ES Reversal Method
- Thread starter no_pm_please
- Start date
I mean no offense towards you or anyone.
This is your responsibility - and it is mine to protect myself from stupid ideas.
Chris
I realize this was not a repsonse to me, I just felt you hit the nail on the head here and why I got so irritated.
My appologies to NPP for an unnesassary diversion.
I wanted reply mechtrader before, but I did not want make this thread off the topic and becoming flaming war.
It's great that mechtrader coded the NPP's method and backtested. However, many of us here are also looking at the method too. And probably see the same thing as he did. But one should not dismiss an idea too quickly, if you think the basic idea is sound. (If not, then you don't have to coming to this thread.)
Before you dismiss the method, you should at least first make sure what you implemented are identical or similar to what NPP proposed. For example a few of us are struggling with the discrepency of different time frame CCI Average in TS. (I see MechTrader's screen is also from TS2000i, and the shape of 30Min CCI Average is very similar to NPP's, which I think may be wrong.) Also it is not very clear to me what the exact exit rule is that NPP has in mind. So without these issues cleared and try to compare the code with manual entries, how can one quickly dismiss the method. And before one get the basic rules in place, it may not be a good idea to start to optimize away.
My main objective in looking at a method is not necessarilly whether it made money over the whole history or not. But whether the method worked what it intended to do: catch the reversal in non-trending days.
As many have pointed out, many of Mechtrader's ideas are good and known to many of us. Such as: relax the stop loss point, avoid certain time of the day, filter the trade with some threshold (MechTrader added threshold for more clear buy/sell region, Today NPP modify the rule on the fly to not trade when the overbought/sold condition are too strong.), different CCI length from the default value etc. However, these issues to me are secondary, for later work. First step is to fine tune the basic idea and see whether it works as we intend it to work.
To me there is no Holy Grail in trading, one doesn't expect to find a strategy that will work and make money for you right away. To me it is a process finding some basic sound ideas and constantly tweaking and experimenting.
Now hopefully let's get back to the strategy.
It's great that mechtrader coded the NPP's method and backtested. However, many of us here are also looking at the method too. And probably see the same thing as he did. But one should not dismiss an idea too quickly, if you think the basic idea is sound. (If not, then you don't have to coming to this thread.)
Before you dismiss the method, you should at least first make sure what you implemented are identical or similar to what NPP proposed. For example a few of us are struggling with the discrepency of different time frame CCI Average in TS. (I see MechTrader's screen is also from TS2000i, and the shape of 30Min CCI Average is very similar to NPP's, which I think may be wrong.) Also it is not very clear to me what the exact exit rule is that NPP has in mind. So without these issues cleared and try to compare the code with manual entries, how can one quickly dismiss the method. And before one get the basic rules in place, it may not be a good idea to start to optimize away.
My main objective in looking at a method is not necessarilly whether it made money over the whole history or not. But whether the method worked what it intended to do: catch the reversal in non-trending days.
As many have pointed out, many of Mechtrader's ideas are good and known to many of us. Such as: relax the stop loss point, avoid certain time of the day, filter the trade with some threshold (MechTrader added threshold for more clear buy/sell region, Today NPP modify the rule on the fly to not trade when the overbought/sold condition are too strong.), different CCI length from the default value etc. However, these issues to me are secondary, for later work. First step is to fine tune the basic idea and see whether it works as we intend it to work.
To me there is no Holy Grail in trading, one doesn't expect to find a strategy that will work and make money for you right away. To me it is a process finding some basic sound ideas and constantly tweaking and experimenting.
Now hopefully let's get back to the strategy.
I started to code this one in WLD but the trading rules weren't complete. I have to say that I agree with the thrust of MechTrader's points. This type of trading is a statistical game and it would be foolish to risk real money without even testing to see if your idea had some grounding in past market behavior.
I also have to agree that March was an anomolous month for a lot of different types of systems. So, I wouldn't be using it as a benchmark for a strategy at this time.
