Pivot points

[freqtrade]

Pivot points

[xmatthias]

This indicator will not do "daily" at the moment - but will use a rolling approach.

You can make it use a rolling of 1d (timeperiod parameter) - but then it'll be windows from 00:00-24:00, 1:00 - 0:59, ...

As an alternative point (Edited in) - try using informative pair (same pair, '1d' resolution) - calculate the pivot point on that and then merge back to the original dataframe - that should get you exactly what you need.

---

additional note:
I don't think the "daily" approach is suitable for crypto anyway - that's made for traditional markets, which have a daily open time - where "daily" is really relevant, as it has a meaning for the market cycle.

Crypto markets however are "always open" - so if you'd use "daily" windows (like in the screenshot above)- which timezone will you use for crypto markets, which are always open?

Your own (for convenience)? UTC? or maybe US, or Asia, or some random European timezone?

Therefore, using a rolling daily parameter (as this indicator is doing) is actually a better approach in my opinion.

Now that does not mean it cannot be changed to also do the "batched daily" - but it's currently not implemented - and i don't think that for the forseeable future, i'll have time (nor will) to implement this - so if you' think it's a must, i'd encourage doing a Pull request adding this feature.

[aescsa2]

I agree you, but anyways the problem might be almost solved using weekly changes for the suports and resistance values I guess, there would be a problem of just hours on every last day of the week?, Hope you excuse me about my inexperience, but how could I use those week timeframes for the calculations of the indicator, to keep those supports and resistances values unchanged, and still be using 5 minute charts for my trading? Thanks again for your time :)

[xmatthias]

you can't really use weekly candles with 5m candles - it's too long, so after ~3 days, your weekly candle doesn't have an overlapping candle to join to the 500 5m candles.

However, using weekly candles is NOT what tradingview talks about, but about daily candles (where the described method works fine).

You should be able to repurpose parts of this sample - obviously with a lot of changes to actually get pivot points instead of RSI as shown there.

Best plot the results afterwards (freqtrade plot-dataframe) - so you see what you'll get.

[MichaelMacca96]

Hi xmatthias,

I Implemented the pivots_points function and as you suggested, I
".....try using informative pair (same pair, '1d' resolution) - calculate the pivot point on that and then merge back to the original dataframe - that should get you exactly what you need....."

It works but after a few hours, the Pivots Lines on FreqUI disappear!

I'm working with a timeframe of 1m and I want, if possible, to have the Pivots_point on a 1d resolution.

Thanks!

Best Regards!

[xmatthias]

Merging assumes that at least one common time is available.
That's not the case between 1m and 1d, as the 2 are too far apart (you get 1000 1m candles, which will not give you a 00:00 candle in the last part of the day).

i doubt this makes sense though - if you look at 1d data, you'll not need signals at the 1m timeframe (which is usually way too noisy to get good signals out anyway) - so best consider switching to a slightly bigger timeframe like 5m or 15m.

[MichaelMacca96]

Perfect! Thanks for your help!