Hi! Bob from
dvcforless.com here. Can you share with me the specific listing(s)? If it is a resort with very few listings, e.g.
Disneyland Hotel, the star ranking is less reliable. We are planning on overhauling the algorithm to account for cases like this and similar ones but just haven't gotten around to it yet. We're working on some other features first that we plan to release soon. I do think the stars are very useful in the vast majority of cases but I DO see the same issues you all see.
Listings are grouped by resort and "point bands", and we're looking back between 60-X days to establish market rate (varies depending on how many lists are available to compare agains). Point bands are ranges of points that help group small-med-large-XL contracts together. There are a few other parameters we look at as well.
There are pros and cons to our current approach. The cons are most obvious when looking at a resort with very few listings, or when two similar contracts are in different "point bands" despite being only a few points apart. A few months ago we worked out a new algorithm with the help from a data scientist that found us here on Disboards (thank you H). The new algorithm will eliminate these deficiencies, and change how we value stripped/banked points a bit. It is a big change so it will take some time to implement. Once we finish the current features we're working on. I think we'll end up targeting this.
By the way, one of the new features we're working on will allow us to notify you of new listings/listing-changes much much faster without us being a bad citizen to the brokers by hammering their site. We chose to prioritize this feature because of the feedback here on Disboards, so I do want you to know our roadmap follows the feedback we get from you all.