Finally, I have yet to see a CCI based system that when all the airy fairy stuff is shaken out actually is profitable. Having said that, I think it's great that people want to work together to examine an idea in a more rigorous fashion.
Just my $.02.
I also have to agree that March was an anomolous month for a lot of different types of systems. So, I wouldn't be using it as a benchmark for a strategy at this time.
Finally, I have yet to see a CCI based system that when all the airy fairy stuff is shaken out actually is profitable. Having said that, I think it's great that people want to work together to examine an idea in a more rigorous fashion.
Just my $.02.
Here is my attempt to code NPP's reversal method in TS8.1 code.
I have made many of the quantities parameters, so that we can change them or optimize them later. I also borrowed an idea from Mechtrader, added a parameter that will stop open position if the new hi/lo is too far.
What I am not too clear is the exit rule.
Just want to share our work like NPP did starting this thread. Hopefully we can get more feedbacks and get more interesting insights. (Don't blame me if you lose money on this
)
I have made many of the quantities parameters, so that we can change them or optimize them later. I also borrowed an idea from Mechtrader, added a parameter that will stop open position if the new hi/lo is too far.
What I am not too clear is the exit rule.
Just want to share our work like NPP did starting this thread. Hopefully we can get more feedbacks and get more interesting insights. (Don't blame me if you lose money on this
)I`m not to surprised about the controversity and emotional reactions, the human ego is always a strong factor and the wish to beeing right especially a male characteristic. Mark Douglas describes in "The disciplined Trader" (I believe it was...)that the most constant losers are doctors, ingenieurs, bankers(!), businessmen, the groups beeing very successful in the normal life. So beeing smart is not the problem, this is a group of smart people. This is a group of people with strong ego, firm belief in their opinions, and used to fight for their opinions. That`s fine to get some good position in society, but the 100% way to lose in markets. You better have no opinions, there are just facts, if you like it or not, the market is not interested in opinions.
Me I don`t feel knowing much about the market, I`m definitely shure I don`t know the most NOT. It`s a dangerous state of mind if you feel like mastertrader#1, checking the game...
So cause this discussion scratched some egos, we saw some egoreactions, I´m not surprised, just normal.
(......I have never been a big fan of mechanical back-testing. I think there is an element of "feel" to any method before it becomes successful. I appreciate the backtesting though......)
That`s what i stated some pages before:
"especially on intraday timeframes you can NOT trade indicatorsignals mechanical, it needs a lot of experience to decide if the market fits the useability of the indicator, or to which % this is a good or bad signal."
But this thread started as a complete mechanical outruled stuff. I would not call it "feel"what you need, this "system" is not yet a system, it`s just the Trigger, you must have some good reason, why market should turn, a SETUP, may be any daily Stituation pointing to a range day or a higher probability of a late morning reversal, cause yesterday was already a trendday or something in this way, or A/D issues, TICK, TRIN, I don`t know, it`s a lot of work to find out, for me I would expect some months of work to check this in detail. To fade trends and capturing Tops/Bottoms is the King Discipline, the harder way, I prefer to go with the trend, cause the chances are more in my favour.
If you have any proofed method pointing to higher probability of rangedays or morningreversals use the "system", but DON`T try to trade it mechanically day in, day out. And better don`t try fading afternoon breakouts, trends are more constant in the afternoon, tested and written in lot of literature, or you have a very good (proofed) reason to do so....
Good Luck in your trading, think about the ego stuff or better read Mark Douglas (Emule..;-))
cheers, Michael
Me I don`t feel knowing much about the market, I`m definitely shure I don`t know the most NOT. It`s a dangerous state of mind if you feel like mastertrader#1, checking the game...
So cause this discussion scratched some egos, we saw some egoreactions, I´m not surprised, just normal.
