Feels like every single show that has some element of sci-fi just recommends me GoT and Walking Dead, kinda boring when I in this instance wanted to find shows similar to Loki and just got these ones, again.
3 Likes
Agreed. Mine are crap too!
2 Likes
These arenât personalized. We use the related items from TMDB if they have them. If not, we do suggestions based on the genre and popularity. Room for improvement though, I agree.
5 Likes
What about something like what anime sites do? People can add recommendations as part of reviews, or it can be a separate area dedicated to recommendations where people fill out why something is recommended. For example, for âStar Trek: The Next Generationâ you might recommend âStar Trek: Voyagerâ and âStar Trek: Deep Space Nineâ. People can then either âlike itâ or âagreeâ (only positive confirmation), and each time someone makes a repeat recommendation, their comment gets added. That all gets aggregated into a list of recommendations.
5 Likes
Try using couchmoney, it links to your Trakt account and you can filter recommendations. It might offer you more/better choices.
You can then find these in your âlistsâ section of Trakt.
3 Likes
Would you be fine with someone building a recommendations engine based on your data? Would everything one needs to be available through the API already? Or could be made available anonymized somehow?
Why not? If you can make it with the stuff in the API. Like couchmoney does.
I canât seem to find anyway to figure out who has recommended/reviewed what via the API: https://trakt.docs.apiary.io
Why do you wanna make it if couchmoney already works fabulous for recommendations?
Like itâs fine to also make it, but if you donât know how to with info from the API 
This is what the creator of couchmoney mentions in their site. Maybe you need data outside of Trakt too? No clue.
How does it really work though?
Basically, it finds people with similarly weird tastes to you, and recommends films and TV that they like more than average people.
Lists are ordered by confidence. The first item is the strongest recommendation.
It looks for âdiamonds in the roughâ: lesser-known titles you may have not heard of that you may like. It recommends these more than well-loved blockbusters that you probably know about.
This evolved from the board game recommender I wrote for Reddit users, which Iâve operated for free since 2014 from my home computer.
I used to license my recommendation software to clients, and now I use it for my personal enjoyment. Iâm happy if other people find it useful.
It doesnât use an off-the-shelf engine. It was written from scratch in Java. It uses tens of millions of film and television ratings to compile its recommendations for you.
1 Like
AGREED! I mean idk if Iâm in this boat alone, but I barely even look at my recommendations anymore because they are all sub 50 percent movies I would be interested in. Or, of course, every marvel or lord of the rings movie ever! 
In my opinion, couch moneyâs recommendations are subpar as well. Iâve tried multiple types of lists from it, and none of them have too many movies I would be interested in. I have a great system for recommendations, mathematically and theoretically speaking lol, problem is I know nothing about coding etc. if someone wants to help me build it though I have done some tests with it and I know it would work better than most options out there, shoot me a line!
We hear you and agree the recommendations need a lot of work on Trakt. A new recommendation engine is on our roadmap and weâre hoping to have some progress on that early next year. Itâs not an easy problem to solve, but weâre continuing to learn and figure it out.
5 Likes
Couchmoney is really awesome. I integrated several lists into kodi. Love it.
1 Like
Hey Justin: has it been implemented yet? My recommends are still garbage. One would think there would be a llm somewhere to do this, with all the hype around AI?