# C vs TCD end points

It is very difficult to know if a vs b is favoring one side or the other. That is, if pivot formula A really is better, how do you know? Since no matter what you do, there are bound to be some variances in strength areas, I feel like you have to really know what you are trying to compare. I’ve wondered as I’m sure many others have about the difference and significance between using closes and using H/Ls. This by no means is exhaustive, but more like a test for a factor within a test.

This is tested for conditions of length >=10, since I think lower lengths cause a lot of issues of being caught within a flat. I was also considering fractal lengths of 7, but there are too many possibilities and its difficult to find significance.

Using Close values:

Using TCDs:

Hmm… Round 1 to the TCDs I think..

Completed for potential possibilites:

One of the times when low probability is still a good thing, because either way theres a way to take advantage of it.

# New Wave work

A lot more filtering opportunity now, although I’m not sure I will know how to progress and filter out the correct ones

Also, finished the first version of my pivot dashboard!

Edit:

The new post-waves are also weaker

There are simply a lot more flats now than there were in my old model. This isn’t necessarily good or bad, it’s just a function of the modeling technique. Given the way that the pivots function, strings of flats are more likely to occur, ex before a string of 3 flats was almost for sure to be broken on the next wave. Here, it’s actually almost common. Time filtering may play a role here, although I feel like it’s a very rigid way to do it so I likely won’t. Instead, since I have a lot more wave legs to play with, I think I’ll try a more elaborate wave sequencing filtering technique.

# PTZ Pivot in Semi MTF

This was my first early attempt at MTF and PTZ /Pivot combinations. Decent odds, but definitely not tradable. This first table is the type of “strategy” I would run maybe a few months ago. I might have done this exact statistic if not for the need for the pivot to already be in tact.

PTZ odds certainly favor bigger periods and in my opinion this is further proof of the need for a varying h value. By this I mean that 24, 28, 30, or some “true” h value doesn’t matter, but there is SOME minimum number that is needed, and it’s definitely higher than 1. Losers here are not time based (failed PTZ due to having a completed right side with no follow through) but pivot based. How far the pivot is from price varies, but I didn’t want to even bother with the poor success rate and the likelihood of the losses being much bigger than the winners.

Second attempt here, with min requirements for both PTZ and pivot. Turns out to be a bit better. I have 2-4 elements at play here: the PTZ number, the pivot price, the length of the current pivot, and the duration of the trade. THe last one I have not looked at all, the others I have but they’re.. slightly tricky. It’s difficult to know  to trade a minimum of both, or a minimum of each. Ex. min of 20 PTZ, 10 Pivot vs 20 PTZ, 20 Pivot. I didn’t expect much out of this one, but it did turn out to be a bit better.

Longs only here, but my own results on shorts are roughly the same)

Max requirements. Same story. At this point I realized that what I was probably looking for is some combination of numbers that aren’t identical. Rather, it’s probably some function of PTZ/Pivot that makes it work (if at all)

Past 24 (which is what I was looking at this week) is decently good, although I definitely think I need an additional layer now. I think Supplementary pivots are in order.

I wouldn’t necessarily call this “k” as it has been in the past, but still some sort of optimal value with very high precision. This week trading an 8 dollar account (LOL) I was about to get 20/23 accuracy using the k of 5. According to this chart I was slightly behind, but I did miss a few winners due to sleeping. Time to put some work into my dashboard for this one.

# To do list:

Tried out a few trades today and I’m fairly satisfied with it. My main goal is still to create it in a multidimensional frame. Currently it’s a bit rigid due to the stats I used to configure it. Those settings work, but I need to find all settings.

1. Complete for all useful h values (10-240, possibly more). This one shouldn’t be too difficult.
2. Multiple time frame recognition and calculations. This one will be a bit more challenging. I will need it to account for instances where two time frames are not in agreement with each other.
3. Create a secondary “constrictor” pivot to be more aggressive with stop losses. This one is optional, but if I don’t complete this one, I want to complete one similar.

