How it works

How Optix works

We grade your decision, not your luck, and rate your predictions against the market’s.

Make a prediction. See the chance the market gives it. Get graded on your decision, not your luck.

1Predict
What happens, by when, and how sure you are.
2Compare with the market
Your chance next to the market’s, from real prices.
3Get graded on the decision
Win or lose. Then your rating adds them up.

What a thesis is

Your prediction, written down before you know the answer: what happens, by when, how sure you are. Then you pick one of three moves.

An example thesis

“NVDA closes above $220 by Friday. I’m 70% sure.”

We read this as: NVDA closes at or above $220 on Friday. That exact claim gets priced and graded.

Place with OptixPaper: Pretend money on Optix Paper, filled at real quotes. A filled order counts toward your rating at half the weight of a real-money trade.Real: Real money: you placed it at your broker or approved it in Optix’s trading ticket. Optix records it; nothing here sends a real-money order.
Paper now · real money in Optix. An Optix Paper order counts half. Real money counts in full.
Paper tradePaper: Pretend money on Optix Paper, filled at real quotes. A filled order counts toward your rating at half the weight of a real-money trade.
Optix Paper, pretend money. Filled at real quotes, it counts at half the weight of a real-money trade, and it’s sized the same way.
Just predictPrediction: No trade. Only your chance is scored, against the market’s.
No trade, just your chance. Only your chance is scored, against the market’s. When no trade here beats doing nothing after costs, Optix suggests this one.

Optix never trades on its own: every order waits for your tap, and real money needs your approval in the trading ticket.

How placing works

Review: Optix re-quotes the exact contracts and shows its review: the limit, the cost, the cash left. Nothing is placed yet.
Place order: The one tap that sends an Optix Paper order. It shows “Working” until it fills, with Cancel, and the thesis opens at Optix’s own fill price.
Real money: Nothing here sends it. Optix’s trading ticket opens with the same contracts and you approve the exact order there. The fill comes back here and gets graded.
Placed elsewhere: “Already placed it elsewhere? Log it” records a trade you made at your broker, at the size you took.

After you act

While it’s open: Your exit plan watches for you. We alert you only when a stop loss, break price or take profit hits.
At the deadline: You get a review: the grade for your decision, and whether your chance or the market’s was closer.
Over time: Your record adds them up: your rating, what you said vs. what happened, and where your money leaks.

What Optix is and isn’t

Optix is

A notebook. Write a thesis before you know the answer: what happens, by when, and how sure you are.
A calculator. It reads the market’s chance from real prices, then suggests a trade, and a size set by how sure you are.
A coach. Every thesis gets a review with plain reasons. Over time, your rating shows whether your chances beat the market’s.
Places trades. Optix Paper in one tap when you confirm it, or real money through the trading ticket. Every fill comes back here and gets graded.

Optix isn’t

A broker. It holds no money. Real-money orders go through the trading ticket, where you approve each one.
Autopilot. It never trades or sells on its own: an Optix Paper order goes out only when you tap Place order, and alerts only fire on the exits you set. An AI agent can send orders only under automatic rules you turn on in the Optix terminal’s Agents settings (Agents here shows how to connect one); those trades come back here tagged Agent and are graded like yours.
Copy trading. You can’t copy anyone’s trades. You can challenge their thesis with your own chance.
Investment advice. Optix is educational. A suggestion is arithmetic on your own numbers.

The $10K challenge

A free trading challenge on a simulated $10,000 account at real market prices. Pass a season and you’re Funded for it: a mark on your profile, not cash.

Season 0 is by invite: codes go out with Optix’s posts and emails, and the Challenge tab takes yours with three questions. Your run gets 20 trading sessions of its own, from the day it starts. Each trade is a thesis first, then an order on your run: long calls and puts, debit spreads and butterflies, up to $200 of risk each, checked against the rules before it’s placed. Reach $10,800 with every other rule met and you pass that moment. Equity at or under $9,200 ends the run. A filled challenge trade counts three-quarters toward your rating.

The Season 0 rules

The $10K challenge

Season 0 rules

Start: $10,000Simulated. Free to enter.
Target: +8%Reach $10,800 to pass.
Max loss: 8%Equity at or under $9,200 ends the run.
Daily loss: 3%Down $300 from the day’s start ends the run.
Per trade: 2%Up to $200 of defined risk: long calls and puts, debit spreads and butterflies, no same-day expiries.
To pass: 20 tradesEach risking $50 or more, on 10 trading days or more. No day over 40% of the profit, a of 70 or more and at most 2 .
Who: US 18+US residents 18 or older, in the 50 states and DC. One account each.

