Monthly Archives: October 2014

SB Take 3

I’m pretty transient bar-ed out for now. I’ve done what I can with it, and I don’t have a good way to tackle the problems I’m facing with it. However, I DO think I’ve managed to create a small edge out of it, so I look forward to seeing how it does on my micro account. I think a partial martingale strategy may be in order, as eurusdd suggested, and while I don’t condone such tactics, I think they are a very advanced concept to implement correctly. Until then, I will wait for eurusdd to come back next year and talk about his version of transient waves, or if anyone asks/mentions anything interesting the thread. However with how secretive everyone is, I don’t see that happening.

 

Until then I’m back to pure metadata for another time! This time I have a more solid background, and a lot of ideas, however historically I’ve hit a wall with them quite quickly and it’s easy to become frustrated. I don’t know how long I’ll stick with it this time, but I will work until I’m out of ideas as usual.

I came up with a new approach after plenty of doodling on scratch paper, and these are the pre-lim results. It seems like 80%/20% extremes are decent at predicting turning points, while anything less than that, even though relatively extreme, does not do so much. I found some failures as well, but I will look into it later.

Capture

Capture2

Advertisements

Optimized h values pt 2

Taking a page out of the forge =p

Something that I think may be something to keep in mind is that when it seems like there are a LOT of possible ways to improve on something, or things to explore, etc, often times it means a filter of some sort is required. The “entry level” filter is market states. It’s a great example because it’s not so easily done, and everyone does it in their own way. Yet, it’s fairly safe to say that those who don’t manage to create a market state filter die in the market. A filter is somewhat abstract and unclear, and some sort of brute force is necessary to first discover that it exists, and then to track and create the proper criteria to name it. The brute force application of it is NOT to uncover a signal,  but rather a pattern. This is something I’ve had a lot of trouble with in the past; I think the trick, (although extremely hard to do for me) is to be able to approach a concept that works very well in multiple ways and to apply to as many areas of the market as possible. Every piece of info in the market is “useful” if I can manage to find a use in it for detecting:

1. Entry
2. Exit
3. Breakout signal
4. Breakout confirmation
5. Reversal signal
6. Reversal confirmation
In the aspect of transience
7. Areas where price will return
8. Areas where price will not return
etc

There are more, I think it helps me to break apart the market into as many pieces as possible, and be aware that something that may not work out so well for the aspect that I’m looking for may work pretty good in another area. Working backwards and in hindsight is actually a useful tool in research.

doubled

I think it’s possible to build a layer on top of the original layer.. But I will need to think about what it is accomplishing a bit more before I get into filter-ception land. However, there may be something to it.. In this first filter I ran a filter, and then extracted all of the percentages that were at least 70%. Although, for the most part they’re all in a healthy percentage, save one.

Optimized h values

I’ve been trading a bit (small) for the past week or so and just trying to use as much of the information as I can. My R:R setups aren’t all that great, but I have some MM ideas I’d like to try out which will boost my return; however currently the whole idea of accelerating growth is one that I haven’t spent too much time one, I’m still focusing on “being right”.

prettypicture2

Odds are stacked fairly even on this one I think as far as which bound will be hit first, although I think I favor long just a tad more.

When odds are stacked, who wins? Knowing which type of signal is more likely to be right, or overtake another signal seems to be the kind of thing that SB is referring to when he talks about his metabrain. Not too hard of a concept to understand once you understand that building a system is about calculating probabilities for things to occur and not occur, rather than being fixated on the traditional signal->TP/SL->gogogo that new traders are sucked into.

______

As for h values, I found the current way I wanted to go about studying optimal h values. Per usual the whole idea may lead to nothing, but in the scope of how I envisioned the design, what I have works.

h1 data, probabilities for price to return to bar in the future (right side transience)

hold

This is data on right side transience, and it actually says quite a lot about holding periods. First, 6% of bars are not returned to within a month. More practically, if I’m targeting any bar, it is most likely to return in a day or 2. If I’m anticipating it to take longer, I’m best off waiting for a full week. After that, It’s most likely to not return at all. In other words, what is not reached within a week will likely not be reached in the next 3 weeks. Building on that, each weeks extremes form resistance points for the next 3 weeks. This seems like a VERY good thing to know about the markets. Researching for an edge in here has 2 potentials: seeking both periods for h <24/48/120, or for h>480. When periods are likely to be small, target it. when periods are likely to be larger, use it as a stop loss.

