I decided to apply my prediction logic for BTC short-time price forecast — predicting based on 4 hours of a price move and predicting the next 30 mins.
Unfortunately, this updates only when my PC is on. (eventually, I would like to migrate these computations to AWS)
How to access -
Login -> Click “Get Realtime”
I’ve asked the same questions several times — how to use this application?
So, I uploaded a guide video to explain the usage. Here is the link.
Announcement for prediction model update
The prediction model for SP500/Currency/T-Bill(Rates)/Cryptos has been updated. (Asof: Sep 9 2022 to Sep 16 2022)
Announcement for crypto enhancement (50 cryptos -> 150 cryptos)
I added another 100 crypto prediction targets to Tripole.app
Please check https://www.tripole.app/
Announcement for GUI update
Now, Tripole.app introduced Vuetify library as below.
So, as I described in the previous article, we find Fedfund rates are an effective measure to enhance prediction accuracy for T-Bill rates.
Is there anything else to improve the predictions?
First, US CPI rates as an input (shifting 2 months — since CPI in July is not available in July).
Then, I added Real M2 Money Stock(also shifting 2 months)
Here are the outcomes. Interestingly, Fedfund+CPI improves performance slightly, but Fedfund+Real M2 generates the best outcome than Fedfund+CPI+Real M2.
Now, tripole.app was updated. Using Fedfund rates and Real M2 to predict T-Bills.
T-Bill predictions are now available in tripole.app
I computed based on 12/24/36months historical records with Fedfund rates in this prediction.
Why do I apply Fedfunds rate in these predictions?
Unlike my previous article(Fedfunds rates were NOT effective input to predict SP500 shares), clear improvement was observed with Fedfund rates in this experiment.
Model update announcement — from Sep 2nd, 2022 to Sep 9th, 2022.
Computed RMSEs are as below
Also, increased crypto coverage from 20 to 50.
