Trading eminis with cycles

[theplumber]

Originally posted by Anson
Hi Plumber,

Thanks for the charts.
I have been playing around with the Cyclepack for the last 2 days.

Tried the 2 mins, 4 mins, 5 mins, 15mins and Daily charts. Do I remember correctly that you once said that Weekly charts seems to be the best for you?

I'm still in the midst of experimenting with all the parameters... there are so many possibilities!!

I seem to have a preference for 256 data window so far... is your experience better with the 512/1024?

For some reason.. the fitness/cycles don't seem to change much between 512 and 1024. Is it just me?

Do you tweak the "bars backward" setting much for fitness and notice any improvement in reliability? I seem to be lost in a world of settings.

I'll experiment more and hope to be able to add some stuff here sometime soon.

For the days trading I've found the 256 data window with 0 filter and 5 MA on a 5 minute chart the best. Below is what I have for practicality, in other words can this software make me money today? and the answer is yes. I'll care about next week/month another time so make me money today or else. The left side is non-updated (from last trade Thursday) and the right side is updated after the first bar is complete, b/c of the gap. If you've ever worked with models intraday you would see first off that gaps play havoc on models. I've seen a model show up morning but b/c of a strong gap up it satisfied the requirements of an up morning. For this reason I update and will flip back and forth to see which one has it right. The chart below also has XT in the price pane and plain projector in the indicator window. They both have the same settings but they give different projections so I like to have an alternate.

Now for larger time frames, the best will ALWAYS be the weekly. Cycles have a tendency to fluctuate and disapear from time to time on shorter time frames ( I would suggest taking about a year to study cycles to see what I mean) but one hedge fund I know of uses them to time the futes using daily and 60 minute time frames. Then you would be looking for the best swing strength cycles . The data windows for weeklies doesn't matter, they are more or less perfect, for lack of a better word.The dailies and 60 or 30 minutes you will have to backtest with different settings to get the best for each stock/index.

[theplumber]

This was updated at the lows Thursday with the forecasted purple model. This is again 30 minute, 256 data window with 0 filter and 5 MA. I think the trick though after finding what data window and other settings to use will be when to update. This is a HUGE trick with the Pesavento map from Ensign and is no different with the re-calc with the cycle pack. I think Divergence has a suggestion on their website to update at pivots but finding the low or high in real time can be difficult.

[theplumber]

Some more cycles. On the left is the daily S&P and the right is 60 minutes. The reasons for anchoring where they are is based on the important information (which is my opinion and not any rule per say). I feel I have to use the analyzer as well as the models to get a handle on where the models could be "wrong" , what I mean is just late or early vs where the raw cycles have peaks and valleys. Both these models are using the 512 data window

[theplumber]

A one of a kind item being auctioned soon http://www.minyanville.com/company/press.htm from here http://www.critterschoiceawards.com/

[theplumber]

I know this has nothing to do with cycles but I'm finding some interesting patterns emerge the past few weeks. The 15 minutes befoer the open used to be quiet, not the past few weeks. Also as at home trader I know I will always be less informed than others so seeing what they do with their money is the most telling way to find direction. I've noticed that when large volume spikes along with large tick buying/selling differences mark so good bottoms and tops, again people putting their money on the line. Some very good indicators on the market but what actually happens is highly overlooked and can give further confidence to using size in a trade. Anyways this is just my 2 cents and what I've seen for patterns below.

[theplumber]

Here's an interesting comparison of Ensigns Pesavento map and the cyclepack. Both are 24 hour models and updated both 2 hours before the open. What is a regular in raw cycle models (raw meaning without feeding the data into a neural network like I do and find a little more accurate) is for a intraday cycle to invert, and both cycle models have the same inversion late in the day. This I've grown accustomed to and just use the pivots as just that on large cycle projection, like was shown late today..Just another tidbit of info as to the way I've used both the cyclepack and Pesavento map intraday.

[theplumber]

A followup to yesterdays chart

[brianb]

Do you prefer to update the Pesavento map using the overnite data in it's calculation? I see what appears to be 24 hr data in the Ensign chart and RTH data for the cyclepack chart. I've tried the P maps using both 24 hr and RTH session data on many symbols over a 3 week period and much prefer the RTH only results.

[theplumber]

Originally posted by brianb
Do you prefer to update the Pesavento map using the overnite data in it's calculation? I see what appears to be 24 hr data in the Ensign chart and RTH data for the cyclepack chart. I've tried the P maps using both 24 hr and RTH session data on many symbols over a 3 week period and much prefer the RTH only results.