I may be starting to understand the notion of finding a pattern that exists for over 95% of cases. I’m looking at a 99.5%, but I want more! Or at least, a more complete version. There is a difference between 97% being recurrent, and 97% of possible transient bars becoming recurrent. The first is a fact of the market. The second is a money maker in the market.

There are two more things to check for:

1. Which is the more correct “trigger” for the trend change?
2. Is it more accurate to project for the next period or just the current period?

Not quite significant enough yet but I suspect HH/LL might be better than C.. Very early stages, though!

# Double filtered pivots and TZ

Since I do consider this kind of an area that’s not fully explored and I understand the basic concepts well, I keep going back to transient bars. I end up describing them in so many different ways because well, it just kinda works out that way. Here i’m using them as momentum breaks, synced in line with pivots as stop points.

What I’m basically doing is trying to stack high probability odds with other high probability odds that sync well with each other.

Consider minimum profit criteria and what already favors us (stats for my own h value and TF, not universal):

1. The probability that a potential high or low PTZ bar is broken (trend continuing) is about 84.5-86.5%
2. The probability that the CLOSE of a PTZ is broken (trend continuing but not as ambitious) is about 97.00-98.00%.
3. The probability that the FIRST PTZ is the TZ of that trend move is about 12.50-14.50%

Further, consider this observation: In a subjective view, when the trend is down, how often does number 3 occur?

So when your trend “definer” is correct, how often do you get the opportunity to grab a couple of pips?

Might be ready for another testing run soon..

# MTF Everything and pivots

I really want everything I do to be within an MTF context, although sometimes I don’t know if that’s a good thing or a bad thing. On one hand, it makes perfect sense. On the other hand, maybe it’s not necessary when all you need is 10 or 20 pips. How can you tell anyhow?

As for the actual theory, I’ve recently been thinking that transient bars and pure probabilities favor trend following tactics. Wonderful. The issue has been time and time again, where to put the stop? I think Pivots are a way to solve this, and I’ve been putting in some time into thinking of what an MTF pivot looks like and how it plays out (and how I can see it play out in a way that’s quantifiable). The point of no return as I’ve studied it seems to be better studied under not the context of “launch bar”, but rather more like a “decision node”. These are better referred to as simple Support/Resistance areas. But In real life, I think these must be zones, not actual prices. I think to maintain flexibility, purely using a single price point for S/R is incorrect (using single price points still has its uses though). The reason I say “real life” is because the theory that I have is that true market makers (banks) aren’t stalking specific prices. They’re looking to get in and out at various prices, and often at whatever they can get. Unless True movers are staring at screens like we are, they’re more likely doing other things and make their entries based on “good enough”. That’s not to say they’re willy nilly about it, but I do think they’re more “discretionary” as we would say. Thus, to create a mechanical system for an occurrence that is truly discretionary is a bit tough to do (though I maintain the stance that it works).

So! For pivots to “work”, I think that in a MTF context, it should end up displaying a zone of prices that should be recognized as Support and Resistance zones. I should actually expect then that the percentage of pass through/no pass through to be okay or decent (not great) and for an additional filter to bring out the value in it. I will say though that often times I feel like I create goals and plans for things that I don’t actually have the resources or capabilities to accomplish. I still think they’re good to think about though hah.

The current pivot work and stats I’ve done look not too great as a start (the ones shared), but after playing with them a little, there’s actually quite an initial edge in them. Very interesting!

# Traditional V1 SR

I’m calling the pivot that I’m working with now Traditional V1, since it’s just the traditional formula, but on a rolling basis instead of fixed.

The next question I had was, “Do pivots function more as Support and resistance levels or more like momentum launch points?” Also worth asking is, which one is more important to me? If I could engineer one, which would I rather have it do?

I calculated the length of time price spends above/below a pivot before switching:

These results look a little similar to the initial transient bar results. 1s and 2s make up a large portion of the data. This means that at least in this pivot, the points act more like support/resistance than launch points.

V2: