Trade Kalshi like a desk,
not a hobby.
EVDESK is the professional way to trade prediction markets. It is a real terminal that prices, sizes and places every order type on Kalshi, with one-click Kelly sizing on any market. The model’s probability sets the stake. One press sends the order to Kalshi from your own account, signed with your own key. Behind it sits a proprietary signal engine that prices only what it can defend and shows its work on every call.
| Market | Action | Price | Fee | Model | Edge | Size | Closes | Open on the desk |
|---|---|---|---|---|---|---|---|---|
| A sample of the board appears here. The rows are real Kalshi markets from an earlier pass, saved as they were priced. The live board, with every row, is on the desk. Start free trial → | ||||||||
An edge you cannot take
is not an edge.
A model price measured against the midpoint is easy to publish. The midpoint sits halfway between the bid and the ask. But nobody will give you that price, so you can’t trade it. Every number the signal engine puts on the board is quoted at the price your order would go in at. It comes after the fee that entry pays, and it only counts the size actually waiting there. When the model can’t defend a number, it stays off the board.
That number is EV, expected value: what a position is worth on average across every way the world can go, measured at a price you can get. It’s the first word of our name because it’s the only number a desk can be built on.
- 01At an entry, not at a midpoint. The edge is priced where your order would go in, on the side the model would take. That means posting your own price and waiting, or taking a price already on offer.
- 02After the fee that entry pays. Kalshi charges a fee whenever an order takes a price already on offer. An order that waits to be matched pays nothing. That holds on the great majority of Kalshi’s market series (its families of related markets), but not all. So the desk takes the fee from that exact series’ own schedule.
- 03Against the size that is waiting. Every row carries the size waiting at the current price, and orders are capped to it. An edge on size the order book can’t show you only exists on paper.
- 04Withheld when it can’t be defended. An idea that fails a single check doesn’t reach the board. The model would sooner say nothing than post a number it can’t stand behind.
Faster than
working the exchange screen.
Kalshi built a great marketplace. We built the cockpit that professional traders bolt onto it. It finds, prices, sizes and places the trade on Kalshi, with your own key. Every order goes out on your press and not before.
One-click ladder tickets
See the whole order book at once: every buy and sell order waiting on the market, and how many contracts sit at each price. Click a price and the ticket is set up for you, sized to what is really there and capped to your limits. Every row on the main board shows that size too, under BOOK.
Every order type, protected
You get every order type Kalshi offers, starting with limit and market. GTC stays open until you cancel it, with an optional expiry. IOC fills right away or cancels itself. FOK fills in full or not at all. Post-only waits in line and never pays the taker fee. Reduce-only is there for exits. One press places any of them on Kalshi. On top of that sits the Adverse-Fill Shield. Tell the desk the worst price you would accept. If the market starts moving toward it while your order waits, the desk warns you before you get filled on the wrong side. Cancel is one press, from the desk.
Fee-aware limit optimizer
Kalshi charges a fee when you take a price someone else posted. Post your own price and wait, and on most markets there is no fee at all. Before you place an order, the desk shows both routes side by side, with the fee for that exact market: take it now and pay the fee, or post and wait for free. You choose with the numbers in front of you.
The details that make it a desk.
One keystroke (Cmd-K) pulls up any market, from anywhere on the desk.
Every price level for an event on one screen, so the mispriced ones stand out.
Every trade is logged for you, with the result filled in when the market settles.
See when each position settles and your money frees up.
See how much you have riding on each theme and each event, against the limits you set.
Prices, profit-and-loss and available size refresh on every row, in real time.
The desk re-checks your positions against the live book every 6 seconds. It flags one the moment a price move eats into the edge your entry claimed.
Your key, your Kalshi account. We never hold your funds and never hold your key. Every order is signed in your browser with your own key and goes to Kalshi on your press.
Flip the halt switch. Nothing is sent until you resume, and Cancel keeps working. Your resting orders at Kalshi stay where they are until you cancel them.
Trade your own ideas,
professionally sized.
The desk isn't only for our signals. Search any market on Kalshi and bring your own thesis. The same models that price our board will check it. Then trade it with the machinery of a professional operation behind you.
- 01Conviction sizing. Bring any market and the desk sizes the order for you. It works out the Kelly-optimal size (Kelly is the classic bet-sizing formula) from your bankroll and your own probability. The size is capped to what is waiting in the market. You can edit it before you send.
- 02Kelly, desk-wide or off. Set your bankroll and Kelly fraction once and every default order size across the desk recomputes live. Switch Kelly off and the desk stops suggesting a size at all. The quantity opens empty and nothing refills it behind your back.