A simulated account at real market prices. Every order is checked against these rules before it’s placed, and your equity is checked on every price mark.

Read the full rules

Your chance and the market’s

Real prices carry a chance for your claim: the market’s chance. It’s the number to beat.

The example thesis

You say 70%. The market says 45%.

A 25-pt gap: you think this is far more likely than the market does.

Where it comes from. For a stock, option prices give the chance it closes beyond your target on your deadline. For an event, it’s what Kalshi’s traders pay. An empty market has no price, never 0% or 100%.

Your number stays yours. Your grade and your rating use the chance you said. Suggested sizes use an , pulled toward the market’s until your record backs your number.

The Best fit and its size

The is the trade worth the most to you. One knob sets its size: how sure you are.

Optix prices shares, call options and spreads at every usable price, and never suggests a trade that loses money when you’re exactly right. It ranks the rest by : the profit you expect, charged for the swings, so a long shot can’t win on its average alone.

You never type a size. Every thesis shows one line and one chip that names what set it, for example “Suggested: risk $120 (1.2% of your $10,000).” and “Set by your chance”. Two limits can only bring it down, the same for real money and paper.

How sure you are: Your chance against the market’s, trusted as far as your record backs it, on your trading money. Your sizing style (Careful, Standard or Bold) sets how much of it to use. No edge after costs means no size: Just predict.
Your per-trade limit: A setting: never risk more than 1%, 2%, 3% or 5% of your trading money on one trade. 3% of $10,000 is $300, so no suggestion risks more.
A max loss, if you want one: Optional, for one trade. Write “risk $150” in your words, or tap Set a max loss. It can only make the size smaller.

How size is graded

Going lower: A paper trade, an Optix Paper order or a Replay round can take less than the suggestion, never more.
Logging a real trade: Optix records what you actually placed, at any size. Above the suggestion, it says so.
How size is graded: One Size reason per review. Over your limit by more than 25%, or more than the math allows at your own chance, is a Mistake; over 2× your limit, or 3× what the math allows at your own chance, is a Blunder. Optix’s own suggestion is never charged.

Set your trading money and limit

Grades judge the decision, not the luck

Every trade gets one grade, for the decision. The result doesn’t decide it: a good decision can lose money.

Grades: the decision against the result
The decision	Lost money	Made money
Good decision	Good Sound, and it didn’t work this time. Still Good.	Good Sound, and it worked.
Bad decision	MistakeBlunder A clear error, and it cost you. A serious one is a Blunder.	Lucky You made money despite a clear error.

Prediction No trade to grade. Your chance is scored against the market’s.

Nobody knows the true odds of one trade. So we grade what you decided with what you knew.

Grades judge the decision. Your rating also checks whether your chance beat the market’s.

What a grade looks at

Each review checks four things, each scored out of 100. Every reason, with its numbers, is in The math.

Prediction: Do your chances beat the market’s? 50 = even with it.
Trade choice: Did you pick the best-fit trade for your thesis?
Entry price: Did you buy near the fair price?
Risk: Did you risk the suggested amount, or less, and keep your losses inside your limits?
Depth: How much your record covers: counted theses, different stocks and events, months and kinds of trade.

Your rating and your level

Your rating measures skill against the market, which sits at 1500. Profit never counts.

1500 = the market’s level. Good decisions alone put you a little above it. Beating the market’s chances moves you higher. It’s public after 20 trades a broker, Optix Paper or the challenge filled; until then it’s provisional and stays off your profile and the leaderboard. Your level marks what your graded trades show, Rookie to Elite. It never limits what you trade or how much, and once you reach a level it’s yours.

1 RookieWhere everyone starts
2 Trader20 graded trades
3 Sharp60 graded trades
4 Elite150 graded trades

Each level also asks for different days, a process score and, from Sharp, chances that beat the market’s.

