Historic daily generation Vs Solcast prediction to refine daily PV forecast

Has anyone looked into getting Solcast tweaked to closer match your actual production for a given forecast, time of day and date based on historical performance data?

Solcast is great for around 9 months of the year for me but from mid November until mid January I get a lot of shading from my neighbours roof.

Without doing some work on actual Vs prediction for this time of year it's difficult to optimise daily usage, especially on the sunny days.

I've had a dig through various apps and online but can't find anything that gets any closer than Solcast currently is. Can anyone recommend a tool that will kick out a PV forecast and allow for time of year dependant shading?

have you used the tuning on the SolCast site?

Seye it's a dead feature so you can't use it anymore for domestic rooftop sites - there is a link on their site.

I just check, you are correct https://articles.solcast.com.au/en/articles/4945263-pv-tuning-discontinued

You could change the efficiency of yout site. So maybe just have 2 settings. One for the summer and one for when your panels are shaded. You can still do that in the rooftop settings. 2 mins admin twice a year

I also see in the link above that SolCast are now allowing a second site for rooftop users. Thats been asked for in this forum in the past. GE site doesnt accommodate that yet but it may be interesting if anyone can have a go and see if it works as a standalone feature. I only have 1 array so cant try it

Seye it would bring it a bit closer running the 2nd rooftop option, its is an exact science however and one that changes daily as the azimuth of the sun increases towards summer.

yandards tbh, I used to try and get mine very accurate. I was pretty sucessful, most of the time. I don't bother now, I really am just using using SolCast to trigger one of several SOC percent requests overnight. 25,50,75,100. As any spare PV power due to overcharing overnight goes to my hot water so I dont mind provided its not too much. I prefer that apprach now, just in case the forecast is wrong. I have been caught out by that in the past, and as the price difference between off peak and peak is widening further I dont want to be caught out again as 1 day in error cost 3 days of savings,

I use my own automation for the charging based off SolCast with a very blunt calculation that fudges the prediction based on my own experience. I am very conservative and expect less PV than the forecast. Worst case is I get lots of hot water and also stored energy in the battery carrying over to the next day ( which just means charge less tomorrow night )

I do know that some people are just having a seasonal value for overnight SOC, not even using SolCast. Same kind of broad settings. The difference between that very simple approach and a finely tuned algorithm is very small in practice. I know that as I have done both ( just for fun )

"Perfect is the ememy of the good" ;-)
So the question is "what is good enough"?

Completely agree that chasing perfect is pointless, still in my 1st 6 months of the system so working out how to get the most out of it. At this time of year it's just fully charge the battery regardless so simple for the same reason you cite.

Forecasts are just that, it's just there will eventually be enough data to give a reliable enough figure to do what you have self automated, ie an overnight top up in quarter increments.

I'm not in a position to run any other devices at the moment, my old Synology NAS is too old to run Home Assistant via a VM or anything else and aside from that it's doing what I want to it to.

I do consider the current option to directly drive development of the beta portal an amazing opportunity - if this was a larger company we simply wouldn't have this level of input.

My synology is old too (DS215j) . It doesnt have docker etc.
Even older versions of DSM can run NodeJS, which means they can run NodeRed. I know that as it thats how I run one of my instances. A little bit geeky, but quite straight forward
https://www.rs-online.com/designspark/installing-node-red-on-a-synology-nas

NodeRed can do all the automation you need, and more. Plenty of flows have been written by the community.
Later those flows can be incorporated into HA etc, so its an easy stepping stone should you wish to go down that route.

I have had my system just over a year, so I have gone through the cycle of intensive engagement with it and the learning that goes with that :-) Now its mainly just on auto with a few broad tweaks every few months. To be fair, its a lot easier and simpler on "go" compared to "Agile"

Interesting about the node red stuff, hopefully my DS412+ can handle it all - just need to make time to look at it.

Whole point is to offload as much as possible to reliable automation, I'm going to be away with work a lot more again soon and it just needs to work without my daily fiddling.