- 03Your rules, applied. Every guard ships with a sensible default and an on/off switch. That covers the Adverse-Fill Shield, exposure caps and discipline modes. The desk applies the rules you set.
- 04Run diligence on your own idea. Press Run Diligence on any market and EVDESK checks it the same way it checks the calls on the board. It uses our proprietary models, fitted on years of real market history. Alongside them, a frontier AI model reads Kalshi’s settlement clause word for word and works the sources. You get a probability, the edge after fees at a price the book can fill, and the reasoning behind it. Then you decide, and the desk sizes the order.
An AI analyst that shows its work.
Every 2.5 minutes the signal engine re-runs its numbers on every market it follows against the live quotes. It kills the junk with fixed rules and publishes what survives. An idea that fails a single check doesn’t reach the board. The engine hunts EV, expected value: the gap between a contract’s true value and what the market is charging for it. The probabilities come from models we fitted ourselves, on real history. What is underneath →
- 01A candidate shows its work. It quotes Kalshi’s own settlement clause word for word and names the source that decides it. It also shows the edge after fees, at a price the book can fill. Treat it as a hypothesis to check.
- 02EV after fees, at real size. The desk computes the edge after Kalshi fees and caps it to the size sitting in the book. That makes it a number you can actually get.
- 03Staleness protection. Every call comes with a limit: the most it says to pay. If the price moves against you by half the edge the call claimed, the desk locks the call. It also locks if the price moves by the full edge in either direction. It stays locked until the next model run confirms it. You can still trade the market yourself.
- 04Kill discipline, invisibly. The signal engine checks every row on every pass, before the board is built. It kills a market when the sportsbooks already price it the same way, when fees eat the whole edge, when the wording is ambiguous, or when it is an outright trap. Each check names the reason it fired.
- 05The written diligence. Press Run Diligence on any market and a frontier AI model writes up the case, working from our own model’s price. You get the argument in prose, the sources it read, and the wording that could go either way. It runs on Analyst credits.
The reasoning printed here is the model’s own, on the second row of the sample board.
Models built for one exchange,
and nothing else.
Most “AI trading” products are a general model asked a general question. Ours was built the other way round. Public frontier models sit inside our own proprietary layer of fifteen in-house engines. We fitted those engines over years to what happened on this one exchange.
The models are fitted on real history. The baseball record holds 37,027 games, each with its inning-by-inning line score and the umpire behind the plate. Underneath them sit 194,380 pitching and batting starts. The football record holds 7,276 completed games, with the rest days, roof and wind each one was played in.
Behind the weather markets sit decades of daily readings at the exact stations those markets settle against. The rain record reaches back to 1940. The temperature record is shallower. The table beside this counts the two separately, because they are not the same archive.
That took a long time to build, and it keeps building. Every model here earns its place. We test each one against what happened, again and again. We hold the data to the same standard as the models. When one temperature feed turned out not to be the data these markets settle on, we didn’t patch around it. We taught the model to refuse those days. The data has to be settlement grade, meaning the same numbers the market pays out on. If it isn’t, the model doesn’t get to make a call.
Games, 2010 to 2025
Per-start detail behind every pitching market, 2022 to 2025
Per-start detail behind every batter prop, 2022 to 2025
Completed games, 1999 to 2025. Each has its rest days and result, plus roof and wind where the record carries them. The 272 games already scheduled for 2026 carry everything but the score
Daily precipitation at the 18 stations these markets settle against. The deepest holds 31,439 daily readings, back to 1 January 1940. Coverage is uneven: nine stations cover ninety per cent or more of the days since, and four stopped recording decades ago
Daily highs and lows at 22 stations, each series with its own start date. The rain record is the deeper one. The longest high-temperature series here begins in 1970, and the longest low in 1990
Settled values, saved before the public window drops them
The hardest thing to teach a model is when to say nothing.
Ours says no far more often than it says yes, and every refusal comes with its reason. A model that only finds opportunities is a salesman.
The archive cannot be bought.
Kalshi’s public data keeps settled outcomes for a rolling window of roughly seventy days, and nothing older. There is no multi-year archive to buy, scrape or rebuild. Not for us, not for a competitor, and not for a funded team starting next year.
We capture that depth one day at a time, into an archive that is ours alone. The capture runs every day the service runs, and a day it misses can’t be recovered. The only way to earn depth is to have started, and we have. Our settlement archive has run continuously since May 2026. The daily price record behind it reaches back to October 2023.
Four lanes. Each with a reason
the market gets it wrong.
The signal engine doesn't wander the whole board hoping. Its proprietary lane models hunt where mispricing has a cause: a repeatable, checkable reason the crowd's price drifts from the evidence. Every card on the desk names its lane, its reasoning and its sources. A card expires the moment the price moves against it by half the edge it claimed.
