22 screens / about 9 minutes / narrated by your browser
Screens captured live from the beta instance, signed in as a demo account. Every position shown is fictional seed data - no real book, no real money, and nothing here is advice. Narration is spoken by your own browser; nothing is sent anywhere. Space plays and pauses, arrow keys step.
This is Trader. It is an options book built by someone who runs ratio spreads, calendars, covered calls and short puts, and it is opinionated about what it puts on screen. This walkthrough assumes you already know what extrinsic value is and what rolling a position means, so it skips the basics and goes straight at what each page computes and why you would trust the number. Everything you are about to see is a demo book. The positions are fictional and nothing here is advice.
You get in with an invite code, and your account arrives preloaded with a fictional book rather than an empty page: a share portfolio, a covered call expiring this week, a call ratio, an iron condor about five weeks out, a calendar into an earnings print, and a short put that has already been rolled once. That last one is deliberate. Position software tends to look fine right up until you roll something, so the demo book ships with a roll already in it and you can check the accounting before you trust it with your own trades.
Home is triage. Book value with the day's change, realized month to date, how many campaigns are open and how much is at risk in them. Underneath: anything expiring inside a week, and your live campaigns with realized and unrealized split apart. One habit of this app shows up here and everywhere else: each panel loads on its own, and when one cannot load it says feed unavailable rather than showing you a zero. Zero and I-don't-know are different answers and they never get confused.
The book. What is worth explaining is the marking. Options always mark at the mid, because the options book freezes at the close and the mid stays honest overnight. Stock outside regular hours marks at the last trade instead, because after hours the market makers pull their quotes and leave a bid-ask so wide that the midpoint becomes fiction - a thirty-seven eighty bid against a forty-one forty offer will cheerfully tell you the position is worth thousands more than anyone would pay. And a leg with no quote at all stays blank the whole way up through every subtotal. It never quietly becomes zero.
The day column has three rules worth knowing. It measures against the last trading session rather than the calendar, so a weekend view does not compare Friday against Thursday. A leg you opened during the session measures against your own fill, not against a prior close it never traded at. And money booked on something you closed today gets added back into the account and total rows. Look at AMD: that day figure includes the profit on contracts that left the book this morning. On a day you rolled something, the total still reads honestly from yesterday to today.
Expanded, the book groups by ticker and then by structure - the calendar, the covered call, the ratio - with each leg's size, basis, mark and gain. The per-leg actions sit on the right: close, edit, move between groups, roll, or hand the whole structure over to the builder. Look at the AMD front leg: it shows a size of one against a two-lot basis, because half of it has already been bought back. The page keeps the contracts you still hold and the ones you have booked separate rather than averaging them into one confusing line.
Getting positions in does not mean filling out forms. Describe the fills in plain English, paste a positions table straight out of your broker, or paste a screenshot of an order confirmation. It reads whichever you give it and shows a preview of everything it thinks it found, including which existing leg each fill matches against - and that preview is editable. Nothing reaches your book until you apply it, and anything that looks like a duplicate defaults to skip, so pasting the same confirmation twice does not silently double your position.
The exit desk is for positions coming up on a decision. The column that matters is extrinsic - what you are being paid to wait. A short collects it if the stock sits still; a long bleeds it. Look at the KO ninety call: nine cents of extrinsic with three days left. There is essentially nothing left to collect on that short, which is the whole argument for deciding about it now rather than carrying assignment risk into expiry for a few cents. Beside it sits what it would cost to flatten the entire group at mid, priced as one combo rather than leg by leg.
One detail that tells you whether to trust a page like this. If any leg of a group cannot be priced, that group is dropped out of the desk-wide close total entirely, and the exclusion is counted on screen - you can see it in the header, one group excluded, unpriced legs. Adding up a spread that is missing a leg produces something that looks exactly like a combo price and is not one. The number is either complete or it is absent and labelled.
Assignment intelligence rides on the same rows. Any short that goes in the money gets flagged. If its extrinsic falls below the upcoming dividend with an ex-date within about three days, it gets flagged harder, because that is precisely the setup where someone exercises early to capture the dividend and you find yourself short stock. There is also a pin flag when spot is sitting within half a percent of a short strike on expiry day. These are conditions the desk reports, not opinions it holds - it shows what is true and otherwise stays quiet.
Exposure answers the question a line-item portfolio cannot: what are you actually long? Every position becomes delta, then delta notional, then gets multiplied by that name's beta against the index - so the whole book is denominated in index-equivalent dollars and can genuinely be added up. Names group by theme rather than by sector, because sector classification files bitcoin balance sheets under Technology and collapses every fund you own into a single bucket. Leveraged funds use their stated multiple against the base index rather than their own regression, since a three-times fund's own returns drag decay and reset artifacts into the estimate.
