GOLD, Gold and XAUUSD - page 127

New comment
 

A mini-trend is ideal for taking profits.

Choosing a chart for a mini-trend isn’t just...


EXAMPLE on the M30 chart:


An upward mini-trend...

 
Serqey Nikitin #:
It’s very good for locking in profits.

There are no open positions on the chart, so there’s nothing to lock in.

Oh yes, that’s the Gurus for you – they only base their advice on historical data.

диванный аналитик - сам не торгует, но пытается сформулировать умные изречения . (не vnik до сих пор)
 
AxelQuant #:
So, the lower the exponent, the larger the brick size needs to be to filter out the mean-reversion noise? Is that right?

The UnformityFactorScript automatically displays the best estimate of the current timeframe/custom chart' data, and prints the factor to the Experts tab. Values 0f 50.0 and above are deemed ranging markets, while values below 50.0 are deemed trending markets. In actual practice, the factors of Renko charts are not quite as intuitive as those of traditional timeframes (nor the table that you previously posted). The factor can go up or down as you progressively increase the Renko brick size. If you're able to deduce a factor of 47.0 or so, you'll see some obvious trending price runs.

 
Ryan L Johnson #:
I see, so you are using market memory information independently of building the Renko bars. I was thinking that if we increase the bar sizes during periods of mean reversion, it could protect us from whipsaws by isolating the trend. However, the way you do it, using it as a filter, likely works well too.
And by the way, Hurst exponent values indicate memory the other way around: where 0 means perfect mean reversion, 1 is a pure trend, and 0.5, as usual, represents complete random walk.

I've never built Renko charts before and used to analyzing charts in relation to time. I'm curious, what does it feel like to trade outside the dimension of time? Do you completely avoid peeking at standard charts to see, for example, how much time has passed since a high or low, or do you take a comprehensive approach and look at both?
 
AxelQuant #:
[S]o you are using market memory information independently of building the Renko bars.

Not quite. What I'm doing is rather stupid-simple. Please allow me to clarify:

  1. I open an M! chart─unlimited history.
  2. I attach a Renko chart generator.
  3. I use the Fibo Sequence to start building Renko charts:
    1. If I want to be scalping, I start at 3 big points/pips─depending on the specific instrument.
    2. If I want to be swing trading, I start at 55 big points/pips.
    3. I run the UnformityFactorScript on the Renko chart, and note the factor number from the MT5 Experts tab.
    4. If the first run factor is not several "big points" below 50.0 for a trend strategy, or several "big points" above 50.0 for a range strategy, I go on to the next Fibo Sequence number (see #4, below).
  4. I reset the Renko chart generator to that next Fibo Sequence number, and repeat the process until I reach an acceptable factor.
  5. I apply my strategy (EA) to that acceptable Renko chart.
  6. Finally, I do extensive backtesting on Renko OHLC data─my EA's are coded specifically for doing so.

Some underlying assumptions, herein, are that more historic data is always better than less, and history tends to repeat itself. To be clear, I don't worry about recent data sampling/modelling with this method.

 
AxelQuant #:
[W]hat does it feel like to trade outside the dimension of time?

Mechanical, and detached. Price becomes binary. It either goes N points up or N points down.

Of course, my EA has a micromanaging time filter so the dimension of time still hangs around in that regard.

AxelQuant #:
Do you completely avoid peeking at standard charts...

Yes.

I should also probably mention that I'm talking about traditional Renko bricks here. Using gapless/rail-to-rail bricks or even wicks just tends to import noise back into the chart. My EA's analyze two prices─Renko open, and Renko close.
 

Thank you for sharing your strategy, it's really interesting. So, in the end, you find the right block size empirically.

I'm also trying to wrap my head around this: when building Renko charts, we make the chart simpler but lose time-related information. Is there actually anything useful in that lost data, or not? Ideally, if the target isn't met, we just wait until it is, or we take a loss. So it seems like we can manage without time. But personally, I find myself comparing time intervals a whole lot. Anyway, it's a very interesting topic, definitely need to backtest it.

 
AxelQuant #:

...when plotting Renko charts, we simplify the chart but lose information relating to time. Is there actually anything useful in this lost data, or not?...

Other properties emerge:

1) The development equation — an arc on the ordinal event scale of Renko-like charts, beyond which the price will not go (can be used as a support/resistance level)

2) Velocity lines — analogous to the Gann fan, but in the event plane, where the price pauses on each line before jumping to the next.

This is described in an old article by André Duka in his *Quantum Analysis*.

It’s an intriguing concept, but its usefulness remains unproven
 
Ivan Butko #:
Other properties emerge:
Yeah, exactly, that's interesting.
 
AxelQuant #:

Thank you for sharing your strategy, it's really interesting. So, in the end, you find the right block size empirically.

I'm also trying to wrap my head around this: when building Renko charts, we make the chart simpler but lose time-related information. Is there actually anything useful in that lost data, or not? Ideally, if the target isn't met, we just wait until it is, or we take a loss. So it seems like we can manage without time. But personally, I find myself comparing time intervals a whole lot. Anyway, it's a very interesting topic, definitely need to backtest it.

You're most welcome.

Renko trading really is more like a 2 dimensional way to trade, unless we consider volume/tick volume analogous to physical depth─because a good Renko generator will include volume/tick volume in the bricks as well. I guess that would keep with reality wherein time is the 4th dimension. So we could say that 1 of 4 dimensions is omitted from a Renko chart. Anyway, an indicator like the VWAP that calls volume/tick volume and bar position will still run on such a Renko chart.

A good Renko generator will also include timestamps in the bricks, if not in the underlying ticks copied as well. That has good implications for using iTime(), for example, in an EA.

Keep in mind that the way in which Renko bricks print, a 10 pip brick for example, can actually cover a maximum of about 29.9 pips prior to its final close. Of course, that can take more time than say, an M1 bar to print. Some traders don't like Renko because they simply can't wait for their next trade to trigger. IMHO, that is not a valid reason to dismiss any trading logic. But yes, Renko definitely takes some getting used to. My EA was in a trade this morning, price stalled at 2 bricks in floating profit, I went and mowed my lawn, returned to my terminal to see a few more bricks of floating profit, and then my EA closed it while I was washing up. No hurry. This is all a matter of personal preference.

Again, the most accurate way that I've found to backtest a Renko chart is to base all EA analysis on Renko open and close prices, and then test on OHLC data. If you want to add wicks to your Renko bars, the Tester will still have you covered with the HL of course. I have applied Heiken Ashi bars to large Renko bricks in the past. In that case, wicks are necessary. And of course, your EA's order management logic has to reference a tradable symbol─I use an input string for that. Then in the Tester, I select the custom Renko symbol, the M1 timeframe, and run the test. You could potentially test on real tick data if your Renko generator supports that, but I don't get into it.

1...120121122123124125126127128129130
New comment