[R-SIG-Finance] Advice on Forecasting

Nick White n-e-w at qtradr.net
Thu Nov 19 05:02:59 CET 2015


Dan,

this sounds like an interesting problem and one that several members of the
list could help you with. As Ilya said, please give us a reproducible
example so we can help you out.

On Wed, Nov 18, 2015 at 12:11 PM, Dan Mack <dmack10 at verizon.net> wrote:

> Hello,  I could use some advice on a problem I have trying to solve for
> projecting forward some predict functions.  I have a backtest that is
> working very well but would like a more sound projection because I am feel
> I not doing this as well as I could.   I feel I am at the limit of the
> predict() function and need a bridge to a forecast() like function.
>
> I am interested in making a next prediction on a stock going up or down
> and use the following equation.  This is one of many algorithms I use but
> they are all set up the same way.
>
> I run the follow example code to fit and predict the model:
>
>  strat.fit <- glm(DirNDay ~l_UUP.Close + l_FXE.Close + MA50 + +MA10 +
> RSI06 + BIAS10 + BBands05, data=STCK.df,family="binomial")
>
>
> strat.probs <- predict(strat.fit, STCK.test.df,type="response")
>
> I am also trying to predict the direction DirNDay (Direction Next Day).  I
> have moved the next day up t+1 against day t EOD Data.  This collectively
> works well with many algorithms in a stacking scheme.
>
> What I would really like to do is to be able use my algorithms (a
> combination of regression classifiers, decision trees and meta learners)
> and get them to do a forecast like function like used with a time series.
>  An example would be to use something like an ADABoost to fit and then
> merge it with an ARIMA like function.   Any suggested methods or functions
> (packages) would be helpful.
>
> Thanks,  Dan
>         [[alternative HTML version deleted]]
>
> _______________________________________________
> R-SIG-Finance at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only. If you want to post, subscribe first.
> -- Also note that this is not the r-help list where general R questions
> should go.
>

	[[alternative HTML version deleted]]



More information about the R-SIG-Finance mailing list