Crypto Token Analyzer: Filters, Indicators and a Safety Check

Overview

This application is a search and analysis tool for cryptocurrency tokens. Dozens of new tokens are created every day on different networks and checking them by hand is not possible; this application takes the new and trending tokens, filters them with rules I set, runs price and indicator analysis on what is left, and shows a short list of the ones that pass the chosen conditions. I built this project in 2024. The application is an analysis and screening tool, not financial advice and not a guarantee of profit.

The pictures on this page were taken from running the application on live market data.

What the application does

• Analyse one token: you give the network and the address of a token, and the application shows the candlestick chart and the analysis in several time frames (1, 5 and 15 minutes and 1 hour) side by side.

• New and trending tokens: it fetches the list of new pools and of trending pools from many networks (about 200), including Solana.

• Filters: minimum market cap and liquidity, the age of the token (for example "between one day and ten years"), the number of buyers, the price change over five minutes, one hour, six hours and twenty-four hours, and the trade size (to calculate the price impact of a small trade).

• Analysis and screening: each token is measured with several tests, and only the tokens that pass all the enabled tests go into the list. Each test can be switched on and off separately.

• Saved settings: all the filters and tests for both modes (new and trending) are kept in one settings page and a database.

The diagram below shows the workflow.

The analyses I implemented

• Linear regression on the closing price and a forecast of the next few candles.

• Exponential moving average (EMA) over the short term and comparing the price with it.

• The Prophet time-series model to forecast 15 minutes ahead.

• RSI and MACD with the ratio of the last value to the average of several recent candles, to see whether the trend is gaining speed.

• Price and volume trend: the average of the last five candles against the whole period.

• Empirical probability of up and down: the share of candles that went up or down.

• A percentage stop-loss and the price impact of a small buy on the pool, to judge risk.

• Each test can be switched on or off in the settings, and the combination of them makes the condition to enter the list.

Token safety check (anti-scam)

New tokens are full of scams and "rug pulls". So I also wrote a safety check that fetches the report of the token from the RugCheck service and rejects it if: one holder owns more than 20% of the token, the locked or burned liquidity of the pool is below a floor (an amount and a percentage), or the risk score is high. New tokens from the DexScreener profiles also pass this check.

Reading straight from the Solana chain

For Pump.fun tokens (a platform for making tokens quickly on Solana), I studied and tested how to listen for token creation from the chain itself: I listen to the logs of the Pump.fun program, catch the moment of creation (the mint instruction), parse the transaction and get the address of the token and of the pool, so I can check the token from its very first second. For this part I used open-source examples and studied them. I also wrote a small tool to see the balance of wallets on Solana.

Collecting and storing

• The data of trending pools and price candles (OHLC) is stored in an SQLite database so it can be analysed later.

• A separate archiver takes the data at intervals and makes CSV files to test models.

• To survive API errors, it retries with a delay and a random wait.

Design

A dark and simple interface, with a financial candlestick chart (Chart.js with the Financial extension) and a table that shows each indicator in green or red. The application can be installed on a phone like an app (PWA).

Tools

• Python, Flask, SQLite, Pandas, NumPy

• scikit-learn, Prophet, the ta library (RSI and MACD)

• Chart.js with a candlestick chart, PWA

• The GeckoTerminal, DexScreener and RugCheck APIs, Solana Web3 (Node.js and TypeScript)

Result

A tool that automates the repetitive work of finding and screening new tokens, with adjustable filters, several price analyses and a safety check. Its output is a short list for further review, and the final decision stays with the user.