Five Pro surfaces, one terminal.
Each surface is a job a serious prediction-market trader already does by hand, and a reason that job is harder than it looks. Reading the same event across venues, where the price you see is shaped as much by fee schedules and settlement rules as by any real disagreement. Acting on a probability view with no reliable way to know whether your views are any good. Quoting both sides of a binary while informed flow waits to pick off a stale quote. Sizing a hedge against a contract that pays the same dollar whether the event grazes you or wipes you out. Judging, after a few hundred trades, whether what you have is skill or a story. Each one is a research project before it is a position. nijinn builds a dedicated surface for each, in a single terminal.
The trade is the gap net of friction.
The same FOMC question, the same election, the same weather contract trades on Polymarket and on Kalshi at the same time, and the 2 prices drift apart. Most of that drift has nothing to do with disagreement about the outcome. It comes from different fee schedules, different oracle wording, different settlement currency, and the cost of tying up capital until the thing resolves. So the spread you can see is almost never the spread you can keep.
Before any of it is a trade, you have to confirm the 2 contracts actually resolve on the same event, which is where naive title-matching blows up: 2 markets can read identically and still settle opposite ways on a wording technicality. Then you net the gap down by the real costs, taker fees on both legs, slippage at the size you intend to fill, the currency basis between the 2 venues, and the carry on being short a binary all the way to resolution. A 3-point gap can vanish once those come out.
And it does not sit still. A 3% net gap on a 7-day market is a different opportunity from the same 3% on a 90-day market, so the number that matters is the annualised return, not the raw spread. Windows close in minutes, and a gap that is widening on real news is one venue catching up, and trading into it is a loss. The arithmetic is trivial. Everything around it is the work. The Cross-Venue Arbitrage surface does that work continuously, and flags only the gaps that survive it.
Having a view is easy. Knowing it is a good one is the work.
Trading a market means betting that the crowd has the probability wrong and that you have it closer to right. The catch is that 'closer to right' is invisible in the moment. You only find out whether a 60% call was a good call across dozens of resolved markets, and only if you wrote down the 60% before the outcome was known. Skip that step and there is no record to grade, which means no honest answer to the one question that decides whether any of this is worth doing: are you getting better, or just getting lucky.
Even with a genuine edge, the ways to give it back are everywhere. Fees quietly eat a 2-cent edge on the thin ones. Size too large and a single wrong call takes a chunk out of the bankroll you cannot easily earn back. Size too small and the edge never compounds into anything. A whale can drop a position against you minutes after you enter and flip the trade negative. The same conviction that worked for you in politics may be worthless in crypto, and you would not know unless someone was scoring you category by category.
None of this is new to a disciplined trader. The problem is that the discipline only holds if the logging, the scoring and the sizing happen every single time, and by hand they almost never do. Overconfidence does not announce itself. It hides as a feeling until a number finally contradicts it. The Informed Directional surface makes that discipline automatic, so overconfidence shows up as a number in 30 trades, not 300.
A green book is not yet a profitable one.
Market making on a binary means quoting both sides and living off the spread plus the venue rebate, while 3 things try to take it back: inventory that piles up on one side, informed traders who pick off a stale quote, and a $0-or-$1 settlement that turns any leftover inventory into a coin flip. Unlike options, there is no underlying to hedge against, so the position has to be steered through the quotes themselves.
Even pricing the thing is not settled. A binary has at least 4 reasonable fair values, the mid, the depth-weighted mid, the microprice, and the recent trade-flow VWAP, and they agree right up until the book thins out or one side starts loading a directional view, which is exactly the moment you need them to agree. Quote off the wrong one and you are the liquidity everyone else is happy to take.
The deeper trap is that the book can look green while the rebate is the only thing holding it up. Spread capture, rebate, carry on inventory and the cost of being picked off are 4 different stories, and added together they hide which one is paying you. Meanwhile the flow can turn against you, your inventory loading on the wrong side seconds before a 40-point move, and on a contract that settles to a dollar or to nothing, the fill you did not need is the one you remember. The Market Making surface keeps those 4 stories separate and watches the flow turn, so you see the trap forming while you can still pull the quote.
Am I good at this. Can I get out at fair value.
A serious prediction-market trader is running a small business: capital in, risk on, money out over time. Every business gets a periodic health check. This one has nowhere to get one. The audit stack an equity desk takes for granted does not exist for prediction markets, where nothing holds calibration, risk and exit cost in one place, so the questions that decide whether to keep going go unanswered.
It comes down to 3 questions. Am I good, meaning is my read better than the crowd's, or did a few lucky resolutions flatter me. Am I safe, meaning how deep can the next losing streak go and how long to climb back. And can I get out, meaning if I had to close every open position today, what the exit would cost. The first question is the hardest, because it only has an answer if you wrote down your probability before the market resolved. Without that, a win and a good decision look identical after the fact.
These are not the same job. They run on different data, on different clocks, and the tools that exist handle them in pieces: a P&L tracker here, a wallet analyser there, none of them scoring the estimate you made, none of them walking the book to tell you what it costs to leave. Edge does not carry across either. A real knack for politics can sit next to quiet losses in crypto, and a single blended number buries both. Unmeasured skill is just a story you tell yourself after resolution. The Performance Audit surface is the one place that answers all 3, scored against the calls you logged before they resolved.
Sizing is the whole problem.
For a lot of real exposures there is no clean instrument except a prediction market: a tariff decision, a policy path the rates curve does not span, a weather event outside the standard indices, a geopolitical outcome with a date attached. If you carry that risk, the contract is often the only place to lay it off. The trouble starts with sizing: what notional neutralises the exposure, given everything about the contract that does not behave like insurance.
4 features of the contract get in the way, and a prediction market hides none of them well. The payoff is binary, so a 5% tariff and a 50% tariff both pay the same dollar, so one overpays the hedge and the other leaves you short. The contract's resolution wording may not match your actual trigger, which is the real risk here, and 2 clauses can read the same and settle opposite ways. The full premium is locked up until settlement, not paid in instalments like a policy. And a large position in a thin market moves the price against itself as you build it.
On top of that, the person carrying the exposure is usually not a trader. They understand their own risk as dollars at stake, while the market talks back in contracts at 30-odd cents each. So the sizing gets done roughly, or by feel, and a hedge that is 30% too small or 30% too large is a quiet, expensive mistake that only shows up at settlement. The answer that matters is a dollar figure, not a coefficient. The Real-World Hedging surface starts from the dollars at risk and returns the size that offsets them.