Bitcoin Forum
December 11, 2024, 11:36:41 AM *
News: Latest Bitcoin Core release: 28.0 [Torrent]
 
   Home   Help Search Login Register More  
Pages: [1]
  Print  
Author Topic: [ANN] [CoindicatorBot] Machine Learning Based CryptoCurrency Signal Bot  (Read 390 times)
Coindicator (OP)
Newbie
*
Offline Offline

Activity: 5
Merit: 0


View Profile
October 27, 2017, 02:21:25 AM
Last edit: October 27, 2017, 03:02:51 AM by Coindicator
 #1

Forecasttastic!

Nothing acts faster than a good forecasting model. With the support of cloud computing, CoindicatorBot uses cutting-edge machine learning algorithms to screen hundreds of technical indicators 7/24 for all digital currencies and provides the highest quality trading signals in crypto markets.



CoindicatorBot utilizes XGBOOST, a cutting-edge machine learning algorithm that has won many data science competitions, to forecast the price increase based on a model trained from a big dataset of hundreds of technical indicators collected over several months.

You can check out the website for more information about the project.



Follow us from Twitter https://mobile.twitter.com/CoindicatorBot
Coindicator (OP)
Newbie
*
Offline Offline

Activity: 5
Merit: 0


View Profile
October 27, 2017, 02:45:20 AM
 #2

XGBOOST : A Unique Machine Learning Tool

- A Glimpse of Algorithm for Short Term Signals -

CoindicatorBot utilizes XGBOOST, a cutting-edge machine learning algorithm that has won many data science competitions, to forecast the price increase. The forecasting model used by CoindicatorBot for the short term buy signal is specifically designed to predict that a coin’s price is going o increase by at least 3% within the next 3-6 hours time frame after the signal. The forecasting model was calibrated by using a training dataset collected over more than 2 months. The selected features in the training dataset included the following list of variables:

  • The percentage change in buy volume for the past 15, 30, and 60 minutes compared to the average buy volume in the past 24 hours

Following values are computed repeatedly using 15-min, 30-min, 60-min, and 120-min candle stick data (OHLC):

  • The slope of buy volume in the past 4 periods
  • The slope of buy volume in the past 96 periods
  • Stoch RSI Fast K value
  • Stoch RSI Slow D value
  • The moving average of Stoch RSI Fast K
  • The slope of Stoch RSI Fast K in the past 4 period
  • Chaikin Money Flow Index
  • ADX value
  • The slope of ADX in the past 4 periods
  • The sign of SQZ Momentum Index (0:negative, 1:positive)
  • The slope of SQZ momentum index in the past 4 periods
  • The sign of SMI (1: bigger than signal, 0: smaller than signal)
  • The value of SMI
  • The sign of MACD(1: bigger than signal, 0: smaller than signal)
  • The value of MACD
  • On Balance Volume value
  • The slope of OBV in the past 4 periods
  • Ichi cloud indicators (lagspan, baseline, conversion line, Span A, and Span B)
  • Aroon indicator value
  • Slope of Aroon indicator
  • Chaikin Volatility Index
  • The sign of KST (1: bigger than signal, 0: smaller than signal)
  • The slope of KST in the past 4 periods
  • The sign of TDI indices
  • Commodity Channel Index
  • Total volume
  • Buy volume in the past 15, 30, and 50 minutes
  • Percentage change in price in the last 15, 30, 60, and 120 minutes.
  • The total amount of buy orders for the first 50, 100, 150, 200, 250, 300, 350, 400, 450, and 500 buy orders in the tradebook
  • The total amount of sell orders for the first 50, 100, 150, 200, 250, 300, 350, 400, 450, and 500 sell orders in the tradebook
  • Total buy orders in the tradebook
  • Total sell orders in the tradebook
  • Buy Order/Sell Order Ratio for the first 50, 100, 150, 200, 250, 300, 350, 400, 450, and 500 orders in the tradebook
  • Buy Order/Sell Order Ratio for the whole tradebook
  • Number of buy and sell walls in the tradebook
  • Spread of tradebook
  • Weighted and unweighted midprice in the tradebook

In addition to these features, the price increase/decrease within 3 hours was also computed after these indicators collected. The XGBOOST model relates these 169 technical indicators to the price increase/decrease within next three hours using past data and reveals complex relationships among all these indicators and the price increase/decrease within next 3 hours.

