yea it's just a pitiful blank i know - i web guy was working on a project and screwed it up so i just deleted the index page till i can get it fixed.
you can improve your mean reversion strategy by analyzing mean reverting events and trading those events, or by using cointegration of pairs. The former will improve your expected return, and the latter will reduce your drawdowns.
i knw this is not what your saying but it made me think i have a spreadsheet that uses quartiles to basically determine the optimal 75 percentile return on the trade so trading multiples you would take off some there and leave the rest to run. but i see what your saying, reverting is something i have struggled with as far as when to identify is it a direction change or just a pull back. i have kinda learned that once the longer term direction has changed the best time to do a reverting trade is then because the longer the direction holds the more likely a reverting trade is actually a direction change. it's easy to see all this on a chart and super hard to automate it, simply there is a disconnect between what the eye sees and the auto trader can execute.
universal system like many machines that try to be all - just fail - they don't do anything well and everything half ass. i gave up on a universal system long ago, i totally know you can however create a universal indicator that will look the same no matter what you put it on. anything that uses percentages is going to endure thru the years and be universal. but not the systems are specific to es though by a fluke they may work on something else.
there's a theory (w/ lots of research) that mean reversion intraday is driven by volume imbalance, where low volume periods intraday cause price to drift on a random walk, skewing away from the trend. you can id time periods (intraday) where this is most likely, during periods of time (days between earnings announcements), and skewed to the momentum of the underlying security. biggest challenge is always drawdowns. :\
A phenomenon that often occurs, there will come a mkt change signal (EOD) which a few days later appears to be a dummy. The dummy is actually the dummy, ie, the original mkt change signal takes a while to play out with inbetween action looking ln hindsight to be pullback. Difficult to explain but we get a signal earlier on which then shortly after looks false but it wasn't, it just takes longer to play out than expected. It's like a shake out where both bulls and bears get torched because both sides get fooled.
if you are using wide stops with small targets - which i think you are : correct me if i am wrong - then you are trading high probability [ using indicators helps in this since indicators confirm ] .......trying to reduce drawdown is contrary to whole trading strategy. so just accept it. the only way to reduce drawdown is to shift to a low probability trade
You mention that you have about 3 possible exit strategies. One of the 3 is the hard stop and that is the least favorable. Can you talk about the other 2 possible (more favorable) stops or exit strategies and how you determine them? Hopefully I didn't overlook any comments you already made about this. Thanks, -Al
1.) hard money profit target as you mentioned i always have the above on most every system these below i optimize to determine what the 2 best are for a total of 3. every market likes different stop methods. 2.) %trailing stop on an optimized percentage, usually around 1-2.5 %. 3.) fast moving average change of direction of and optimized length usually around 8-12. 4.) standard deviation break of an optimized value, usually around 1-3 devs. 5.) channel breakout usually 8-20 look back. 6.) played around with most every type these are the ones i keep coming back to. oscillators all sorts of stuff comes to mind. i do have one stop method that is a stop and reverse and it's quite complicated i need a whole other thread just for that stop. i only use it on bonds and some other markets i don't trade any longer. so add the fixed and pick two above and that is my three stop methods.