An edge measured in cents survives only when execution costs less than the edge.
The scanner finds the edge.
The desk captures it.
The scanner and the desk were built as one system. Every call the scanner publishes is priced the way the desk prices your order: same fee math, same size caps, same rule for when a call goes stale.
Your on-call analyst, on every market.
The Analyst is a small window in the corner of your desk. It already knows the market you’re looking at: its rules, the live order book, and the model’s own work on it. It sees nothing from your account. Ask it anything. Or press Run Diligence on any market on Kalshi for the full report. You get the probability, the edge after fees, the sources, and the resolution clause that decides it.
- 01Context-aware. No copy-pasting tickers into a chatbot. It sees the market you’re on and answers in seconds, with that market’s numbers.
- 02Our models decide. Frontier AI does the reading and the research. Our own proprietary fitted models and diligence rules make the call. They are built on registries of the sources that settle each market, rules for facts that keep changing, and fee math checked against Kalshi’s published fee schedule. What is underneath →
- 03Watch, and it stays current. Switch Watch on for a market you have run diligence on, and the desk re-runs the diligence for you every day. It also re-runs it the moment the price moves against the call by half the edge it claimed. Each refresh is a diligence run, about 20 credits. It’s off until you turn it on.
- 04Fresh reports are free. If anyone ran Diligence on a market in the last 24 hours, you get that report instantly and use no credits, unless the price has moved enough to need a new one. Reports carry no names. Your own questions and answers stay private to you.
- 05Credits for the work done. A credit is a cent of Analyst work. A question uses about 6 credits and never more than 30. A full diligence run uses about 20 and never more than 60. A monthly allowance comes with the desk. Need more? Add 2,000 credits for $20, or switch on auto top-up with a monthly limit you set. How credits work →
Ask it about any market on the board, in plain words.
It answers from the exchange’s own settlement clause and from sources it can cite. When it can’t answer, it says so. It doesn’t invent figures or guess probabilities.
Off, the way it ships. It starts running, and spending, only when you switch it on.
Capped at 500 seats.
On purpose.
Prediction markets are thin. Each one can absorb only so much money before a signal stops being tradeable. An uncapped signal service destroys its own product with every seat it sells. So we cap ours. We’d rather turn subscribers away than dilute what they’re paying for.
- 01Set well below what the system can carry. On any given day EVDESK can serve 700 to 800 subscribers, depending on volume, the markets open and the trades available. We cap at 500 to leave a wide margin. Demand doesn’t raise it.
- 02Everything is real. Checkout enforces the cap. When seats are full, the seat list is a real queue. Every subscription is a FULL DESK seat, drawing from the same 500.
- 03What the cap counts. A seat is one subscription. Nothing else is rationed: the markets you trade in your own account and your use of the Analyst are not counted against it.
- 04Who else wants the same fill. Everyone acting on the same signal competes for the same fills. The 500-seat cap keeps that crowd small.
- What EVDESK can carry on any given day
- 700 to 800
- What EVDESK will ever sell
- 500
- Held back as margin
- 200 to 300
Daily capacity moves with volume, the markets open and the trades on the board. We cap well below the low end, so even a thin day has room for every seat.
Start free trial →Up and trading in minutes.
Take a seat
Pick a size, agree in one box, and your seat is yours. Your first three days are free. The desk is live the moment you log in, with the full signal feed and every tool. Start free trial →
Connect your Kalshi API key
Optional. The key never leaves your browser. It is stored on your own device and requests are signed locally, so only the signed request leaves. Unlinking wipes it. A key with trading permission lets the desk place, amend and cancel orders on Kalshi when you press. A read-only key only mirrors your balance and positions. We don’t hold your funds. They stay on Kalshi. Prefer to just read signals? Skip this step.
Set your risk, then trade
Set your bankroll and your Kelly fraction. The fraction is the dial for how aggressively each trade is sized. Every order size adapts. One press places the order on Kalshi. The models research and the desk prices. You decide, and you press.
One product in
two sizes.
FULL DESK is the whole thing: terminal, signal engine, and the Analyst. The two sizes differ only in monthly Analyst credits. It’s priced like a professional tool. For an active trader, the taker fees saved by posting instead of taking can cover the price. Run your numbers. Seats are open now.
