Yeah, I tried this out. It’s a very good post. The thing I was most surprised about is how little you have to mess around with k means to get it to classify candles properly.
hey clark, just wanted to say thanks. this should be one of the more populated threads here on BP. obviously there arent enough “1/100’s” on here
any chance you could publish the newer version with the percentages for the forecasts? would make the handling a bit easier and faster.
also a suggestion: could be handy to let the forecast work the other way around (more like a backcast ;)). lets say the input is the outcome you want, for example state “A”. now you can choose a backtest range, lets say 3, and it gives you the sequence of 3 states prior to A with the highest probability. could make the search for “relevant” patterns a bit more efficient. not sure how useful it will be in the end though, would need to try that out.
I will upload the newest version as soon as I get the chance. I finished it a couple weeks ago; just haven’t had the time to upload it (was traveling).
yes, how and where do i get the RAW data mt4 uses to plot the charts, i think im somewhat capable of writing some scripts to extract specific data id be looking for if i can just get the raw/flat files used and some info on the format they’re in. Any links or such ?