(......I have never been a big fan of mechanical back-testing. I think there is an element of "feel" to any method before it becomes successful. I appreciate the backtesting though......)
That`s what i stated some pages before:
"especially on intraday timeframes you can NOT trade indicatorsignals mechanical, it needs a lot of experience to decide if the market fits the useability of the indicator, or to which % this is a good or bad signal."
But this thread started as a complete mechanical outruled stuff. I would not call it "feel"what you need, this "system" is not yet a system, it`s just the Trigger, you must have some good reason, why market should turn, a SETUP, may be any daily Stituation pointing to a range day or a higher probability of a late morning reversal, cause yesterday was already a trendday or something in this way, or A/D issues, TICK, TRIN, I don`t know, it`s a lot of work to find out, for me I would expect some months of work to check this in detail. To fade trends and capturing Tops/Bottoms is the King Discipline, the harder way, I prefer to go with the trend, cause the chances are more in my favour.
If you have any proofed method pointing to higher probability of rangedays or morningreversals use the "system", but DON`T try to trade it mechanically day in, day out. And better don`t try fading afternoon breakouts, trends are more constant in the afternoon, tested and written in lot of literature, or you have a very good (proofed) reason to do so....
Good Luck in your trading, think about the ego stuff or better read Mark Douglas (Emule..;-))
cheers, Michael
In summary:
-This is no_pm_please's journal.
-He devised a new trading method, tried it, and as he did in the past came here to share and discuss it.
-mechtrader41 disagrees with this method and put a lot of work in backtesting it
-no_pm_please asked for this code to backtest it himself
Let's keep our eyes on the ball, please!
In the mean time let's give both guys credit for sharing their work and ideas.
-This is no_pm_please's journal.
-He devised a new trading method, tried it, and as he did in the past came here to share and discuss it.
-mechtrader41 disagrees with this method and put a lot of work in backtesting it
-no_pm_please asked for this code to backtest it himself
Let's keep our eyes on the ball, please!
In the mean time let's give both guys credit for sharing their work and ideas.
Well, I am not a doctor, or engineer and definitely not a banker.
I read NPP's strategy and applied it today the first time on YM. Did not follow it blindly. I did use the 5 min CCI average with the LRC and watched it with other market internals as TICK and TRIN and the BB and Stoch squeeze.
The 1:55 high was 11291 and the CCI cross was at around 11283. when I entered my short. I did not cover at the next CCI cross up because most of the ticks were hovering around zero or in the negative territory. So I waited for the next CCI cross and stayed short until the next low bar of the day which was at 3:50 and covered at 11205.
I did use only a minimal number of cars to test this system. It worked today. Again, I am pretty new at this, but my bottom line is to see the big picture and not to stick 100% to the rules.
Thank you NPP.
I read NPP's strategy and applied it today the first time on YM. Did not follow it blindly. I did use the 5 min CCI average with the LRC and watched it with other market internals as TICK and TRIN and the BB and Stoch squeeze.
The 1:55 high was 11291 and the CCI cross was at around 11283. when I entered my short. I did not cover at the next CCI cross up because most of the ticks were hovering around zero or in the negative territory. So I waited for the next CCI cross and stayed short until the next low bar of the day which was at 3:50 and covered at 11205.
I did use only a minimal number of cars to test this system. It worked today. Again, I am pretty new at this, but my bottom line is to see the big picture and not to stick 100% to the rules.
Thank you NPP.
Quote from ww_nyc:
Here is my attempt to code NPP's reversal method in TS8.1 code.
I have made many of the quantities parameters, so that we can change them or optimize them later. I also borrowed an idea from Mechtrader, added a parameter that will stop open position if the new hi/lo is too far.
What I am not too clear is the exit rule.
Just want to share our work like NPP did starting this thread. Hopefully we can get more feedbacks and get more interesting insights. (Don't blame me if you lose money on this
Thank you for the code. I converted it to TS2000i.
Attached is the code in 2000i format for others to try.