The bot runs the algorithm every 10 minutes and computes the same indicators for all coins, then use the calibrated model to predict whether or not the price will increase by at least 3% within next 3-6 hours time window. If the prediction score assigned to a coin is above a certain threshold, a detailed technical PDF report is created and the signal is sent to all subscribers at the same time. Note that these signals have probabilistic nature, so it is not always guaranteed that the price will increase as predicted.
Coindicator (OP)
Newbie
*
Offline Offline

Activity: 5
Merit: 0


View Profile
October 27, 2017, 02:52:27 AM
 #3

FREQUENTLY ASKED QUESTIONS

What type of signals CoindicatorBot is sending?

CoindicatorBot send BUY signals based on a machine learning algorithm, XGBOOST.  There are two types of signals CoindicatorBot sends: a short-term signal and a long-term signal.

Short Term Signals: The short-term signal predicts that price will increase by at least 3% WITHIN 3-6 hours after the signal. These signals are not valid after a couple hours. When the signal is sent out, it is because XGBOOST model predicts that there is a high probability the price will go up WITHIN next 3-6 hours. Why the term WITHIN is emphasized? Because, it is typically misunderstood. Keep in mind that there is not a certain time frame about when the model predicts it will go up. It can go up within 15-20 minutes just after the signal , or it can go up after a few hours. The model by design only predicts it will go up WITHIN 3-6 hours. Also, once the goal is achieved (3% increase is reached), the signal does not guarantee what would happen after that point. So, the price may go down just after increasing by at least 3%.

Note that these signals are probabilistic in nature. There is a probability-like score computed by the model for each coin every 5 minutes. If this score is above a certain level, then the signal is sent out. Since these are model-based probabilistic predictions, there is no guarantee the price will increase at the expected level for all signals. So, there will be some bad signals. If the signal is bad, the price may not go up by 3% or may not even increase at all.

My goal is to achieve 80% accuracy in my signals. It means 80% of the signals will see a 3% increase in price WITHIN the three hours after the signal. This is a goal I have not been able to achieve but it’s close. I constantly work on to improve the model performance and to achieve the goal.

I publish weekly performance reports if you want to check how it performed in the past study the signals to develop a trading strategy for yourself. However, note that past success does not guarantee future success for constantly changing market conditions. I do my best to maintain the model so it keeps accuracy rates as high as possible.

Long Term Signals: I am also working to develop a long term signal which will predict the price increase of at least 30%-50% within next 24 hours after the signal. This is under development now, and CoindicatorBot will send these signals when the model is ready. It will be announced when it is ready. The expected timeline is sometimes around December 2017.

Does CoindicatorBot send Sell Signals or tell its subscribers when to sell?

No, CoindicatorBot does not send sell signals or does not tell when you should/must sell.

What exchanges CoindicatorBot is sending signals?

CoindicatorBot currently sends signals only for Poloniex. The model for Bittrex is currently under development, and I expect to start sending Bittrex signals sometime in late December 2017 or early January 2018. I also plan to extend the service for other exchanges if Polo and Bittrex signals are successful.

How do I subscribe to your service?

The signals are currently delivered using the Telegram messaging app. You should first install the Telegram app to your computer or phone. Once you install the Telegram app, you can go to the website for the link to start interacting with the bot. First, you choose your language and the bot will give you the instructions for several commands.

