Create your own neural networks to predict market values using a very easy no-code interface and join us in creating the largest distributed AI to consistently beat the market. Or just browse the predictions we publish...
I like the concept. Can you help with a clarifying question? Data is as important as the AI models, when making predictions. How do you get the data that lets your AI predict the stock movements?
@perdamgaard I use an API that provides me with the historical data.
It's quite expensive, so I have a proxy caching the requests I made, to avoid ask for the same data more than once.
@antonio_fernandez_porrua
OK - got it. It is interesting if this can give you a high enough confidence level. Have you thought about also including economic parameters?
@perdamgaard Yes, I would need to find a source, and they might not be available for every resolution.
Also new data sources mean new data parsers and mappers, but I think it is worth it.
Those are the hypothesis that I need to validate with the market:
* A lot of market and crypto investors are inclined to do their own analysis rather than blindly trust third-party analysis. Those investors would be creating their own AIs if they had the chance. (like a no-code tool)
* For this AI problem, a combination of hundreds of small AI's predictions can be more accurate than the prediction of a big AI.
* I need to offer the tool for free because big data analysis can give me much more value than users' subscriptions for premium features.
So what do you think about the product? and the hypothesis?
Report
@antonio_fernandez_porrua Great concept! Since you're using weighted average, the aggregate prediction isn't as useful as it could be. But it's a great start.
Antonio, hi there.
I'm not a technical person. I can only ask/advise on strategic & business issues. But still, wonder whether you've already validated your hypotheses? As I can see, you have some data now from different users. Is aggregated data accurate?
Anyway, it all makes sense. I'll be glad to hear more professional points of view than mine) I will be following the thread. From a business perspective, the idea has a lot of potential.
@yaroslava Aggregated data is not very accurate yet. Currently it's just weighted mean.
I have in the road map to replace it.
Report
Very interesting, tried it out a little but started to wonder, is data from the stock market, like price movements of different instruments the data you can use as auxiliary data, or is there some other data available?
It would make it easier to use if would somehow specify what is available in the search.
@aleksi_halsas1 yes, it's price movements. The thing is you can create your IA to predict ethereum using ethereum price and using bitcoin price too, because you may think that they correlate each other.
Instead of using bigger timeframes, using wider datasets. Because maybe in order to predict the tomorrow price of SYMBOL1, the yesterday price of SYMBOL2 and SYMBOL3 give us more information than the price of SYMBOL1 one year ago.
Looks amazing, I really like the concept. The searching bar looks like it's not working, tho... you might want to check it out. I put only "variable = high" and I did not fill any other inputs and I got the "nothing found" message.
@stzdio Thanks.
My fault. The way I did it, the symbol field should have something. I need to hide/disable the button when it's empty, or use it as a filter of the feed when empty so:
You enter a symbol or text -> it searches.
You enter nothing -> it filters the feed.
I'm creating the issue in github right now.
Thank you very much.
Kaddara
exnge
Kaddara
exnge
exnge
Blend Social
exnge
exnge
Neomad
exnge
exnge
Neomad
exnge
exnge