Category Archives: Statistic

RHS #2.8 Swing ratios as predictors

As it turns out, using the current swing ratio to determine if the swing should continue or not doesn’t have the support I thought it did. visually it still looks strong, but I’ll need to find another metric to measure it against the normal ratios to get it to shine.
Another idea I have is using the end swing point to predict the type of wave that will occur next. This is a bit easier to test if it works in the A/B structure because 3 swing stats are pretty easy to get. I normally wouldn’t have thought there would be any edge in this, but I thought I noticed something interesting about swings that has extreme ratios and the swing that came 2 legs after.

Following 3rd swing based on first swing, regardless of middle swing:
3rd swing
This is just the regular data set.

lr

Found some differences on down swings that aren’t there on up swings, which is kind of unfortunate.. Differences in rejection swings doesn’t matter too much, mostly because they’re rare to begin with, which makes the trading opportunities that come with them even more rare. There’s a small edge in DT swings leading to another DT/DJ swing (broken level) rather than DR by about 10%. Not too large..

 

RHS #2.7 Swing Ratios

t2
Each type leads to one of 3 other types. Total of 6 types, plus dom/non-dom, makes this one set out of 12.
These ratios still don’t tell me about the reversal probability, but rather about the continuation probability and when price may not have to continue. I think I definitely need a new TCD completely to try to capture reversal moves. Distinct Vega or Vega prime maybe? hmm..

The way I’ve been using this is pretty simple; if the current wave ratio is in an extreme in terms of the histogram, then price should make a new extreme on the chart.

current

I find the 0-140 ratios to work much better than the 100-500 ratios, and since the dom and non-dom are roughly flipped, I pretty much always have a ratio that is in the favorable range that I can use. The histogram for the 100-500 range is much more spread out, and harder to be accurate.

I still need to look at how these ratios play out in the charts, but my gut hypothesis is that ~.84 ratio suggests that price is a bit over-extended (being in a ~5% extreme), but needs to “fix” the ratio to be in an acceptable range. These numbers are averaged over the length of the wave, so this would mean that if price wanted to create a “correct” ratio, it would need to make a new extreme to print a new number, but only after some time, in order to drag the average down. In other words, the expectation is pullback or pause, followed by new extreme.

c price

We’ll see how this one plays out. If these ratios really work like this it would be awesome. Not quite what I was originally aiming for, but still big.
Edit:
fail
Not quite there. Some good opportunity for profits if I wait for a good R:R probability which I did get. but the map did not complete..

Edit2: I’ve been studying the ratios over wave time. Oh boy does the rabbit hole go deep.

RHS #2.5 Fill bars as R/S

The number is referring to the timing of the fill bar – 1 is the first fill bar that appears, 2 is the second fill bar that appears, etc. Probability that fill bar holds as support/resistance.
fill break
Kind of interesting to note that the 3rd-5th bars have a higher probability to hold than the first two.

Number of bars it takes for the fill bar to be broken:
number count
This one was kind of a bummer, I hoped that it would be much more clustered towards the lower count because it would map the overall map much more clear. As it stands though, I take it as a sign of no edge. Remember that markets have to move in order, so a completely uniform distribution is a very rare event. If they were more clustered, it would mean I could take this approach to the market:
-New wave occurs (still have to distinguish between the starting point of the new wave vs mini retracement in bigger wave)
-Wait for over/under fill
-If overfilling, wait for high of bar to be taken out, likely in next or 2 bars later (note: not the case here)
-Short
The probability to break is decent, but if I want to have an accurate measure of when it will break, I’d like to to wait until at least an 80% probability, which is 5 bars. To me it’s just too late, because 5 bars vastly increases the probability that the wave has already ended because if price is not going down (breaking the bar) then it’s going up and extending the wave.

Overall, there is a 86% (233/271) that the first over/under filled bar will be a continuation bar. In up moves, an underfilled bar will lead to at least 1 more higher high before the termination of the wave and vice versa for shorts. aka, there is a 14% chance that the first bar that is extreme filled is actually the extreme itself (done this before). It continues that there is a 66% chance that this bar is broken on the other wise (further pullback). So the most likely flow of price in these structures is to make an extreme, pullback, and continue, sort of like a mini-swing within the swing. These are more tradeable points, rather than the map given by the overall swing.

fill bar entries
These are kinda cool to look at, may be some extra things to pick up and test.

Rabbit Hole Series #2.2

PP
Calculating pip profit from the first signal entry to end of wave. Negative numbers, of course, for the entries that occur at the end of the wave. There were a few (<10 or so) that had a “valid entry” but occurred near the end of the wave anyhow, so the potential was quite low. The overall summary of this single aspect is roughly: 98% of the time if you take the first signal that appears in the true wave, there is a 70% chance to make at least 5 pips. Seems okay, but is missing the possibly large error that the system would require assumption of where the current wave start point is. Sometimes one would perhaps think that a UR is starting when it’s really just a small bounce in a continuation DT.]

The upside is that the large majority of the profits are in the over 25 pips category, suggesting that overall, this TCD does give the correct signal about where price can go. I think this is an okay start depending on what I can add on to it, I still need to:
Look at the other two scenarios: oversold in down moves and overbought in uptrend moves
Look at the TCD crosses or create a new TCD to interact with this one.

Left side SR zone

Haven’t quite worked on confirmation statistics, working on a couple of extra parts for now before bringing it together. Considering that catching retracements are easier than extensions, I think it’s important to work on the safest entry first.

To be specific, this is where the specific price, low in this case, lands within a single bar. Because of this I don’t think 5% is very different than 10%, maybe like 3-6 pips or something, but I think looking at the 25/50/75 marks are worth something just for eyeballing.

percent in bar

The extreme of the swing bar occurs within 75%, or close to 80%, of 1 side of the 50-50 line, the lower half. This is nice as it creates high probability zones, but still problematic and currently not as useful because there are a LOT of them.

Capture

I’m noticing that the extreme bar, if you trace it to the left, it seems to be very close to or in the wick of another bar.. Considering the 1/4 1/2 1/4 rule, I wonder why I feel like I’m seeing it a lot. I’m also noticing that ends tend to appear in areas where there is “air” or a portion of right-side transience. (there’s also a TCD reversal at that swing point)

air and wick space

Back to work..

Edit: about a 50% probability that the extreme is occurring inside a wick..

3