FULL DESK
- Live quotes on any Kalshi market. Every order type priced and guarded, plus the Adverse-Fill Shield
- One of 500 seats on the full signal feed, with reasoning and sources on every call
- Fee-and-depth-aware EV board, with junk killed before it is built
- One click on an AI call sizes it by Kelly and prices it on the cheaper fee side. Your press places it on Kalshi, and the journal logs it
- Fee-aware limit optimizer, and signal entries guarded against going stale
- The Analyst, an automatic trade journal with CSV export, an exposure dashboard and a capital calendar
- 1,500 Analyst credits a month, about 250 questions or 75 diligence runsHow credits work →
FULL DESK+
- Everything in the standard FULL DESK
- 3,500 Analyst credits a month, about 580 questions or 175 diligence runsHow credits work →
- Put more markets on Watch without needing a top-up
- The only difference between the plans is how much research you can run. The desk itself is not tiered
Taker fees add up.
Posting skips them.
Kalshi charges a fee every time an order fills at once. An order that waits in the book pays nothing on most series, if it fills. The desk prices both routes on every ticket, so the cheaper one is one press.
Kalshi’s standard taker fee is 0.07 × contracts × price × (1 - price), rounded up per order, and some series charge less. A posted order pays no fee on most series, a quarter of it on some and half on combos. An illustration from the Kalshi fee schedule.
Taking all 25 orders costs $364 a month in fees.
Orders that wait in the book pay no fee on most series, but may not fill.
Fill now: $3.36 an order. Post and wait: $0 on most series, if it fills.
Ticket: 200 at 60c = $120 an order.
0.07 × 200 contracts × 0.60 × 0.40 = $3.36 a taken order. $3.36 × 25 a week × 4.33 weeks = $363.72 a month.
Fair questions get straight answers.
What does "EV" mean?
Expected value: the difference between a contract’s true value and what it costs. If a contract has a true 71% chance of paying $1 but trades at 60c, buying it is +11c of EV per contract. That gap is the edge the desk measures. It is an estimate, not a promise. Any single trade can lose. Our models estimate the true probability from the history we fitted them on. They subtract fees, cap the edge to the size waiting to trade, and show you the result. The estimate can be wrong. That is why each one comes with its evidence and expires the moment the market moves against it. The desk only recommends the side it expects to win, never the longshot. EV is also the first half of our name: EVDESK is the desk built to find it.
What is an Analyst credit?
One cent ($0.01) of Analyst work. Each action uses credits for the work it actually does. A question uses about 6, never more than 30. A full diligence run uses about 20, never more than 60. FULL DESK includes 1,500 credits a month and FULL DESK+ includes 3,500. Both refill on the 1st. If you run low, add 2,000 credits for $20, or switch on auto top-up with a monthly limit you pick. Nothing is charged beyond your plan unless you add credits or switch auto top-up on. Your questions stay private either way. The Analyst never uses one of the 500 seats. See the whole thing on one panel →
How does my own research stay current?
Switch on Watch for a market you have run diligence on, and the desk keeps that verdict fresh. It re-runs the diligence every day. It also re-runs it right away if the price moves against you by half the edge the verdict claimed. That way, a call you made on Tuesday isn’t quietly wrong by Friday. Each refresh is a diligence run: about 20 credits, never more than 60. Watch is off until you switch it on, one market at a time. The desk only spends the credits you asked it to.
Is the edge guaranteed?
No, and be suspicious of anyone who says otherwise. We publish theories of mispricing with the evidence behind them. Every call carries a freshness guard that blocks it the moment it goes stale. Markets adapt, edges decay, and any individual trade can lose. EVDESK is a research and desk tool. Every trading decision is yours, and no order goes out until you press Place order.
Does EVDESK place orders for me?
Only when you press. The order sheet on the desk is the ticket: side, price, size, type, time in force, cost and fee. Press Place order and the desk signs the request in your browser with your own Kalshi key. Then it sends that one order to Kalshi. Cancel and Modify work the same way, one press each. Nothing places orders on a schedule. The halt switch stops every send but leaves Cancel working.
Do you hold my money, or my key?
Neither. Your funds stay in your own Kalshi account. EVDESK never holds them and can’t move them. Your API key is stored only in your own browser, on your own device. It isn’t sent to our servers. Your browser signs each request, and only the signed request leaves your device. Our relay forwards it to Kalshi unchanged and does not keep it. We don’t see or hold the key, and unlinking wipes it from the browser.
Won't too many subscribers erase the edge?
That’s why FULL DESK is hard-capped at 500 seats. On any given day the system can carry 700 to 800 subscribers. The exact number depends on volume, the markets open and the trades available. We cap at 500 to keep a wide margin, and demand doesn’t move it. EV is always measured against the size waiting to trade right now. The signal engine scans thousands of markets and publishes only the few that pass every check, so the board is not one crowded name. When it’s full, it’s full.
