AlessandroAceSame here. I have a separate account for my monthly rent, which is debited the 1st business day of every month (usually the 1st actual day of the month). I top-up the account as soon as I get paid on the 21st, but for some reason the graph thinks I’ll have a stack of money left after the 1st and that the money is gradually put into the account during the month.
FreekWizardWhen the world moved from algorithms to A"I", we moved from hard logic to black boxes. Not always better :)
It would help if they would include scheduled transactions into the learning
and it would help even more if we could mark expenses (as well as salary payments) as "recurring" (so that they show up in the "scheduled" category with a frequency that we can define) and those where factored in as well - because then the only expenses that actually need to be predicted in the graph are regular but not plannable expenses like groceries
Amber 🇳🇱🇬🇧WizardWow, today I noticed the graph shows a very very accurate prediction 😍
So I guess it just needed some more time to learn or maybe the latest update improved the feature, but anyways: I’m impressed!
Keep up the good work 🌈❤️
For me the graph is not very accurate either. It seems to somehow take an average of spendings on a certain day, but not take into account whether this is a recurring payment.
JakobProdigyRight now my graph shows a spending of 2000 euros tomorrow and then it goes up by a few hundred euros every few days. My payment history is of course nothing like that. But... well I guess it may just need some more time. Or my spending is just so unpredictable 🙃
MartijnChampNever, unless they change the algorithm. For example, I have one account that I transfer the same amount of money to once a month, every month. At the end of the month it’s empty due to debit requests from various companies. The graph thinks I’m going to add extra money at random times during the month. It’s all over the place. And this is a very predictable pattern, the ING bank could predict stuff like this with much more accuracy. I disable the graphs for all my accounts for this reason.
ThijsProdigy@Martijn-Yellow-Monkey#152132 Exactly the same for me. I've been adding exactly the same amount to my account every month for months, but the bunq algorithm still doesn't seem to recognize it.
Amber 🇳🇱🇬🇧Wizard@Martijn-Yellow-Monkey#152132 Well I actually see a lot of improvement. No random additions are predicted anymore to accounts I only add money to once a month. Just like you described actually, I have a few of these accounts where I add the exact amount to that will be deducted during the month. And finally, the graph really shows me a realistic prediction. It’s harder for my ‘random transactions’ account, but even that one is not all over the place anymore and shows a somewhat reasonable prediction.
@Ambiebambi#152305 Unfortunately this still isn’t the case for me. I have certain accounts (like an account each for holidays, clothes and hobby) to which I add x amount every month. I sometimes do take some money out, but if I look at the graph at any given time it’s prediction is that tomorrow I have 0 euro left, and then it slowly ramps up again. This makes no sense because 1. I hardly ever take the full amount out in one go, and 2. I never slowly ramp the amount up, but always add a lump sum.
The graph in its current state, even for pretty passive accounts, is simply not useful and currently feels more like a gimmick than a useful feature.
To address this, my suggestions would be that but takes away some of the ‘noise’ from the subscription. If they see I add 100 euro every month, than that’s a reliable transaction, but if they cannot verify the date, they should try and estimate the date and not make a ‘build up’ of 10 days 10 euro. Further, if there is an inconsistent amount taken from the account, it should not reflect in the graph for the next month.
I would also suggest to look at ING their ‘look forward’ feature. That is imo a more useful interpretation of a similar concept.
Miro KnejpRookieThe graph is completely broken for me. Even the historical graph (which I expect to represent actual balance and not come from an algorithm) is so wrong, it's not funny anymore. It thinks my account has 3x the balance it actually has, and instead of going down where I spent money it goes up. The tooltip keeps saying "Prediction" for historical data. That alone is just wrong to me.
I can accept that the future prediction is unreliable, but please, make the historical graph accurate and show your actual historical balance. Without that the graph has zero information value because I cannot accept any predictions, however accurate they may be, if I have to assume it's based on historical data that is utterly wrong.
As a first step I'd be great if the event history showed the new balance on every transaction, since that is information that is currently inaccessible to me without doing the math by hand. That might be easier to implement than fixing the graph and would help a lot by itself.
JakobProdigy@Miroslav-Silver-Falcon#159988 Historical balance should be accurate. That is definitely not "predicted" or anything. If that doesn't work for you, it's a bug that should be fixed.
