The Really Useless Thread

I believe .066x holds. It's a flip for me if it doesn't.

Looking for momentum play at .049x to 1.02 and then .98.

0640 unlocks on probability side higher levels more likely than not, not a good bear level to go by. Looking at directional bear agressive levels should be around current or a bit higher/lower levels.

This is the most sarcastic thread of all time.

1.061x & 1.051x...maybe too micro

Earleone,
if I understood correctly the order which you used, worked in the following way:
- you put a limit order at 1.0615. When this level was hit, a sell stop order was activated at 1.0605 with stop loss at 1.0630 ?

Yes, that is correct.

Nicely played XXX.000ish level break on JPY delivering more than not expect playoff. Tracking 200ish GJ for counter play atm otherwise all along JPY bull target to go along.

I read this thread so much I should play too sometimes. Upside: above 1.066 possibly to 1.1 ish, but unlikely. Downside: normal pullback after prior break of 1.045ish (which does not look like a bottom). I'm short with a stop above 066 (decent RR), if that hits will reevaluate. But watching dollar and yen.

From the 23 of nov this flip area has been support and resistance, a daily close below may be of interest.

Since 22 of nov this flip area has given resistance and yet to show support .

Shorted @ .062x ... S/l .063x

I think momentum north would be 1.064x and south the momentum play would be 1.041x

A method with high probability . Counting the Swing high / Low. . If we include a new level to HH LH LL HL named NSH and NSL ( NSH = New structure high , NSL = New structure low ) Kind of ZigZag counting with expectations.

When a NSH or NSL is made make the expectation that a NSH always is followed by a HL for a new NSH and a NSL always is followed by a LH we will on most timeframes have a accuracy between 70 and 90% ( depending on instrument , trending etc ) .

I attach a chart on 1H timeframe and daily for EURUSD here.

Looking at 4H or higher we are expecting a LH followed by a NSL . But all will leed to the same at the end. and when the NSH / NSL expectations is broken we have a change in trend for the tf we trade. The time to participate in the market with a trade is mostly when we fade an HL after NSH or fading a LH after a NSL.

Nice analysis. This is related to the Dow theory, if I am not mistaken
I have a question for the first picture: why don't you count the top at 1.06149 on 05.01 as swing high? Based on what we determine, which is a swing high and which simply a high?

In true time I would have done with hindsight you take what became the swing high/low. Anyway it doesnt matter much. All will guide you to the same result at the end. Just try to look at the clear structure high and low swings.

Lets see if we get 2180 tap on GU to clear some shorts, late micro shorts already squeezed. Unlocking some higher levels if we get there.

In fact he asked good question. It does not lead to same results at the end becouse you are adding noise effect to the pattern that you seek, which means that your performance will be distributed away from the statistical expectation and curve (you wouldnt even know if you dont do the homework). When you start fitting pattern x into pattern y it by mathematical default leads to noise in performance, the key of every strategy should be noise removal as much as humanly possible.
Ill have some time later today will try to post some examples. No disrespect to your post by the way, its great thing to approach things like that as trader, but its even more important to understand where back-forward testing failures are, and how noise affects them. Laters.

I actually disagree with that. All will lead you to the same because the NSH if putted on the place he mentioned would have been moved to the place I actually putted it after the fact. This is some off the difference by using NSH and NSL instead or in addition to HH and LL ( because a HH is not necessarily NSH ) The level he pointed out became a HH but did not become an NSH after the fact.

It is using the HH and LL that will add noise. Not the NSH and NSL .

I have marked a HL now in the attached chart. But it can still be moved. As long as we get an swing low above the last HL from yesterday it will be moved there if it happens before a NSH.

If you look the daily chart in my previous sample the NSH / NSL expectations is not broken since Mai month. But it dont tell us when/ where to take our entrys. Just expectations. That is why FA and having a map like Sisse has been showing us is a big advantage in addition ( or as standalone )

The noise will be factors in your pattern picking that are random inputs or adjustments, such approach as you described above needs to be very neatly defined with deep set of rules otherwise by all means you will have large noise impact on performance. Any highs and lows approach will have noise unless you adress additional variables or put things in context.

First thing that you land on is what L/H is actually L/H that you pick. If you dont have robust rules to pick them you have noise. A good way to clear a lot of noise would be to define them by volatility scale, just one example. If you believe that volatility is mean reverting then that would be a good input into rules.

Every strategy eventually has noise even if you try your best to reduce it, but certain strategies are less exposed to noise weakness and some are more. With strategy like nsl/H you are weak to noise simply becouse the value for your expection is not very precise on price. Meaning if NSL is formed at which price are you expecting LH to be formed? To clean noise for such approach what you could do is to only be interested in precise area, lets say 100% or 50% retrace after NSL/H and then run statistical performance count what happens around there. What youll find is that numbers will be quite low, but that will reduce noise and will actually give you better oportunity to tweak other variables of strategy to build positive expectation approach around it.

The whole point of statistical backtest is that you need to reduce noise as much as possible in order for statistical results to have actual mean. This is key reason why many backtests of variety of systems simply are worth 0 becouse there were key variables NOT adressed for backtest, which then perfectly reflects that once such trader starts to trade live markets results are completly off backtesting expectations becouse in live markets those noise variables will change performance drastically.




If you adress at least fe[h]w additional needed variables to make statistical count relevance (which is the whole point of approach you described above since you do not have cleanly defined "reversal" expectation) then you will find that statistical data will be much lower i can guarantee you that, i have been doing same mistakes in past on thousands of backtests and lot of them were eventually proven to be worthless due to oversimplyfing very noisy method . Noise within your own approach (unless i missed something if so feel free to point it out):


Again the whole relevance of post i just made is about statistical data quallity and actual relevance of your expectation which is based of your count. If you have lots of noise your expectations are actually going to be underperforming or overperforming out of expected curve, with lots of noise it can be severe.

Took a eu short at 607 last night. Very light. Stop at 670. I'll take half off at 545 if I get it. Entering slow. If we get <545 then back up near that entry and I don't stop out I'll add more. If south from here, well I'll have a tiny ticket for the ride down. At these prices I won't buy a big one. I don't think she's done climbing. But if I'm wrong about that it'd be OK. If I'm right the stop won't hurt too much. And if we bounce around in the tight 520/650 mini range it may be an opportunity to build a small short position where it's a little rich for my blood to go all in at once.