Lawrence, Thanks for the insightful post Will you please post some results from pre-1997 and also for 3 distinctively bull runs since 1997-1998. It will be very interesting to look at same model performing very differently in different bull runs. I will post some of my own results later today. But I have mostly results from post 2008 crisis period.
What's the best ES trading system these days? That's an easy one: whatever algos are able to shadow the Fed's daily POMO injections at 1030 and/or its 3:30PM just-to-be-safe ramp into the close. (sorry, it just had to be said.)
The best system is the one that gives you a buy signal at the end of a down swing and a sell signal at the top of an up swing. Rinse and repeat throughout the day. Possible? Not sure, but I'm working on it.
Not much time today to dig things up. Pre 1997-1998 S&P range is 6 to 8 pts. 10 pts day are very rare. That couple with VIX < 10 means trading opportunities are just different from what is happening now. Here is an example where the edge of the model collapsed once Fed intervention became the norm. http://www.daytradingbias.com/?p=37679 Here is different one coming from Lawrence Connor where the edge collapsed during the 2002-2003 Internet dot bomb. It just recovered recently since 2010. http://www.daytradingbias.com/?p=87953 I will have more time to post by the end of week.
Me too. 1. Avoid bear flags in down channels. 2. Buy the channel break to the upside. When channel finally breaks. 3. Take all 1st single swing low retraces. May become a channel but loss is reasonable. 4. Now on 2nd leg down you know , from the 1st swing, what is a channel break and so on. 5. Use an "x" bar low of close for reference. 6. Measure all swings from this. 7. Better swing highs than swing lows is GO for trend. Count mag and number of them. 8. Enter and exit on standard error bands * "x" on the swing high/low 9. Trade with tend. 9. Optimize all the above. This system worked great on accident and health multi-lime insurers in 2012. Sucks with everythng now.
The first model is pointless example - it trades 20-30 times a year. Therefore, in 10 years it probably traded 250 times - how someone can even develop a model like that and say it has any positive expectation is beyond me. Also, the idea behind the model of simple moving average crossover would surely ring a few alarm bells to the developer. Moreover, the developer must surely have some data on max draw-down expected, expected losses in a row, losing months in a row or any other idea on when to pull the model before the account gets wiped out. I am pretty sure the big money is in swing trading, but for the pikers like me i think we have to rely on intraday strategies - these can be developed with smaller draw-downs and gives more trades to define the system more clearly so developer has an idea of when to deactivate the model. Anyway GMST, this is the best i have been able to come up with so far - this is 5 models which trade long/short or hedged on the YM and ES. I have been trading this live since August now and so far its doing okay. I would be happy if i can even get a one third of the back-testing performance.
Paskewitz Asset Management, LLC Contrarian 3X Stock Index Program " http://www.managedfutures.com/program_performance.aspx?fundtype=&productId=32812 Strategy Description The fully-systematic contrarian program employs multiple models to forecast short and intermediate term tops and bottoms in the S&P 500 index Yr/YTD Program 2013 -9.59 2012 6.07 2011 8.01 2010 -6.60 2009 3.03 2008 32.38 2007 39.20 2006 38.98 2005 14.05 2004 17.30 2003 -0.14 "