How the rating counts

Public at 20: Before 20 filled trades it’s mostly a guess, so it stays off your profile and the leaderboard: you see it marked provisional, with its likely range. Your first blind round or graded trade starts it.
A likely range: Your record shows where your rating likely sits, like “likely 1505–1759”. The range narrows as theses add up.
Simulated money counts less: A challenge trade counts three-quarters and an Optix Paper order half. A paper trade is an Optix Paper order.
Predictions: Just predict and Matches entries count in your forecast score, a second rating of your chances against the market’s. They never move this one. Games count in neither.
Blind rounds: The first five each day count. Your first 10 count even before your first trade; after that, blind rounds stay within a fifth of the rating.
A blowup counts double: Losing more than 1.5× your limit in force (the smaller of your per-trade limit and any max loss you set) counts double against your rating. Suggestions never go past your limit, so this happens when you size over it yourself or a price jumps past your stop.
One event counts once: Theses on the same earnings report, game or Fed meeting count as one.
Selling early: Your decision is graded when you sell. Your chance still waits for the deadline, so selling early neither dodges nor banks it.
Beta: These weights are beta: they may change after Season 0.
Logged or imported: A trade you log or import is graded like any other, losers included, and stays on your record outside the rating: no broker, Optix Paper or the challenge saw it fill. Imports are graded against your per-trade limit, because Optix didn’t size them.

When to trade less

After 100 or more counted theses, if your chances haven’t beaten the market’s, Optix says so plainly. For example:

“After 140 theses, your chances haven’t beaten the market’s. That’s worth knowing. Just predicting for a while costs nothing and keeps your record growing.”

What each level asks

The four levels, what each asks for and what it unlocks
Level	To get there	Unlocks
1Rookie	Where everyone starts	All of Optix, from day one
2Trader	20 graded trades · On 5 different days · Process score 60 or more	Your level shows next to your name on the leaderboard and your profile.
3Sharp	60 graded trades · On 15 different days · Process score 75 or more · Chances closer than the market’s	You can be invited to be an Optix creator.
4Elite	150 graded trades · On 40 different days · Process score 85 or more · Ahead of the market	You’re listed on the Levels page with every Elite trader.

What counts: Trades a broker, Optix Paper or the challenge filled, once they’re graded. Predictions, trades you type in, imports and Replay don’t.
Money: Your per-trade limit and any max loss you set size every trade, real money or paper. Your level never does.
Kept: Once you reach a level, it’s yours. A bad stretch never takes it away.

See your level and what’s next

Replay and blind Replay

Practice on real past earnings reports: you see only what was known before, give your chance, then see what happened.

Replay

Five past reports a day. You set your chance before the reveal, and it never defaults to the market’s. Past results can be looked up, so Replay has its own rating and its rounds never touch your main one.

Play a round

Blind Replay

A real earnings report with the company, the dates and the price level taken out. Make your call, then see which company it was.

Blind rounds count toward your rating again: the first five each day. Your first 10 count even before your first trade; after that, blind rounds stay within a fifth of it.

Play blind

Either way, a round is sized like any trade: your chance sets the size, under your per-trade limit.

Matches and 1v1s

Scheduled events everyone predicts before they lock: earnings reports, the Fed, inflation, jobs and big games.

Where the numbers come from

Real prices where they exist, a labeled estimate where they don’t, and never an invented price.

Our promises

What Optix will never do, so it can’t quietly turn into the thing it’s meant to train you out of.

The math

Every formula behind the pages above, for anyone who wants to check our work.

The market’s chance

Read from real prices, the same way for everyone.

p = N(d2), where d2 = [ ln(S/K) + (r − q − σ²/2)·T ] ÷ (σ × sqrt(T))

Stocks: the chance, under the options’ own prices, that the stock closes beyond your target K on the first listed expiry on or after your deadline. S is the price now, σ the implied volatility at that strike, T the years left. Down theses use 1 − p.

p = (C(K1) − C(K2)) ÷ (K2 − K1)

When both strikes around your target have tight quotes (gap at most 10% of the midpoint), the call spread between them is a near-direct quote on “ends above K”. We use it when it agrees with N(d2) within 10 points.

p = (Yes bid + Yes ask) ÷ 2

Events: Kalshi’s midpoint between the best Yes bid and ask, else the last trade. A No thesis uses 1 − p. Bid 44¢, ask 46¢ → 45%.

p = P(close > S × (1 + m)) + P(close < S × (1 − m))

Earnings matches: the chance the first close after the report is more than m away, either way, read from the option chain at the first expiry after the report.

Closes, not touches. We score where it closes on your deadline, not whether it touches the price on the way. Touching a price is roughly twice as likely as closing beyond it.

