BigBubba,
I do not have a "specific" amount of time that the system has to forward test on. The following are the steps I take:
#1 I only "backtest" on 2/3's of my data. This means if I am going to make a modification to the strategy those modifications are only tested on this data.
#2 If the results from step 1 are to my liking, I then test on the last 1/3 of my previously untested data. You have to be careful not to move to step 2 often, because if you do then in effect you are still "testing" on this data which makes it less valuable. So as a general rule, I will only test on this data if I am very confident that I want to eventually move live with this strategy.
#3 Assuming the results are still looking good, I will move it live to a sim account. Here I will leave it for at least a few weeks. There are a few things I look for in this stage. Everyday it runs live on the sim account, the data is stored for future historical testing. So, the number one thing I look for is after the day ends, if I backtest on the data it stored the results must MATCH or exceed what the backtesting engine said it should be. The reason this is important is because occasionally something tweaks the backtesting engine for a certain strategy that produces abnormal results. So if I am going to go live with a strategy I need to verify that this is not the case. A few weeks of testing can easily verify this. The only other thing I look for in this stage is that the strategy results remain in the ballpark of what they should be. Such as equity curve not taking any sudden dives....
If all of these things line up, then that strategy is promoted to a potential live account strategy. Then I just chose the one I like the best.
I do not have a "specific" amount of time that the system has to forward test on. The following are the steps I take:
#1 I only "backtest" on 2/3's of my data. This means if I am going to make a modification to the strategy those modifications are only tested on this data.
#2 If the results from step 1 are to my liking, I then test on the last 1/3 of my previously untested data. You have to be careful not to move to step 2 often, because if you do then in effect you are still "testing" on this data which makes it less valuable. So as a general rule, I will only test on this data if I am very confident that I want to eventually move live with this strategy.
#3 Assuming the results are still looking good, I will move it live to a sim account. Here I will leave it for at least a few weeks. There are a few things I look for in this stage. Everyday it runs live on the sim account, the data is stored for future historical testing. So, the number one thing I look for is after the day ends, if I backtest on the data it stored the results must MATCH or exceed what the backtesting engine said it should be. The reason this is important is because occasionally something tweaks the backtesting engine for a certain strategy that produces abnormal results. So if I am going to go live with a strategy I need to verify that this is not the case. A few weeks of testing can easily verify this. The only other thing I look for in this stage is that the strategy results remain in the ballpark of what they should be. Such as equity curve not taking any sudden dives....
If all of these things line up, then that strategy is promoted to a potential live account strategy. Then I just chose the one I like the best.
