ES Reversal Method

Well, it looks like another round where everyone looks at the same market with different software setups and has different results.

Gotta love this game.

I looked at the 30 min. discrepency. Yes, it exists. I plugged all the CCI(20) numbers for the past 15 periods into excel and tried to come up with the average cci number. No matter how I computed the average it didn't match. Then I thought maybe I didn't have the cci average calc. figured out so I setup a screen with the cci average indicator and another where I plotted the cci(20) and another plot of the 10 period average. Results were identical. So there you have it. The thread for practical purposes is probably going to be useless to most people because we're never going to get the same signals.

I also tried using ESM6 instead of the perpetual contract. The results were nearly identical.
 

Attachments

I tried to refresh my data from Globalserver but it justs gives me the same data and doesn't give me the first 10 min. of Friday's trading. Says it downloaded fine, but no changes in the data are seen.

Seems like a waste of time, so I give up.

Will continue with the thread tomorrow as the trades unfold. If I see something where a change to the rules could improve the results, I won't hesitate to make the change. March was a great month for this method. Lets see what April brings...
 
Yes, I can replicate TradeStation's CCI average. The 30Min CCI only has values every 30 minutes. If I print out the 30Min CCI values at every 30 minutes, take the 10 period average, I get the exact same CCIAvg as TradeStation's. Following are the raw data if I use ES.D symbol for the end of 3/31:

Time, CCI, TS CCIAvg, Manual CCIAvg
16:15, -36.26, -108.44, -108.443
16:00, -130.2, -116.07, -116.072
15:30, -100.19, -109.29, -109.286
15:00, -83.54, -105.48
14:30, -144.74, -98.82
14:00, -195.83, -93.22
13:30, -135.05, -79.28
13:00, -76.66, -71.85
12:30, -83.75, -76.77
12:00, -98.21, -80.04
11:30, -112.55, -88.97
11:00, -62.34, -90.54

In EasyLanguage, in order to do the average correctly, you need to declare the variable to be the same frequency as Data 2:
Var:
myCCI(0,Data2);

Then the Average(myCCI, 10) will be correct.

Quote from jerryz:

have you tried creating your own 30 min CCI indicator to replicate Tradestation's? i can replicate the 30min CCI but the 10 period average of the 30min CCI does not match Tradestation's. if i calculate the average manually on a piece of paper it matches my indicator but it does not match Tradestation's.
More...
 
I think NPP had given very detailed instructions. Although it is for i2000, in TS8.1 it is very similar.

1. Insert a second symbol to the chart, make it ES.D 30 min
2. Insert a CCIAvg indicator, then format it to use the second symbol, i.e. 30 minutes
3. If you want to it to look better, you can actually hide the 30 min chart by formatting the symbol->Scaling->Subgraph->Hidden

Quote from milan47:

would you please tell me how did you insert the 30 min CCI under the 5 min ES chart and the 5 min CCI on TS 8.1?
thank you:confused:
More...
 
Quote from ww_nyc:

I think NPP had given very detailed instructions. Although it is for i2000, in TS8.1 it is very similar.

1. Insert a second symbol to the chart, make it ES.D 30 min
2. Insert a CCIAvg indicator, then format it to use the second symbol, i.e. 30 minutes
3. If you want to it to look better, you can actually hide the 30 min chart by formatting the symbol->Scaling->Subgraph->Hidden
More...


Thank you, it worked fine.
 
NPP,
The chart you show is for the 5 minute CCI and CCIAvg. My CCI and CCIAvg value and Chart are very close to yours using TS 8.1 and its data feed. They are close but not identical. (yours at the end of 3/31 is 204.77, 8.51, mine is 203.55, 8.54. The shape of the curves are very similar.) But I think the signal from the 5 minutes should be very similar.

My problem is with the shape and the look of your CCI and CCIAvg for 30 minutes chart you posted before. The CCIAvg for 30 minutes is very different from mine (CCI values however, looks similar). I did verified the TS8.1's CCIAvg calculation to be correct if you calculate the average only in the 30 minutes timeframe.

Quote from no_pm_please:

Well, it looks like another round where everyone looks at the same market with different software setups and has different results.

Gotta love this game.

I looked at the 30 min. discrepency. Yes, it exists. I plugged all the CCI(20) numbers for the past 15 periods into excel and tried to come up with the average cci number. No matter how I computed the average it didn't match. Then I thought maybe I didn't have the cci average calc. figured out so I setup a screen with the cci average indicator and another where I plotted the cci(20) and another plot of the 10 period average. Results were identical. So there you have it. The thread for practical purposes is probably going to be useless to most people because we're never going to get the same signals.

I also tried using ESM6 instead of the perpetual contract. The results were nearly identical.
More...
 
NPP,

.....March was a great month for this method. Lets see what April brings...

Yes, that`s right, march worked fine, but why don`t any of you make simply a backtest? You all should a little bit reduce the euphoric mood about a strategy working one single month, do some serious work, go back the charts by hand the last 6-9 months and work a couple of days to check the historic outcome, or a lot using Tradestation, do some backtest.
Do you really expect a VERY simple indicatorstrategy will make money in the S&P? The industry has the best systemdevelopers, money can buy, working years, hours a day...and look what`s available in the systemsector for rent and buy to meet reality again..

march was great, that`s right, but in tested 1267 trades from may 2004, this system earned ~ 400,-$ without commissions...

That`s my last post here, take care, it`s definitely not the way to develop a strategy, looking on things working the last 3-4 weeks....

Michael
 

Attachments

this is what's happening in TS2000i.

to replicate TS2000i's 30min CCI average exactly, you have to use this code:

Code: 
Plot1(cci(20) of data2,"CCI",red);
Plot2(average(plot1,10),"CCIAvg",green);

strangely, however, Plot2 is not plotting the 10 period average. if you calculate the numbers in Excel, your averages will not match Plot2.

to plot the 10 period average that does match Excel, you have to use this code:

Code: 
value1 = cci(20) of data2;
Plot1(value1,"CCI",red);
Plot2(average(value1,10),"CCIAvg",green);

these two pieces of code should plot the same thing, yet they are not. does anyone know what's going on?
 
I really feel i have to keep people from losing money, may be it helps to be more constructive.

1. Why using the default Length of (20,10)? It`s just the default of the CCIAV in TS. The same with the default 14 in most other indicators, it was an error of Wilder to use 14, he wanted half the mooncycle, but there are only 10 tradingdays, defaults mean nothing.

The optimized Parameters by keeping the AV 10, are 30 for the 5 min CCI and 25 for the 30 min, by testing the numbers of allowed # of trades/day, the best was 1(one) trade per day.

With your basic rules and trading every time of day, 1 trade/day, 30 and 25 for CCI, setexitonclose and realistic 20,-$ commissions the results are here
 

Attachments

You must log in or register to reply here.
Back
Top