How this works
Everything here takes about ten minutes a day. Read the bit for whichever job you've been given — the app tells you which one you have each morning.
The short version
Every market day the app picks two photographs. One of them stands for "the stock market went up today", the other for "it went down". Nobody is told which is which.
One person (the Viewer) tries to describe the photograph they'll be shown at the end of the day, before it has been chosen. They only get a number to work from.
A second person (the Tasker) reads that description, looks at both photographs, and says which one it matches better. Only then does the app reveal which photo meant up and which meant down — and that becomes the day's guess about the market.
At the close of trading the app checks the cash S&P 500. UP means today's close finished above the previous trading day's close. DOWN means it finished below. An unchanged close is void, the same as a holiday.
Your day
| 7am | A notification arrives. Your jobs for the day are on the Home screen. |
| Morning | If you're the Viewer, do your session and submit it. |
| Once a transcript lands | If you're the Tasker for someone, score it and make the call. |
| Just after 4pm New York | The market closes (9pm UK most of the year). Results appear, and the Viewer sees their photo. There's a Get results button on Home if you don't want to wait for the alert. |
Most days you'll have both jobs — viewing for one person and judging for another. You never judge your own session.
If you're the Viewer
you describeWhat you're given
An eight-digit number, like 4271-8830. That's it. The number means nothing in itself — it's just a label, so that you and the photo are talking about the same thing.
What to do
Sit somewhere quiet with the number in front of you, and write down whatever impressions turn up. Aim for plain descriptions rather than conclusions:
Good: "cold, wet, rough grey surface, something tall and narrow, a
repeating pattern, water moving."
Less good: "I think it's a lighthouse in Scotland."
The moment you name a thing, your mind starts filling in details to fit the name, and those invented details are what spoil a session. If a name pops into your head, write it down in a separate list — the form has a place for it — then go back to describing.
Five to fifteen minutes is plenty. Sketches help — a CRV template runs to two pages, and the form takes a photo of each. Attach both if you filled both in; one is fine if that's all you did.
The one hard rule
Don't look at any market news, prices, or charts before you submit. Not the app, not your phone's finance screen, nothing. If you already know what the market did, the whole trial is worthless — you'd just be describing the photo you expect to be shown.
Afterwards
Once you press submit, that's you finished. You can't edit it, and you won't see the photos until the market has closed.
When results come in, you're shown one photograph: the one attached to what the market actually did. Compare it against what you wrote. That single image is the whole point of the exercise — it's how you learn what your own impressions feel like when they're right.
Your tasker may also have left you notes on how they scored it. Those appear underneath, and they're worth reading before you decide for yourself how well you did — it's easy to talk yourself into a match once you've seen the picture.
If you're the Tasker
you judgeStep 1 — score the photos, blind
You'll see the other person's description and two photographs, labelled A and B. Give each photo a score out of 7 for how well it matches what they wrote.
At this point you have no idea which photo means "up" and which means "down", and that's deliberate — it stops your own opinion about the market leaking into the scoring. Judge only what's in front of you: does this description sound like this picture?
Two or three clear hits on distinctive details beat a vague overall feeling. If neither photo really matches, low scores for both are a perfectly good answer.
Step 2 — make the call
Once your scores are locked in, the app reveals which photo was "up" and which was "down". If the description matched the "up" photo better, the day's prediction is up.
Then you choose: UP, DOWN, or PASS. Pass is a real option and not a failure — if the scores were close, or the transcript was thin, passing is the honest call. Passes are recorded separately and don't count against your hit rate.
There's also a notes box on that screen. Whatever you write there goes to the viewer once the trial resolves — not before. It's the most useful thing you can give them: the photo tells them what the target was, but only you can tell them which of their descriptions actually landed and which led nowhere. Your scores are locked by then, so say what you like.
About the money side
No money is staked anywhere. When a trial resolves, the app shows what a small bet would have returned, at £1 a point, from the previous trading day's close to today's close, so the numbers mean something concrete. It's an illustration only — nothing is placed, and nobody's account is touched.
Real trading on this would need far more evidence than a handful of trials, and the run of results below 50% is exactly as informative as the run above it.
Odds and ends
Why some days say VOID
US holidays (Labor Day, Christmas, and so on) are skipped the same way weekends are — no new targets that morning. If a session was already run on a closed day, or today's close finished exactly level with the previous trading day's close, it's voided: no photo, and it doesn't count in anyone's figures.
If you miss a day
Nothing breaks. The trial resolves without you and is excluded from your stats. Entry closes when the result comes in, so a missed session can't be filled in later.
What the Stats page is telling you
| The viewing | How often your description matched the photo tied to the real outcome. |
| The judging | How often you, as Tasker, scored the real-outcome photo higher. |
| The trading | The up/down calls, counted twice over — see below. |
Every trade has two authors, so the same trades appear in two tables. Calls made is your record as the Tasker: you read someone's session and made the call. Sessions worth following is your record as the Viewer: how those trades did when your session was the one being read, whoever judged it. The last table crosses the two, so you can see whether a particular pairing does better than either person does on average.
Coin-flipping scores 50% in the long run. Anything close to 50% over a few dozen trials is just noise — don't read a story into a good week.
Pools
You're in one or more pools. A pool is just the group you're paired with: sessions rotate inside it, and its Stats page shows only that group's people. What another pool is doing doesn't appear anywhere on your screen.
Changing your PIN
Tap your name in the top corner and you can set a new one yourself, or leave the box blank and let the app pick one. It's shown once and then stored scrambled.
If you've forgotten it and you're signed out, nobody can look it up — not even Marvin. Ask him and he'll set you a new one.