Every automated trading tool makes one decision before any other: who holds the pen. We made it early, and it has shaped every feature since. The engine drafts. You click.
That sounds like a slogan, so let me say what it means in practice.
What "drafting" means
When OptionAgent proposes a trade, it hands you a complete order, not a hint. The structure and the legs, the size, a limit price, the per-leg greeks, the margin it would consume, and a few sentences of reasoning in plain English — including what would make the idea wrong. If a suggestion can't show its work, it doesn't get shown.
The same holds for the Portfolio Assistant. Ask it to draft an SPX order and it writes one, with the strikes, the price and why. It will not send it. There is no setting to let it.
What "clicking" means
Nothing reaches your broker without one of two things from you:
- An approval, for a suggestion or an assistant draft. You review it, change the size or the strikes if you like, and send it yourself.
- A switch, for a strategy. A strategy trades on its own only after you switched it on, under rules you set, on an account you chose — and it stops the moment you switch it off. Paper first, if you want to watch it for a while.
Both are visible. There is no third path.
What this rules out
A lot, deliberately.
- No trade "the model decided" that you can't trace to a rule you set or a button you pressed.
- No hidden sizing. Margin and a per-order notional cap are checked before an order leaves the desk, and you see the numbers.
- No acting on bad data. If the market feed stalls or restarts, the engine holds rather than fire a target or a stop on a stale quote.
- No surprises after a restart: stops and targets are reconciled with the broker, and the first login of every session waits for your approval on your phone.
It also rules out some things people ask for. We don't let the assistant place orders, even with a confirmation prompt. We don't let a strategy change its own rules mid-session. We don't "learn your preferences" and act on them. Each of those would move the pen.
Why we think this is right
The honest answer is that we don't trust automation we can't read, and we build the tool we would run our own money with. A rule-based strategy that we backtested on years of minute-level option data, that trades a structure we understand, inside caps we set, with every fill reported to us as it happens — that we will run. An opaque model with a send button, we won't.
There is a second reason. Markets have hours where the right action is to do nothing, and software is bad at nothing. Keeping the final click human keeps that option on the table.
What it looks like day to day
You open the app to a briefing written from your own accounts: what moved, what expires this week, what needs a decision today. The ideas waiting for you sit next to it, each with its reasoning. Your strategies, if you run any, have already done their morning's work and told you about it in your chat channel.
Then you decide. That's the product.