I understand what you're saying here... sort of an anti-mean-reversion strategy. The idea being that once a move reaches a certain level of extremity, it tends to go further rather than stopping. At first glance this does seem to be a valid concept. Extreme bars do happen. I looked at 70 hourly bars on ES; in that sample the "short end" was smaller than 7 points 94% of the time, while the long end ran to 12 points as much as 34% of the time. However, I'm not sure where the predictive edge lies here. It seems like this strategy on 60min bars would be identical to an MA cross entry on a smaller time frame, and with any change in volatility you quickly run up a succession of big losses. Your stop would need to be many multiples of your profit target. I did a series of calculations in Excel and the system's expectancy seems to be invariably negative, even without accounting for commissions and other costs. There just is no case where you can be 90% sure that if price ran X, it'll run an additional 2X or whatever. Not enough to cover the losses, anyway. Do you incorporate any other elements into the analysis? Have you tried this on futures or stocks?
This is not the case. Bar average size changes massively with changes in volatility and there are many outliers.
i guess im not looking for extreme bars im using the bars stats where it extends past the short side of the bar if it hits that then there is a high chance the move will continue for at least a specified number of points or pips. every vehicle has its own personality. i only use the ema to show which direction the moves tend to fade so i will not place entrance on both sides past the open but only on the side that agrees with the bias and price in relation to ema. this is what i consider to be one of the beauties of it since its statistical nature it will change as new bars develope lets say you use statistics over the last 21 bars or 41 bars you stay relatively current with vol changes you update every new bar and drop the oldest... but this 97%ish of the time filters out consolidation and chop so your not confused and getting thrown around i actually make money through "chop". the proof in the the candle structure everything can be adjusted for this to work as all vehicles have their own personalities ask most traders who've been trading certain vehicles for years at the can with extreme accuracy tell you what pair/stock/vehicle it is without seeing the print. this just takes advantage of that personality note that this is hard to test the candle your trading in one direction not two...and thats w/ bias... then you establish your statistical marks... such as run 5 points take take at 10 points from open. stop at reverse 5 points from open... risk:reward=1:1 if it works > 50% of the time your profitable ... statistically it works 80-90 ... while i work with FX market this does work in stocks as well with adjustments for your comm. you have to have a large account its easier in FX. i have tried it in stocks and it works ive successfully done it on apple and some of the other high vol equities. anyother elements.. no i did write an indicator that does read price level and macd adjsuted it for fx but i find it easier not to use indicators.
their are outliers im not saying this is 100% im just saying its profitable.. yes vol does change and so statistics should be run rerun every so many bars with the oldest being dropped and the newest added for three levels which is what ill try to get to eventually in here lets say 21 bars 100 bars and 200 bars so you can see average over 200 100 and 21 youll know whether its higher vol or not and you can adjust.
current pending order the top line is open then 2nd line is short line the third is TP note that if price doesnt reverse and hit this price ill never be in the trade thus it will never been a loosing trade