It's just a general observation at this point. I've noticed that trend days seem to be more likely when the opening gap is above or below the previous days high or low. Waiting for the market to come back past the previous close just increases my confidence the day isn't a trend day.
NO PM Thanks...Your system worked today very well....It got you out of a bad trade w/minimal loss...I am impressed with your discipline to trade your system. Thanks for the info...
I plugged in offshore's short system and tried it. It missed most of the trades so it needs to be fixed. It correctly had the 8/8 trade and the first 8/18 trade. It missed the 2 trades on 8/4, trade on 8/6, trade on 8/13, and trade on 8/14. I hope he can get the code to accurately reflect the method. I would love to see some long term results.
no_PM, Where do you see the differences are in the code as compared to how you trade the system? It appears to me that your system should be able to be accurately described in code. Richard
I'm no TS expert, but the weaknesses in offshore's code that I see are: - the system doesn't take the sequence of events necessary for a trade to be initiated into consideration. It just looks for the evidence of the events having occurred. It's important to not look for divergence before the adx crosses above 30, rather than spotting divergence and then adx crosses 30 so a setup is initiated. This is not a valid trade setup. - it doesn't seem to take the momentum into effect at all. Both to confirm the divergence and initiate the wait for the down bar. - the divergence is only detecting large long term (20 bar) divergences. In real trading a divergence of 2 or 3 bars is easy to see with the eye. - the method does not need to be above the upper keltner channel, however trades can't be initiated when the market is trading more than the stop amount from the high bar before entry. - the exit limit price is wrong since it's calculated on the previous bar. The target exit point moves intrabar and somehow this needs to be taken into consideration. - the time for initiating a trade should go out to 230 and outstanding trades need to be exited on the close
no_PM, For the benefit of fully understanding how to trade the system and also since I would like to perform a backtest using Wealth-Lab (I'm still working on getting a divergence recognition function working), I really can't say I understand your first too points above. And more specifically how they might be incorporated in code. Would you mind explaining these two points further? Richard
TS isn't perfect... it will miss trades.. I did see some trades that were missed throughout the whole test. I did a hand count when I did this and looked over the last year and found that when ADX>30 there was less than a 50% chance of the market moving between keltner channels. It was something I wanted to see for myself. If this is the case, then out of the 50% of times when the market moves between keltners, the divergence indicator must capture the turns correctly. This does not take into account "being close" to the keltner to initiate a trade. I was objective and it had to touch or penetrate the keltner band. There seems to be around 15-22 occassions per month when adx reaches above 30. Now the next question would be is what's the probability of getting divergence (subjective) during these times. At that point, it might be able to subjectively assess the accuracy rate of the method. Hope that was clear.