Tuesdays Prediction: -3.35 -Result- Open: 866.87 Close: 827.16 Change: -39.71 Prediction: Right Note: I have started paper trading the predictions as well and will post those results from here on. Also, based on observed accuracy I wouldn't suggest trading predictions between -1 and 1. Anything above or below that margin will be paper traded by me. Paper Trade (one ES contract, Market Orders): Sell ES @ 858.25 (9:30am) Buy ES @ 825.00 (4:00pm) Change: -33.25 Profit/Loss: $1,662.50 Overall Profit/Loss: $1,662.50 Next Prediction: Wednesday (02/11/09): 6.08 This means there is a 80% or greater chance the S&P500 will close higher than it opens tomorrow (Wednesday).
What is your dependent variable here. What exactly is your model predicting? (C - O) / O ? lnIC) - ln(O) ? or just C - O ?
Just curious and maybe I missed something but if you are so confident in your "model" why haven't you n-tupled your money and are instead looking for other peoples money? Furthermore, perhaps I haven't read a trading thread in quite some time and I understand you won't provide much info on your 'figures', but can you at least offer some intuition behind your model. Thanks.
Please read the original post for more information. The prediction simply predicts whether the S&P500 will close higher or lower than it opens. The further the number is from zero indicates the signals strength/certainty. Positive number predictions suggest a buy at market open and sell at market close. Negative number predictions suggest a sell at market open and buy at market close. I'm sorry but the way the predictions are created is a secret, not to mention complicated beyond my writing ability to explain. PhalanX is a form of machine learning I created due to the inadequacy of genetic algorithms and neural networks. I simply programed it to output like this because it was the simplest way (for me) to see the predicted direction and certainty in one number.
this should be a fun thread, lotsa luck. dont pay too much attention to most of the posters here, your system will speak for itself.
I haven't "n-tupled" my money because I can't afford the initial capital to invest. Sure, I could go get a small loan and hope that I can satisfy margin requirements long enough to get things rolling but that leaves very little room for error. No system will be right all the time and you MUST be prepared for losses. Don't worry, I don't plan on letting many people have access to it, and those that do, will only have access for a short time. I will never say exactly how it works or what data it uses, however it doesn't rely strictly on generic market data. Part of why it works so well is due to the fact it can use data no other system can use (at least that I know of). I am also still working on setting up a blog and website where I can post the predictions as well. Hopefully that will be done in the next few days.
OK, good luck to you. PS. I was asking for intuition behind the model, not the details of how it works or what data it uses. Regardless, I look forward to your predictions.
Wednesdays Prediction: 6.08 -Result- Open: 827.41 Close: 833.74 Change: +6.33 Prediction: Right Note: I will try to simplify the trading signal from here on with a slightly different format. (buy, sell, wait, etc) Paper Trade (one ES contract, Market Orders): Buy ES @ 829.75 (9:30am) Sell ES @ 832.00 (4:00pm) Change: +2.25 Profit/Loss: $112.50 Overall Profit/Loss: $1,775.00 Next Prediction: Thursday (02/12/09): Prediction: -0.03 Signal: Wait This signal should not be traded. While the prediction is for the market to close lower than it opens, there is very little certainty.