You can use /subscribe command and choose the 1-week trial to start receiving signals. Once your free trial expires after a week, the bot will stop sending the signals and  inform you that the free trial period ended by sending a notification message. If you like the service and continue receiving signals, you can again use /subscribe command and choose a paid membership option. There are currently two membership options:

-1 month membership ($5), and

-3 month membership ($12).

Once you choose a paid membership period, the bot will compute the equivalent amount of XRP at the current price and tell you what amount of XRP to send to what address and using which Tag number. You will have 2 hours to send the payment. Do not forget to write your Tag ID along with the address when sending the payment. When your payment is complete and arrives to the the given address, the bot will send you a notification message indicating that your membership is approved and you will start receiving signals.

You can use other commands such as /lang for changing language settings (12 different languages are available), /remaining to learn when your active membership ends, and /support to get contact information and your unique membership ID if you have any issues about using the bot.

Once I subscribe the service, what kind of signal I will receive? What information is included?

Below is a screenshot for a typical signal message you will receive via Telegram. The signal message includes some quick snapshot of volume and price activity in the last hour, a link to a detailed Technical report, and a link to the exchange website for the coin signaled.

http://www.coindicator.io/wp-content/uploads/2017/09/IMG_0167.png

For instance, the below signal indicates that the total buy volume for $BELAis 1.03 BTC in the past 15 minutes  and this is a 2819.7% increase compared to the average 15-minute buy volume in the past 24 hours. Similarly, the total buy volume is 1.05 BTC in the past 30 minutes and this is a 1389.29% increase compared to the average 30-minute buy volume in the past 24 hours. Finally, the total buy volume in the past 60 minutes is 1.11 BTC and this is a 686.12% increase compared to the average hourly volume in the past 24 hours. The signal also tells that the price did increase by 3.18% in the past 15 minutes, by 0.66% in the past 30 minutes, 0.55% in the past hour.

There is also a link for a more detailed technical report. All technical reports are uploaded to Dropbox and these links are publicly accessible. Subscribers do not have to have a Dropbox account to open the report.  The link will take them directly to the report. You may find it useful to download the official app for Dropbox if you don’t want to open the reports on browser.

What information is included in the technical report?

The technical report is prepared in a standard format for all signals and mainly includes graphics. Below is a list of things included in a technical report for any signal.

– The volume and price activity in the past 15, 30, and 60 minutes and their comparison to corresponding averages in the past 24 hours.

– Plots for market depth analysis. These plots show the total BTC amount of buy and sell orders for the first 100, 200, 300, 400, and 500 orders in the tradebook at the time of the signal, and also their ratios.

– A 15-minute candlestick plot for the past 24 hours and a 4-hour candlestickplot for the past 2 weeks with technical indicators such as  MACD, SMI, SQZMOM, ADX, StochRSI, Cumulative Buy Volume, On Balance Volume Oscillator, Chalkin Money Flow.

– Separate candlestick plots for the same time frames with Ichi Cloud indicators.

Would you make me rich if I strictly follow your signals and buy after every signal?

No, you will probably end up losing all your investment. If you are here to blindly follow the signals for trading, you should stop reading now, forget about this project, and go find something else to do. Predicting future price in cryptocurrency market is a very difficult task because there is a lot of outside factors and noise that statistical models cannot perfectly handle and of course sometimes pure manipulation of market.

Although, I am trying to improve the performance, there will be always false signals and these false signals may sometimes make you lose a lot of money. That is why a detailed technical report is attached to every signal. If you don’t know what these indicators are, you should not be trading at all. I would suggest that you study these indicators (please ask Mr. Google) and then also study each signal with the report coming along with it for a while. Then, you may develop some sense and experience about how to use this tool for your benefit.

Please make sure you read and accept the Terms of Service on the website. These signals are for information only. They are not, and should not be regarded as “investment advice” or as a “recommendation” regarding a course of action. If you don’t agree with these terms, please do not use or subscribe to any service provided by this website.
Pages: [1]
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.19 | SMF © 2006-2009, Simple Machines Valid XHTML 1.0! Valid CSS!