I'm always experimenting with the models and like also to use only market hours but with the Pesavento map I've have seen good accuracy with updating 1-2 hours before the open. The results have been good for me for some time now. It's always been knowing when to update, like just before an important stat 1 hour before the open I like to update 15 minutes before to see what it looks like, and it's been surprisingly accurate with hitting the direction after the numbers(s) come out. The cyclepack has it's best days finding a good pivot, like yesterday's high 1 1/2 hours before the close was a good model for all day today, which makes sense b/c most traders set themselves up for the next day 1- 1 1/2 hours before the close. Also if there is a good pivot the first hour makes a better model, this seems to work for the Pesavento map as well. With the gaps the last week it's been hard to judge for me when to mess with them. Here's a 30 minute cyclepack from last Fridays closing price.Not bad so far if Wednesday is an up day.

[theplumber]

Here's another example of haveing both a regular trading hours and 24 hour model for both peices of software. One thing I've noticed with ES trading with the European markets is the action doesn't heat up until 30-60 minutes after they are open so 2:30 CST I update both 24 hour models and look what it gave me below.

[brianb]

Glad to hear you found an update time that works for you, maybe I'll dig up those old 24 hr charts again and give them another look. I agree knowing when to update is very important. Too much updating or more correctly updating at the wrong time can cause the prediction to sorta flatten out. Maybe it's due to calculating from the wrong points in time, just a guess. I try to update after a major intraday swing has occured, but that's easier said than done . Some days the maps just don't match the prices at all. Those are days I wish I'd stayed in bed. Just a bad day to trade. Gaps are another animal that's rough to time just right. Here's yesterday's map updated 15 minutes into the RTH. Not a bad result.

[theplumber]

Originally posted by brianb
Glad to hear you found an update time that works for you, maybe I'll dig up those old 24 hr charts again and give them another look. I agree knowing when to update is very important. Too much updating or more correctly updating at the wrong time can cause the prediction to sorta flatten out. Maybe it's due to calculating from the wrong points in time, just a guess. I try to update after a major intraday swing has occured, but that's easier said than done . Some days the maps just don't match the prices at all. Those are days I wish I'd stayed in bed. Just a bad day to trade. Gaps are another animal that's rough to time just right. Here's yesterday's map updated 15 minutes into the RTH. Not a bad result.

I wish I could backtest the Pesavento map like I can the cyclepack to show you and everyone else something about what I've seen. One real big confidence trick so to speak is to use both 24 hour and regular trading hours model, then find an important pivot (or time like this Fridays employment report) and watch to see if both model times match in forecast. The average I've seen is around 75% match for Pesavento map and cyclepack they match the 24 hour to regular hours. When they don't match the 24 hour one is the right one 100% of the time. Below is the open anchor point for the cycle pack and gives the same impressive result, even the inversion of the last 1/2 hour.

[theplumber]

Not a bad forecast for the whole week. On the right was the chart I posted yesterday with the model showing a nice rally and me saying "Not bad so far if Wednesday is an up day", well that was an understatement. Talking to my trader friends of mine during the day we all said that $52 oil, FNM mess and possible terror in front of the elction doesn't make them sell then their going to lining up to buy'em. In fact this was an interesting tid bit of info I read today "On Sept. 2nd the SPX crossed back through its positive
trending 200-day moving average after being below it for 30
trading days. This has only happened 6 times since 1950. The
SPX was always higher 60 trading days later with an average
gain of 7.8% - so far it is up 1.5%. " --Snoop Tony Dwyer

On another note, I showed some stats on when the market bottoms in August during an election year the last time was 1920, and Nov was the worst Nov on record (or something like that I'll have to dig around for the actual numbers) and this could be playing out. A 5% Nov dip looks right after squeezing to 1170 or so. Dec of course could be a nice snap back with year end bonuses on the line. Always amazes me how most Hedge funds are not hedged but instead are long bias funds.


One last edit, the Friday action looks like a sell the news or maybe a bad number I don't know but the cycles point down. It isn't in the screenshot I gave above.

[theplumber]

I trying to cut back on my posting these charts, they are just too valuable to me and don't really want to spoil the system. This was my chart updated at the open of the European trade of 2:00 CST am this morning. Needless to say I was long going into the rally, looking to relocate to Hawaii soon

[theplumber]

Been playing around with P&F charts and have noticed the cycleProjector has worked well with it instead of the XT like I use with 5 minute and 30 minute charts. Update line is 2:00 am CST, the start of trading in Europe.