The advent of centralized exchanges (CEXs) has significantly democratized Bitcoin (BTC) acquisition. Understanding the intricate dynamics between CEX activity and BTC price movements is paramount for navigating the complex cryptocurrency market.

The Ÿ CEX Robo-advisor offers a data-centric approach to analyzing these interactions. By employing a robust fair value model and examining the distribution of returns based on CEX feature contracts. Users can leverage this comprehensive analysis to assess CEX market activity and its potential impact on BTC price trends.

Key Features

Flow of Funds Visualization

Spot and Future volume

Real-Time CEX Tracking

Statistical Analysis

Comparative Analysis

Initial Database Structure

The CEX Advisor leverages a database of around 10.000 datapoints updated every 6 hours.

Framework

The Ÿ CEX Robo-advisor offers a daily analysis of CEX flow data, providing users with valuable insights into potential market movements. This analysis is presented in three key sections:

CEX State

Basic Info

The date and time of the last report from Ÿ Cex advisor

Model Output

This section provides a breakdown of what the model expected, the actual price and the different between the 2, the Residual value.

We then present how overvalued or undervalued the current price is, this is given by taking the expected value - the actual value.

Green: Undervalued

Red: Overvalued

Features

This section provides you with some of the features used for the fair value model.

Spot Volume

Futures Volume

Funding Delta

Open Interest

Liquidations

Strategy Indicator

CEX Indicator Graph

This graph is representative of the table below, it lets us know that at specific dates the residual value (was it buy or sell signal). The line joining these dots represents the actual price of BTC at the time.

At the bottom of the graph, we see the residual value over time compared to various dates

Strategy

This section provides a breakdown of how "Residual" is calculated. It also breaks down how to interpret the Residual data if it is overvalued or undervalued.

We also provide a handy breakdown of when it is a buy or sell signal. Essentially, if there is 1.5x deviation (positive or negative) from the residual results. Simply put, Red is overvalued (Sell) and Green is undervalued (buy).

Date

Here users are provided specific dates with the residual values for those days. As stated above, the colour-coded circles help identify under or overvalued values. The percentage presented helps provide an idea of how different in % the residuals are

ŸRA Summary

ŸRA Graph

Shows the distribution of the different predicted value vs actual value returns on the various days. Red indicated values from last week and Green is for Train data. Kindly note the colors for features below have nothing to do with whether or not it's positive or negative for price.

Strat Stats

The strategy is buy if residual above 1.5 standard deviation, sell when below 1.5 standard deviation. The stats are calculated using a dollar cost average strategy buying/selling 5% of total available balance on each signal.

Return is the "Total Return on Investment" for the strategy during the test period

Winrate is representative of the number of wins divided by the total trades

AVG_win & AVG_loss is representative of the average win &/or loss ratio for the strategy

Sharpe is representative of how much return you are getting versus the amount of risk you are taking on. Read More on Sharpe Ratio

Sortino focuses on the downside potential of the investment. It measures the return of an investment relative to the "bad" risk associated with it. Read More on Sortino Ratio

Model Stats

Here we get a look at some of the nitty-gritty details surrounding the model used.

Type: Here we present the type of model we are using to , we are using the Ordinary Least Squares regression model "ols_regression"

MSE (Mean Squared Error): This stat here provides the users with the accuracy of the model in question

R2 (R Squared): This measures how well a statistical model explains the variation of a dependent variable; essentially, it tells you how much of the outcome can be predicted by the inputs.

Datapoints & Features: Features are representative of the input variables used to train/test the model, each datapoint has several features.

ŸAI does not endorse the use of leverage, buy signals are spot DCA'ing only with a max limit of 100% of the balance invested for strategy statistic calculation.
Also note that trading strategies do not consider or suggest shorting. Sell signals are only used to calculate the dollar cost average out of open spot positions.