Monthly Archives: May 2015

Roundabout thoughts

Capture

It’s certainly not perfect, but interesting to note.. How to make it more accurate, or rather, what filters to apply, is an interesting problem to tackle. I find it feasible to be very clever and aggressive with position sizing and entry points to make some great profit, but:

1) testing and
2) long term repetition

are my main concerns. It may be worth tackling now that I’m much more in tune with how things “work” (although still have plenty to learn), as opposed to what I think a lot of newbies try (and I did) when I was first introduced to trading. It’s very easy to get into the “what if I.., what if I..maybe I can..” mindset. The problem is that when one is unaware of their surroundings (close by S/R, unknown conditions/movements/unfoldings, the position is dead.

cap2

There are some interesting “gut” things that occurred in this trade I took. The trend is up and therefore you should only really be comfortable taking longs. As an aside, one of the random thoughts that I had is that from what I’ve done and “confirmed” with others, it’s actually quite hard to know where a continuation wave will end. This may have something to do with what is actually occurring in a (up) trend: people are buying, bandwagoning, adding on, etc and this makes it hard to know how high it will really go. On the other side, retracements of the larger range may be more likely to hold because the people who are currently in sync already have additional limit orders that are waiting. The aggregate of these may be enough to protect the levels. Kind of elementary theory, but thinking it through helps me a bit.

Anyhow, as price unfolded and moved up (and therefore meaning that me being sneaky peaky like was looking to go short) I only had a few spots short, given my current toolbox. The main one I’m using is the simple 5 bar fractal. I’ve considered doing a more rigorous test, but haven’t quite done so (writes in notepad). The trigger for this is just the break of the range. There is plenty of room for a fake break and a disastrous position. But it kinda works out nicely when you can mindfully get some other factors in your position, and it’s great when the range is not too big.

When you need 50-60 pips as a stop and the wave is only really expected to hit 150 or 200 (and often times 100-130), you only get a set up of 2:1 or 2.5:1. Ideally you want something like 4:1 I think. The only two windows for entry here are shown in the two yellow boxes. The more aggressive option is to take the position and then add somewhere in the blue box. To me, the stalling here means that the top is less likely to break higher. Now I am beginning to see that it also functions as a flat as well. Doubling up on risk – but does this new position present better odds than the previous? I’m wondering if the individual sizing of the bars during a move are important to take into account..

Some updated stats on wave movement

I’ve expanded a bit to include the possibility of multiple waves (as alluded to in previous posts) occurring, rather than only looking at 3 leg waves and guessing about the probability of specific legs to follow.

I did some work on what I call single, three, and five wave breaks/stats

wave stats

Our original thought going into these stats should be “trends rule” because that’s what we’ve seen so far. Therefore, it might be alarming that there is a slight edge to a one ‘leg’ wave reversing rather than continuing. However, if we think about what this means a little more, it makes some more sense. This is because in any “trending leg”, that is directional move, retracement, directional move in the same direction, the second leg is broken in the reversed direction to complete the trending state and therefore counting towards the “reversal” probability shown. This also happens in expanding 3 leg moves.

Untitled

From this specific model, 3 wave and 5 wave continuation stats are very similar, so I decided to work with just the 3 wave stats to make it more simple.

Of course, i can’t know a wave is completed until it’s done, but we can use the rule of “almost there” and diminishing probability in a lot of cases to learn more about how price moves.

1

 

If we can expect a certain kind of leg to appear, it becomes cakewalk really. The expanding condition is much harder to capitalize on than the spring due to the natural movement of markets and R:R involved.

If we follow breakout theory there’s actually some good evidence for it:

2

97%, but there’s the 3%, or 4 cases that I’m scared of. In reality, to account for spread, i’m scared of anything that’s lower than 7 pips. The dream is really 10 pips I think. Even so, there are only 13 cases to look at. In those cases, the P/L on an entry – bar close are as follows:

loss

Some notes

Usually I just do this on a notepad saved locally, but sometimes people like reading them 🙂

Premise: check to see where edge is, and during the process, see what needs to be improved.

Current numbers to keep in mind:

  • -LG pivots: daily, weekly – independent pivots. If possible, target as safe exits. Safe SL too.
    -Maybe work on building a position during the week with the ultimate loss at the weekly LG pivot?
  • -5 bar fractal – 24hr fractal (26 bars?) If these occur in specific spots, they are more likely to work.. But which spots?
    • -Time spots. If I only take specific time spots, I will miss out on some trades
    • -Retracement spots. Pure numbers are too vague. Statistically, the best spots to reload are anywhere before 70% retracement of the full range. Perhaps use a reversal trade in those spots? “Once you’ve discovered that you’re in the wrong direction, get into the correct direction”
  • Time element +2 hour buffer from previous wave
  • Natural areas of continuation
  • Likely weekly PA
  • Price has some possible areas of turning based on previous projections

Questions:

  • What is strong on a “entry signal” basis, the 5 bar fractal or time? Or 2 bar time fractal? The biggest issue with the 5 bar fractal break is that the size is too big.. does it need to be?
  • Probability of a time element failing if it meets the requirements?

 

Test:

  • Accuracy of 5 bar breaks to predict next wave
  • Accuracy of 5 bar breaks to predict some retracement or projection (maybe projection of the 5 bar range or current range)
  • Accuracy of 2 bar time fractal to predict next wave
  • Accuracy of 2 bar time fractal to predict some retracement or projection (what range to use?)
  • Accuracy of 5 bar breaks with a time element as well to predict next wave
  • Accuracy of 5 bar breaks with a time element as well to predict some retracement or projection (retracement of the current range)

Probability of completion if trend+retrace+5 bar fractal+”defense” of a level=?

If I’m going to be using the LG pivots, then each of those tests must also take the current 1 day, 2 day, or weekly pivot into account..

A wave cannot shift until PTZ8 is met at the very minimum.

 

Next idea

An idea I want to test out.

current

How to properly construct and test is the question.. Observations first, followed by designing and testing.

Edit: Update

update