Currently I have developed a new model (not technical nor fundamental) that able to predict the price direction for particular stock in the next trading day with accuracy > 60%. For example, the close price for stock ABC is $105 today, the model will predict if ABC will close higher or lower than $105 for next trading day, e.g. "close lower" means the stock will close < $105 in the close of next trading day and vice versa. The model will make the prediction about an hour before market close today. I have randomly select 4 popular stocks and will start live demo on this. FB (FaceBook) MSFT (Microsoft) AMD (Advanced Micro Device) INTC (Intel) Let have fun on this
10/9 Prediction INTC - close lower AMD - close lower FB - close lower MSFT - close lower Note: "INTC - close lower" means the model predict INTC will close lower than the current close price ($52.82 as in 10/9) on next Monday close (10/12)
Calvin, Looking forward to your journal, thanks for bringing it here. Alt methods are interesting to me. How about teasing us with a little about the inputs to your model?
Your model will need some Fine Tuning or more data to be exploitable. For example, by what percentage do the 60% close higher and by what percentage do the 40% close lower? 60% +1 = 6 40% -2 = 8 =(2)
I agree. It's not how many times you are correct that matters; it's how much you make when you are correct vs how much you lose when you are not.
I have to keep the model as "trade" secret. Inputs will be from different sources (not just purely standard close/open/high/low price)