To compensate for my lack of trading skills I started some time ago developing an algorithm for spotting changes and trends in crypto prices. The project started as an internal experiment but I made a decision of publishing one of the approaches I have been now using mostly.
The project is called "The Whirrel Index" which is a combination of price predictions squeezed into a single number that ranges from -100% (price likely to dip steeply) to 100% (price likely to rally high).
The prediction model consists of multiple neural networks that have been trained with technical indicators and price data. Each of these neural networks predicts trends and changes in future average prices in timeframes ranging from 1h to 24 hours. The Whirrel Index number is a single and simplified indicator that combines all these predictions into one number that speculates whether the price will drop up go up.
The data is updated hourly and the accuracy of the prediction models are tested by backtesting them with historic data. Currently the accuracy varies between 60% - 80% depending on the prediction model.
Imo fine tuning the algo is crucial but I'm also now working to find the most optimal strategy. The algo has not made me rich yet but has already helped me to avoid some losses and even make some small profit by knowing when to hold, sell or buy.
It's still in early stage, and as I'm not a professional trader, it would be super helpful to hear feedback especially from people who are more experienced with cryptos, trading etc.
In short there's a stack of prediction models that analyze the data: deep-learning neural network, keyword-based machine learning model, and shortly also I'll include a BERT NLP analyzer
I'm the maker of Product Farm - a virtual Product Hunt simulator. It uses a machine-learning algorithm (neural network) that has been trained with Product Hunt data to artificially rank products. You can use it to test how high ranking your product would get in Product Hunt before submitting. Or then just for fun.
The project is still an early experiment and many things are under construction. But I'm sharing it here already. Let me know what you think about the idea!
I'm the maker of Hacker-AI. Here's a little bit of details and background to the tool.
I have a decade-long background working as a marketing/tech consultant. I've used a similar approach in my projects to save time and remove uncertainty when choosing the content for eg. marketing campaigns or product launches. I have learned how to create tools like this first for myself, but I've also implemented similar tools for companies so that their content and marketing team can perform more effectively.
I'm not a data scientist and this tool is a result of learning from smart people and experimenting with different machine learning and NLP solutions. It uses a combination of feed-forward neural network and bag-of-words analysis to conduct the predictions. In my tests it was able to predict correctly 60%-70% oft times which variation got more points in Hacker News or upvotes in Product Hunt (when using texts that resemble the platform's style). It uses data from Hacker News API and Product Hunt API for training the prediction models.
Thanks for testing! I'll write a description of how it makes the predictions and add it to the tool.
But in short, it uses a machine learning model that I trained with a dataset that contains all stories and comments between 2006 and 2017: https://www.kaggle.com/hacker-news/hacker-news
I've tested various approaches, and currently, the algorithm takes the title as an input and transforms it into an array of numbers between 0 and 1 (each character is a number). Then I give these arrays to the machine learning model (brain.js feed-forward neural network) and the number of scores as an output. After learning and iterating over the data, it spits out the prediction model that I can use to predict the outcome of different title variations.
I've tested the algorithm with approx. 10.000 posts and it has been able to predict 60% of the cases correctly. So, it's not perfect yet, but I use this method in a situation where I don't have any experience of which type of title would work + I don't have time to do "proper" pre-testing.
To compensate for my lack of trading skills I started some time ago developing an algorithm for spotting changes and trends in crypto prices. The project started as an internal experiment but I made a decision of publishing one of the approaches I have been now using mostly.
The project is called "The Whirrel Index" which is a combination of price predictions squeezed into a single number that ranges from -100% (price likely to dip steeply) to 100% (price likely to rally high).
The prediction model consists of multiple neural networks that have been trained with technical indicators and price data. Each of these neural networks predicts trends and changes in future average prices in timeframes ranging from 1h to 24 hours. The Whirrel Index number is a single and simplified indicator that combines all these predictions into one number that speculates whether the price will drop up go up.
The data is updated hourly and the accuracy of the prediction models are tested by backtesting them with historic data. Currently the accuracy varies between 60% - 80% depending on the prediction model.
Imo fine tuning the algo is crucial but I'm also now working to find the most optimal strategy. The algo has not made me rich yet but has already helped me to avoid some losses and even make some small profit by knowing when to hold, sell or buy.
It's still in early stage, and as I'm not a professional trader, it would be super helpful to hear feedback especially from people who are more experienced with cryptos, trading etc.