accuracy level of the algorithm.
2) “History” and “forecast” to select the length of historical database used as base for forecasting analysis and the future period to forecast. These options will be similar to those of Pillar, but with a bigger freedom in choosing “history” (for pillar you only have three possibilities: 2, 5 or 7 year of base).
3) Timeframe and pairs. These 12 algorithms will probably be available on BTC/USD, ETH/BTC, ETH/USD, LTC/BTC, LTC/USD, XMR/BTC, XMR/USD, XRP/BTC, XRP/USD, DASH/USD, DASH/BTC pairs.
They will be based on the same dataset of Pillar.
In a first phase, Cryptolearning will be fully manual. In Q2, we will provide to our users the possibility of automatically regulate some parameters of cryptolearning. In fact, all the 12 algorithms will have some particular parameters — they are very different from one another.
Prior to the publication of the new webapp, we will release some detailed guidelines about each of these 12 algorithms. This will greatly increase the analytic usefulness of these new instruments.
Of course, these algorithms can be useful if used together with other analysis tools — for example fundamental analysis, or Pillar, or Technical Analysis (if you use it). But we believe that a trader with more analysis tools is a more efficient trader. As soon as we can, once the new webapp will be released, we will also provide guidelines about how to better use all our tools (it will be a very interesting research work for all our team). These upgrades will greatly impact on the future of Cryptoforecast and will lead it to a new development stage. It is time to think big!