The shock curve moves the market across a range and reprices every option at each point rather than scaling delta, and that is the entire point. A covered position loses more than delta implies on the way down as its short calls shed delta. A long-call position gains more than delta implies on the way up. The plus and minus ten percent columns on each row are that repricing, and where they disagree with the straight-line estimate the position is convex and the page says so. Volatility is held fixed across the grid, which means it understates a genuine selloff - stated on the page rather than left for you to assume.
This is the page that earns its keep. Look at the XOM short put. Two lines: the July one-fifty, bought back at three-ten against the two-forty credit you sold it for, booking a hundred and forty dollar loss - and the September one-forty-five that replaced it, labelled rolled. Then read the header, because that is the part most software loses: entered for a four-hundred-and-eighty-dollar credit, and the roll itself paid you ninety dollars. A roll is rarely free, and its cash normally disappears inside the new leg's cost basis. Here it is a column of its own.
Same page, the AMD calendar. The short front renders as two lines: one closed at nine-eighty with three hundred and seventy dollars booked, and one still open at the original basis. That is a single position in the database shown as two, because that is honestly what it is - contracts that are gone and contracts still exposed. Merged into one line, the closed half vanishes and the row reads as though you are still carrying the whole thing at the original size.
There are three different cost figures on this page because they answer three different questions. At-risk-now is what can still be lost from here. Entry cost is what the original ticket cost, counting only the legs you actually opened - legs a roll created are excluded, because their cost belongs to the roll line and counting both would double-count every roll you have ever done. And for ratio structures there is a modelled worst case, because a hundred-and-fifty-dollar debit can carry fifteen thousand dollars of uncovered width above the short strike. Cost is context. Cost is not risk.
The builder prices a structure against the live chain, and it prices it under six fill conventions at once - from the natural price, where you cross the spread on every leg, all the way through to mid. On a two-legged spread with wide markets those can be dollars apart per contract, and mid is a price you will simply not get filled at. The stats read off your own editable per-leg prices rather than the model's, so what you see is what the ticket costs you, not what the theory says it should.
Two things about these numbers. When every option leg shares one expiry, max loss and max profit are exact rather than sampled: payoff at expiry is piecewise linear with kinks at the strikes, so the extremes can only live at a kink or out in the tails, and the page labels them at-expiry-exact. Mixed expiries fall back to the grid and say so. And the probability number is labelled model POP - it is lognormal at a blended volatility, not a market-implied probability, and it is labelled that way precisely so you do not read it as one.
The lightning bolt turns on earnings mode. It reads the at-the-money term structure and backs out the market-implied one-event move - the forward variance between the first expiry that carries the print and the next one that does not. Then, for every column at or after the event, it strips that event variance back out of any leg spanning the print. So the profit grid shows the position after the crush instead of pretending the volatility you paid for survives the announcement. A pre-open print is carried by an expiry on that date; an after-close print is not, and the two are handled differently.
The scenario lab is the two-dimensional version: spot down the side, an implied volatility multiplier across the top, at any number of days forward. This is the grid that explains the earnings trade that goes your way and still loses money. The stock moved where you wanted, but every column left of centre is volatility coming out of the position, and for anything long premium that is exactly where the money went. These are model values at your stored entry prices - not executable quotes.
The events page puts the market's implied one-event move next to a deserved move - the median absolute reaction that name has actually printed historically - and flags the ratio as rich or thin. It states two caveats on its own face rather than burying them. Realized spans bracket the print close to close, so the deserved figure runs slightly high and rich therefore errs toward fair. And these are display flags, not trade gates. Nothing on this page decides anything.
Every strategy the app knows about lands in one ledger, and the column that matters is the thesis. What you write when you put the trade on is what you grade the decision against later - not the profit and loss, which you already know by then, but whether the reason you did it turned out to be the reason it worked. If a position was tagged as rule-following or discretionary at fill time, that tag survives every subsequent edit, so the attribution stays honest even when you rewrite the note.
That is the tour. If you want somewhere to start: put a structure you are genuinely considering into the builder and check whether the six fill conventions match what you would actually pay for it. Take a position through the lab and see whether the volatility axis behaves the way you expect. And paste a real broker confirmation into the portfolio page to see whether it reads it correctly. There is a help page covering every screen and a question box in the corner of every page. Tell Dan where it is wrong - that is what the beta is for.