It should be +$50. When I'm having a bad run trading I try to stop it from getting to me emotionally. I stay focused on the money management and forge on Kind Regards, James
James, You have been at it for 3.5 months and are down -$1,000. The Markets rallied in early April to Mid May then took a downturn to this week. A simple moving average crossover system would have outperformed the system you are using by a mile. What are you trying to prove?
Can you back that up by giving me the parameters of a simple moving average crossover system that was profitable over the last 10 years or even 5 years? I only have one Dakota system that is along those lines and it is a very long term trader. Kind Regards, James
In all honesty I haven't back tested simple sma systems. However, since you started this test, the Dow rallied from 12,300 to 13,100 then dropped to 10,830. That is a total move of 3070 Dow points and your system is underwater by $1,000. Seems to me there were opportunities missed, no?
Yes definitely. The hypothetical trade that resulted in a $3k+ loss (assuming no stop loss strategy) is typical of the type of market behaviour that my models fail in. The majority of my models perform well when the price is tending to 'mean revert' rather than trend strongly. I started a project on the Dakota forums about 1 year ago to develop a set of trend following Dakota systems. I haven't given it the time it deserves. If I could anticipate trending market conditions then I could automatically switch to trend following models such as the MA Crossover that you mentioned. Kind Regards, James
I use several mechanical, NN and AI systems. What I have found is at times I have to make a decision to switch my strategies from those that work well in sideways markets (oscillator based and support/resistance strategies do well) to those that work well in trending conditions. It's not an easy decision as to when to switch, it's entirely based on market sentiment and what I am seeing on the charts. Unfortunately cannot be backtested.
There is a discussion taking place in the BioCompSystems forum (not here in ET) about the switching problem. Here is an extract of what one of the users is trying to do: What I'm suggesting goes down another path where the swarm receives multiple performance values, one per system and adapts normally based upon what it makes of that performance. The twist here is that the swarm doesn't only set the parameter values for each system it also chooses which model gets traded should an entry be triggered. Using adaptation rather than optimization, Dakota will be able to look at two or more systems and select the best one to trade or just go flat. So if you have four models that work well under given market conditions - say trending up, trending down. sideways, and hugely volatile, Dakota would select the best performing model's signal to trade. In my opinion, the system would lose money when the transition from one market behavior to another was underway. However, hopefully, the switching capability would smooth the equity curve and make the model-to-model transition mechanical rather than emotional. Further, realistic backtesting should be possible. In response to the post, Carl Cook indicated that the above is a capability which could be implemented now by some of the "super users." As I mentioned in an earlier post, Dakota is being recoded to enhance its capabilities. I hope Carl decides to make the selection process an easy-to-use feature for the coding impaired like myself. Jack
Signal for Mon July 21, 2008 MOC: Short Trading Instructions for Mon July 21: None Open Position: Short 1 ESU8 @ 1260.00 Trading History: 4/1/2008 Bought 1 ESM8 @ 1371.00 4/22/2008 Sold 2 ESM8 @ 1380.25 5/1/2008 Bought 2 ESM8 @ 1412.1 5/7/2008 Sold 2 ESM8 @ 1394.7 5/8/2008 Bought 2 ESM8 @ 1392.5 5/19/2008 Sold 2 ESM8 @ 1429.3 5/27/2008 Bought 2 ESM8 @ 1385.25 6/4/2008 Sold 2 ESM8 @ 1377.00 6/10/2008 Bought 1 ESM8 @1356.75 6/10/2008 Bought 1 ESU8 @ 1359.00 6/18/2008 Sold 2 ESU8 @ 1337.50 6/25/2008 Bought 2 ESU8 @ 1323.00 7/10/2008 Sold 2 ESU8 @ 1255.00 7/17/2008 Bought 2 ESU8 @ 1254.00 7/18/2008 Sold 2 ESU8 @ 1260.00 Profit / Loss From Closed Positions: 4/22/2008 $462.50 5/1/2008 ($1,592.50) 5/7/2008 ($870) 5/8/2008 $110 5/19/2008 $1,840 5/27/2008 $2,202.50 6/4/2008 ($412.50) 6/10/2008 $1,012.50 6/18/2008 ($1,075.00) 6/25/2008 $725.00 7/10/2008 ($3,400) 7/17/2008 $50 7/18/2008 $300 Total Net Profit: ($648.00) Note: This is a hypothetical trading record. Kind Regards, James
Signal for Tue July 22, 2008 MOC: Short Trading Instructions for Tue July 22: None Open Position: Short 1 ESU8 @ 1260.00 Trading History: 4/1/2008 Bought 1 ESM8 @ 1371.00 4/22/2008 Sold 2 ESM8 @ 1380.25 5/1/2008 Bought 2 ESM8 @ 1412.1 5/7/2008 Sold 2 ESM8 @ 1394.7 5/8/2008 Bought 2 ESM8 @ 1392.5 5/19/2008 Sold 2 ESM8 @ 1429.3 5/27/2008 Bought 2 ESM8 @ 1385.25 6/4/2008 Sold 2 ESM8 @ 1377.00 6/10/2008 Bought 1 ESM8 @1356.75 6/10/2008 Bought 1 ESU8 @ 1359.00 6/18/2008 Sold 2 ESU8 @ 1337.50 6/25/2008 Bought 2 ESU8 @ 1323.00 7/10/2008 Sold 2 ESU8 @ 1255.00 7/17/2008 Bought 2 ESU8 @ 1254.00 7/18/2008 Sold 2 ESU8 @ 1260.00 Profit / Loss From Closed Positions: 4/22/2008 $462.50 5/1/2008 ($1,592.50) 5/7/2008 ($870) 5/8/2008 $110 5/19/2008 $1,840 5/27/2008 $2,202.50 6/4/2008 ($412.50) 6/10/2008 $1,012.50 6/18/2008 ($1,075.00) 6/25/2008 $725.00 7/10/2008 ($3,400) 7/17/2008 $50 7/18/2008 $300 Total Net Profit: ($648.00) Note: This is a hypothetical trading record. Kind Regards, James