I'm using AlphaPy and XGBoost to predict the next day's range. It's a simple classifier to predict the probability of the next day's trading range to be above or below its ATR. My question is: What do you consider to be the simplest intraday trend and counter-trend systems? I'm considering ORB for above-range and some simple S/R system for below-range. Not looking for the Holy Grail, just some recommendations. Regards, PTR
R-Breaker was good when it was good and Gann 50% pullbacks still decent. Problem with intraday, they don't last long as unless you are experienced they have to be adapted as volatility changes. As far as counter trend MACD divergences was ok, but unless you very experienced on chart reading, it can be horrible method on a long trend.
What does one have to do with the other? You're trying to see if your ML fitted algo has better results than random?
Right, validation AUC is 0.67, and the probabilities are well-calibrated, so we can apply a filter to trade the correct system.
I like the approach you're taking. I'm always curious how people calculate their values so how did you calculate the validation AUC?
%% Another good thing about intraday; combine many intradays profit with a longer term profit to commission ratio..........100%% sure that is not a prediction; but can be profitable. But 100% sure its not eod[end of day] exactly also. Hope you agree with this. But something below 50+ 200day moving average, like many small caps= that's a downtrend /main trend not a counter trend.So expect better downtrends/ expansion of intraday +closes on the downside. NOT a prediction or commissions ad; even though i look at that in ads LOL.,,
It'll work over the Summer Doldrums which is coming up in a few weeks time, but currently it's crazy and 200pt swings on most markets, so just don't bother. Better off, just joining direction and riding it currently, at other times direction stops and reverses, then your method might be valid for a while.
%% I'm sure the best of the best short term traders could buy small uptrends, in a mostly downtrending market; but even if say QQQ upticks a bit @200 day moving average. Most likely the best way for most to lose money=don't try this @home /catch a falling safe LOL. Parabolic stop + reverse worked on silver uptrend about $50; my main use for it is a sloppy reminder to take profits some months-it seldom works intraday...………………………………………………………………………………………………………………………………...