CandorTrade
AI Trading Journal: AI powered trading psychology coach.
I graduated with a CS degree a few months ago and wanted to build something real — not another tutorial clone sitting unused on GitHub. I'd always been curious about trading, so I started digging into why most retail traders actually lose money. The answer kept coming back the same way: it's rarely the strategy. It's what happens in the moment — fear overriding a plan, chasing a move out of FOMO, moving a stop loss because losing feels unbearable. Every trading journal I looked at recorded WHAT happened. Nothing captured WHY.
So I built CandorTrade — a journal that tracks the psychology behind every trade (emotional state, confidence, whether you actually followed your plan) and uses GPT-4o to find the behavioral patterns costing you money, with the real dollar cost attached.
Here's the part I actually want to talk about, though. A few weeks in, I ran a competitive review of my own product against the bigger players in this space. It flagged something I'd missed: my AI was reporting an "estimated annual cost" for each bad habit — a number that sounded precise but had zero real math behind it. The AI was just... making it up. Confidently.
For a product whose entire pitch is "we tell you the honest truth your overconfidence hides," that was a real problem. I rebuilt it to only report costs traceable to actual trades — real trade count × real average loss, nothing projected, nothing invented. It's a smaller, less impressive-sounding number now. It's also the first version I actually trust.
A few other things from the build:
→ Layered architecture (Repository Pattern) that let me migrate from SQLite to PostgreSQL by changing basically one file
→ Added email-based password reset because one early tester asked for it — first time a stranger's feedback directly shaped a feature
→ Deployed on Railway, live now at: candortrade.in
Where I'm at now: talking to real traders (Mom Test style — asking about their actual habits, not pitching) to figure out what's worth building next. CSV import? A pre-trade behavioral check before you even place the trade? Genuinely don't know yet, which is why I'm here.
If you trade, or if you've ever built something and had to catch your own product cutting a corner — I'd love to hear from you either way.
5 Comments
This reminds me that trust isn't built by making the AI sound more intelligent—it's built by making it comfortable saying less.
Rebuilding the feature around evidence instead of plausible estimates feels like one of those decisions users may never notice directly, but it's exactly the kind of choice that makes a product worth believing over time.
That's a really good way to put it, thank you.
You're right that this is bigger than just one fixed bug. I think I need to check the rest of the app with the same idea in mind — like, does the Weekly Summary always try to find a "pattern" even in a week where there isn't really a strong one? Does the Trader Archetype label sound too confident when someone's only logged 5 trades? I hadn't thought about it this way until you pointed it out.
Really appreciate you taking the time to write this instead of just scrolling past. Adding this to my list of things to fix.
I'm glad it resonated.
Reading your reply made me think about one broader implication of designing around evidence instead of confidence. I'd rather explain it in the context of your product than try to compress it into a few comments.
If you're interested, what's the best email to reach you on?
Happy to keep this going - my email's adityaworks2562@gmail. Genuinely curious what you had in mind, feel free to reach out whenever works for you.
Thanks! I’ve just sent it over.
Looking forward to hearing your thoughts whenever you have a chance.