Who regulates the markets I trade here?
Kalshi is a CFTC-regulated exchange, and EVDESK is analysis and trading software. It is not a broker, not a fund, and not individualized investment advice. You trade your own account at your own discretion. See the full disclosures below. Every subscriber agrees to them before billing begins.
Does the Analyst know a market's exact rules?
It reads the exchange's own settlement clause, word for word. Every market on the desk shows its full title and Kalshi's published rules text, and the Analyst works from that wording. A title is only shorthand. The clause is what pays. That’s why every verdict quotes the clause it hinges on. It also names the source that settles the market and flags any wording that could go either way.
What powers the AI?
Frontier AI (the Claude model family), wrapped in our proprietary diligence system. That system is the part we built, and the part that matters. It starts with our own models, fitted on real history. It keeps a registry of which source settles each market family. A dynamic-field doctrine (rules for facts that keep moving) catches the traps naive models fall into. Then there is the fee and liquidity math, and the discipline to kill a weak call. We won’t pretend we trained our own foundation model. Nobody selling you that is telling the truth either. The edge is the system around the model. Our models price the board on their own numbers. Press Run Diligence and the frontier model writes up the analysis in prose, working from those numbers.
What happens when the desk’s market index falls behind?
The desk says so, on the row it affects. The index re-checks every open market on a cycle the service measures and publishes: an hour’s pause plus however long the last pass took. The countdown on this page is that measured number. Live quotes for anything on your board refresh every six seconds straight from Kalshi. If a price is older than that, it’s labeled with its real age.
Can I see how much there is to trade?
On every row. The board has a BOOK column: the shares waiting at the price you would hit. It reads them straight from Kalshi’s order book, with no averaging or estimating. Every edge we publish is capped to the size waiting to trade. A +9c edge on 109 resting shares is not the same trade as +9c on 820. The board tells you which one you are looking at before you open the ticket.
Is the scanner a separate purchase?
No. One subscription covers both the scanner and the desk, because they were built as one workflow. Every signal is priced with fees already subtracted and capped to the size waiting. The desk prices your order the same way, in one click. Kelly sizing, fee-aware routing and a freshness guard are built in, and the order goes to Kalshi when you press. Trade the signals anywhere else and expect to hand a few cents of each edge back to fees and slippage (the price moving against you as your order fills).
What happens after the free trial?
Checkout is open. Your card is billed at the end of the 3-day trial unless you cancel. Canceling takes two clicks from your account: Cancel, then confirm. There are no emails and no retention games. Cancel mid-cycle and you keep access through the period you paid for. Start free trial →
Can I change or cancel an order after I send it?
Yes. Every resting order shows in Open Orders on the desk with Modify and Cancel, plus Cancel all. Modify changes the price or the size of the order in place on Kalshi. The halt switch stops every new order and leaves Cancel working, so you can always pull an order that is still waiting.
How is this different from odds screens or Discord picks?
Three ways. First, the terminal is a faster way to find, price and size a trade on Kalshi. Second, you can read the reasoning: every call carries its chain of evidence and a reading of the exact clause that settles it. Third, guardrails protect the trade after you find it: a freshness guard, fee-aware routing, and sizing capped to what is waiting in the book. A screen gives you prices. A pick asks for trust. We give you a desk.
The work is real,
and so is the seat cap.
Five hundred seats, deliberately under what the system can carry. The margin is the point.
Required disclosures. Read before subscribing.
No guaranteed profits; no guaranteed edge.
EVDESK identifies theories of potential mispricing in prediction-market contracts. These estimates may be wrong, may reflect no true edge at all, and past performance does not guarantee future results. Trading event contracts involves substantial risk of loss and is not suitable for everyone. Never trade with funds you cannot afford to lose.
Not investment advice.
EVDESK provides research, analytics, and trading-desk software for your own self-directed trading. Nothing on this platform is individualized investment, financial, legal, or tax advice, and no advisor-client relationship is created. All model outputs, probabilities, "EV" figures, and verdicts are automated analytical estimates, provided for informational purposes only.
You are the trader.
Every order is initiated by you, in your own exchange account, under your own exchange agreement. EVDESK is non-custodial, never holds funds, and takes no responsibility for trading outcomes, exchange availability, data errors, or execution results, except for fraud, willful misconduct, and the rights you hold under consumer-protection law, which section 22 of the Subscriber Agreement does not limit and neither does this. By subscribing you agree to the full Subscriber Agreement and Risk Disclosure, including binding acknowledgment of the above.
Honest figures.
Any performance-related figure shown anywhere on this platform is labeled, unedited, and never cherry-picked; hypothetical figures (such as "EV identified") are identified as such.