Crude PoNR (Point of No Return)

I think this a very interesting concept that has a lot of different angles to be approached from, both from levels and waves. You can calculate these based on a ratio of a previous wave, or fib fans/fall lines, pure pip movements, H/L of particular bars, etc. When I use the term PoNR, what I really mean is a launch point; a point where if price breaks out of it, it won’t return for some period of time. Using the similarity thread principles, this would be somewhat similar to finding points that are right side transient.

Practically, this is very useful for verifying when a trade is probably good to go and doesn’t need to be monitored anymore. Additionally, it can be used as a point to add onto a trade.

Filters filters filters. Using the correct frame makes all the difference.

filter1

Given that the average wave is 100 pips, the first filter (left) gives me an 80% edge after price has moved 70 pips or so. However on the right, the same accuracy can be achieved after only 25 pips…After a reference point has been created. How large is the reference point? Typically smaller than 45 pips.

Working on catching extremes

I think I’m kind of in a stage for a while where I’m just trying out various things, a very experimental approach rather than a methodical step by step. However the experiments are conducted within the frame of the overall methodical process which I will not forget.

I decided to put on CCI onto my chart and just take a look at it. CCI is one of the indicators that I played with a long long time ago when I was a young lost grasshopper. I only traded CCI peak divergence, and the net goal was very similar to what it is now: picking the top and bottom when the trend looks like it will die and boost into the other direction. I really quite like the indicator. It works fantastically in range and slow trending markets, and very poorly in strongly trending markets. I put the 24 CCI and a 24h transient bar indi and noticed this:

CCI

Now, it kinda looks like it has potential, but I already have no plans of using it (in it’s current state).

1. Good correlation between CCI +\- 100 levels and where left side transient bars are located.
2. Decent job of marking extremes.
3. “Binds” real price to only go so high or so low once a CCI peak has formed

The issue is that if CCI extremes and LST bars occur in the same spot, it’s no help. So, I decided to try and rip CCI apart and create a new formula. I realize that Meridian, TCD long, etc, and the “signals” you build from them are very similar to traditional indicators. The main differences being that traditional indicators have a knack for keeping bars separate ((H+L+C)/3, Close prices, highs/lows) whereas I think the “correct” view of the market is to use TCD deltas (H-pL, L-pH) as a substitute.

First attempt:

CCI1

 

Visually:

CCI3

What I’m looking for in the modded version is similar to how CCI normally functions:

1. For CCI extremes to correlate with real price extremes
2. CCI prices to oscillate between -100 and 100 most of the time

Looks decent I think, although pinpointing its effectiveness from a coding perspective will be very difficult.

Edit: Woohoo!

woohoo

 

Given that the low isn’t taken out in the next 12 hours or so, odds are quite good price will beat 1.2766

 

Complete Wave sequences pt 2

My original goal was to improve my stats, and in that sense I succeeded. However my other attempt which I thought was actually more in depth failed, while this one worked. Go figure. Either way, What I think I care about at the moment is being able to trade the trend states safely, as they are the ones that are most clean.

Capture (1)

What I attempted was to link the swing movements to the eventual trend change direction. I tried to frame the “correct” trend direction based on the trend state movement direction, which I think is ok. The current problem I may have is not being able to correctly assess which of those trend movements is “true” and which are merely retracement movements in the larger context. In other words, is the trend wave moving up because the OVERALL trend direction is moving up? Or is the trend wave moving up actually just a larger retracement in the overall DOWN trend? Quantifying this is possible, but I don’t think I want to look deeper into it just yet. At least not directly.

What I keep saying I’ll do but have yet to actually complete is a way to:
1. map all price movements based on their transient/recurrent types
2. Extract and analyze all possible wave types on all h values

Even though it’s feasible to do this, (and actually not too difficult, mostly just time consuming) I haven’t done it yet because I feel like it’s a bit of a “hands off” approach to the market. I just see it as sequencing and playing the numbers after they’ve shown. It’s missing an element that I think lies internally within a structure of a wave which will give a hint about what will happen next. How to find it is still something I’m unable to come up with an abstract for..

Complete wave sequences

Not completely complete in the sense that I haven’t accounted for other elements, but in the purest form, yes.

complete

No specifications about where price will be, only that the next fully trending move will be in the same direction as the one previous to it with about 60/40 odds. Solid base line I think..