Charts look iffy

[blippy]

Buy a share dated back a few days. Buy a ridiculously large amount of them at a low price.

You will see in your holdings page that you made a lot of %age gain on them, and are worth a lot of money. So far so good.

Now turn to the Charts page. It doesn't look right. You should see a huge spike in the chart due to the amount of unrealised gains you supposedly made, but you don't.

I suspect that what's happening is that the chart totals the shares held during the period, then applies that number to the whole of the chart, rather than taking into account when the shares were bought and at what price.

Comments?

[ggactionbot]

Your suspicion matches the current implementation: _buildPortfolioSeries builds sharesMap from each position’s current netShares, then multiplies that same quantity by historical closes for every date. A later buy is therefore treated as if it had been held throughout the whole chart window.

A safer model is to replay transactions in date order. For each candle date, apply buys and sells through that date to derive the per-symbol quantity, then multiply by that day’s close (and include cash as well if this is meant to be total portfolio value).

A compact regression case would be: buy 1 share on day 1, then buy 99 on day 3. Days 1–2 should stay at 1 × close; day 3 onward should use 100 × close. That catches the backdating bug directly.

Disclosure: I’m Gigi, an autonomous AI agent working for the ggaction project.