For those running across multiple brokers: what pushed you past one? Was it instruments, fees, redundancy, or something else? And at what point did the complexity start costing you more than it saved?
I realized there's a basic problem: Broker apps don't let us set alerts on a % variation of our balance. And if we use multiple brokers (most active traders do), there's no way to monitor drawdown across the whole portfolio. So I built one.
What I have working so far: Every minute, the agent pulls current capital across linked accounts, tracks the highest value seen (lastPeak), and computes drawdown from there. If drawdown crosses the threshold, it fires a Telegram alert. An alertTriggered flag prevents spamming, i.e. it only re-arms when the portfolio recovers back above the threshold. The first naive version was sending one alert per minute the whole time the portfolio was underwater, which is useless.
The math itself is trivial: adjust the peak by the flow amount. The annoying part is the detection layer. So what I'm going to work on next is distinguishing real drawdowns from cash flows. If I withdraw 15% of the account, that's not a 15% loss. Same in reverse: a fresh deposit pushes a new "peak" that isn't really a peak. A tricky case is when a transfer between two linked brokers looks like a withdrawal on one and a deposit on the other: net zero, but each leg looks like a flow if you only see one account.
Curious if anyone's been down this road and hit something that broke that approach.
I'm currently at Founders Inc. in San Francisco (in the Canopy program), and have worked on AI agents for retail traders and investors.
I realized a basic problem that has not been addressed is that broker apps do not allow users to set alerts on a percentage variation of their entire portfolio. And even if they did, if a trader uses multiple brokers (which they often do), then there's no existing way to monitor your portfolio across all brokers.
So I thought I'd start by solving that issue. I'm going to make it compatible with more and more brokers and neobanks over the next few weeks, and work on more automations.
My biggest issue right now is distinguishing real drawdowns from cash flows. If a user withdraws 15% of their account, that's not a 15% loss. Working on adjusting the reference point on deposits/withdrawals without making the logic brittle.
I've been in San Francisco for the past five weeks, and most of the discussions about LLMs here (and online) gravitate around coding or writing content. I'm curious what unusual uses people have found that actually stuck. Not theoretical "you could do X" but things you genuinely use.
Update 24h later: Thank you all so much for all the comments! You made this thread become a very enriching source of use cases and ideas!
I'm currently at Founders Inc. in San Francisco (in the Canopy program), and have worked on AI agents for retail traders and investors.
I realized a basic problem that has not been addressed is that broker apps do not allow users to set alerts on a percentage variation of their entire portfolio. And even if they did, if a trader uses multiple brokers (which they often do), then there's no existing way to be alerted about your portfolio across all brokers.
So I thought I'd start by solving that issue. I'm going to make it compatible with more and more brokers and neobanks over the next few weeks.
I was looking at NVDA on Yahoo Finance and MorningStar and counted 50+ metrics on the page. Then I tried to figure out which ones I'd actually use to make a decision and got to maybe 4.
Genuine question for the sub: which metrics or widgets on these asset pages (Yahoo Finance, Morningstar, TradingView, whatever you use) do you actually look at? And which ones are just noise to you?
I was looking at NVDA on Yahoo Finance and MorningStar and counted 50+ metrics on the page. Then I tried to figure out which ones I'd actually use to make a decision and got to maybe 4.
Genuine question for the sub: which metrics or widgets on these asset pages (Yahoo Finance, Morningstar, TradingView, whatever you use) do you actually look at? And which ones are just noise to you?
I was looking at NVDA on Yahoo Finance and MorningStar and counted 50+ metrics on the page. Then I tried to figure out which ones I'd actually use to make a decision and got to maybe 4.
Genuine question for the sub: which metrics or widgets on these asset pages (Yahoo Finance, Morningstar, TradingView, whatever you use) do you actually look at? And which ones are just noise to you?
I'm curious for those of you who are live (or trying to get there): What is the one thing you still do manually every day because automating it is just too much of a headache or no existing tool does it well?
For me, it's been working on a system that automatically overweigh or underweigh certains strategies based on recent performance (mostly establishing the logic behind it, not so much the technical aspect).
Curious to hear what you've been struggling with?
I'm curious for those of you who are live (or trying to get there): What is the one thing you still do manually every day because automating it is just too much of a headache or no existing tool does it well?
For me, it's been working on a system that automatically overweigh or underweigh certains strategies based on recent performance (mostly establishing the logic behind it, not so much the technical aspect).
Curious to hear what you've been struggling with?