Adjusting, sizing and ranking

The worked example shows each with its own numbers (See a worked example).

adjusted = sigmoid( logit m + λ × (logit s − logit m) )

m is the market’s chance, s is yours, and λ (reliability) is the share of your past gap, in log-odds, that held up. It’s fitted on your record by maximum likelihood and starts at 50%, worth 20 theses, for everyone.

w(x) = w_market(x) × ( adjusted ÷ m ) above K, × ( (1 − adjusted) ÷ (1 − m) ) below

Jeffrey conditioning: the smallest change to the market’s spread of closes that matches the adjusted chance. Prices above your target K get more of the market’s weight, prices below get less, and the shape on each side stays the market’s.

size = min( k × full Kelly, cap ÷ unit risk ), where full Kelly = the q that maximizes E[ log(1 + q × P&L ÷ W) ], k = ¼, ½ or 1 by your sizing style, and cap = min( per-trade limit, max loss )

Kelly: the bet that grows money fastest if the adjusted chance is exact. Full Kelly is the ceiling, never a target. Careful sizing takes a quarter of it, Standard half and Bold all of it, then the cap in force can only lower it: the smaller of your per-trade limit (a share of W) and a max loss you set (if any), real money or paper.

CE = W × ( exp( E[ log(1 + size × P&L ÷ W) ] ) − 1 )

Worth to you is the certainty equivalent: the sure amount you’d take instead of the bet, at the suggested size, on your trading money W. E[…] averages over every close under the adjusted chance. Under the market’s own chance, every trade should lose about its costs, so any edge comes only from your chance.

Prediction score and edge

Brier = (chance − result)²

The Brier score: how far a chance landed from what happened (1 if it happened, 0 if not). Lower is better. The market gets the same treatment with its own chance.

prediction score = 100 − 100 × Brier

The prediction score turns it around so higher is better, from 0 to 100. Saying 50% every time scores 75.

edge = mean( score_you − score_market ) ± 2 × SE

Edge is your average lead over the market’s prediction score. The ± is a 95% range (two standard errors). Ahead of the market means the whole range is above 0.

match score = score_you − score_market

A match or 1v1 score is the same lead on one question.

The rating

A Glicko-2 rating, played thesis by thesis against one opponent: the market, fixed at 1500. Each thesis is a game scored 0 to 1. Holding 1500 needs an average of 0.5. The λ starting point is half, worth 20 theses.

game = 0.55 − 0.011 × (100 − decision score) + 2.5 × (Brier_market − Brier_you)

A trade: up to 0.55 for the decision (each decision point under 100 costs 0.011), plus 2.5 × how much your Brier beat the market’s, kept between −0.5 and +0.45. Just predict: 0.5 plus the same term. The total stays between 0 and 1.

Game scores for example theses, computed by the engine
The thesis	It happened	It didn’t
Clean trade, a modest chanceYou 50%, market 45%	0.68	0.43
Clean trade, a big chanceYou 70%, market 45%	1.00	0.05
A mistake-level tradeDecision 75/100, you 70%, market 45%	0.73	0.00
A blowupLost past 1.5× your limit	0.00 ×2	0.00 ×2
Just predictYou 70%, market 45%	1.00	0.00

Above 0.5 gains rating against the market. Below 0.5 loses it.

Every grade reason

Every reason cites a number. Minor notes never change the grade on their own.

Entry price

Buy–sell gap paid: (fill − midpoint) ÷ (asking price − midpoint), in dollars. Over 60% on a quote wider than 5% of the price is a minor note. Paying more than the asking price is a mistake.

Risk

Loss kept to the cap: The cap in force is the smaller of your per-trade limit and a max loss you set, real money or paper. Losing past it is a mistake. Past 1.5× is a blowup: a blunder.
Size: One reason, two checks. Risking over 1.25× the cap, or over full Kelly on the chance you said (it grows money slower even if you’re right), is a mistake. Over 2× the cap, or 3× full Kelly, is a blunder. The sentence names whichever bound. Optix’s own suggested size, or less, is never charged.
Your own exits: Holding through your stop loss or break price is a mistake.

Trade choice

Best fit: What your trade was worth per $1 at risk under your chance, against the best on offer. Under 50% of the best is a mistake. Taking the best fit is never charged.
Your own math: A trade that loses money on average under your own chance is a mistake.
Expiry: Expiring before your deadline is a mistake. More than 4× the time you needed is a minor note.
Paying for a move: Buying a single option into earnings when your target sits inside the move options already expect is a mistake you could see at entry.
Thin at your target: An option that makes under 25% of its risk if you’re exactly right is a minor note.
No edge: Trading when nothing paid after costs is a minor note: Just predict was the honest move.

Prediction

Not graded: Whether it happened is the result, not the decision. Your prediction score moves the rating instead.

See it for yourself

The quickest way: the 2-minute demo, or one Replay round on a real past report.

Watch the demo Write a thesis