> For the complete documentation index, see [llms.txt](https://predictex.gitbook.io/docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://predictex.gitbook.io/docs/risk-management.md).

# Risk Management

#### Offering leverage on prediction markets comes with risk

A prediction or sports-betting contract is a bet that settles to one of two prices. While the event is live, the contract trades somewhere between 0¢ and 100¢, and that price can be read as the market's estimate of how likely the outcome is. When the event ends, the price snaps to exactly 100¢ (it happened) or exactly 0¢ (it didn't). Prices can also jump mid-event whenever new information arrives — a team scores, a player gets injured — rather than drifting smoothly.

These jumps are what make leverage hard. In ordinary futures or perpetual markets, prices move in small continuous steps, so an exchange can close out a losing trader before their loss exceeds the margin they posted. A binary contract can leap straight past that margin in a single move, leaving the exchange on the hook for the winnings owed to the other side.

A quick illustration. Suppose the Lakers (home) are at 60% to win and the Knicks (away) at 40%. Both traders use 10× leverage:

* Trader A goes long 1 Lakers contract (pays $1 if the Lakers win). He posts 6¢ margin — 10% of the contract's 60¢ value.
* Trader B goes short 1 contract (effectively backing the Knicks). He posts 4¢ margin — 10% of the 40¢ value on his side.

If the Lakers suddenly jump to \~100% and win, Trader A is owed 40¢, but Trader B only posted 4¢ — the exchange must cover the 36¢ gap. If instead the Lakers collapse to \~0% and lose, Trader B is owed 60¢ while Trader A only posted 6¢ — a 54¢ gap. Either way, the exchange comes up short.

So you can't offer leverage on binary contracts just by lowering collateral and applying normal liquidation rules — the terminal jump guarantees the exchange ends up under-collateralized or exposed to arbitrage. This isn't an edge case; it follows directly from how binary payoffs work. PredictEX solves it with three tools working together: (a) platform stability adjustments, (b) performance adjustments, and (c) a form of ADL. Together they make leveraged betting possible.

#### Platform stability adjustment

The platform stability adjustment is what makes leverage possible. It's calculated on your notional position size and compensates the exchange for the leverage risk described above. It's dynamic: the amount depends on the contract's current price, your selected leverage (Boost), the score and time left to play, the league, and similar factors, all modeled and calibrated on large historical datasets.

The intuition: the more room there is for the price to move gradually — so a losing position can be closed (Knockout) before any sudden jump — the smaller the adjustment. The closer the contract is to a discontinuous resolution, the larger it gets. That's why an identical bet can cost far more with seconds left on the clock than it does hours before tip-off.

In the current flow, the adjustment is charged upfront and baked into the Max Payout shown before you bet — it is not listed as a separate fee. Because it's taken at entry, a leveraged position's cash-out value can start below the amount you put in (your Bet / "You Risk"). It only applies to leveraged bets (Boost greater than 1×).

**Two simplified examples show how this plays out.**

**Example 1** — well before tip-off. Toronto Raptors vs Golden State Warriors hasn't started; odds are 42% / 58%. A trader goes 10× long GSW:

* Posts $100 as the cost basis (Bet / You Risk)
* Opens 1,729 contracts at 0.5776 — a $1,000 notional position
* A $1.7 stability adjustment is baked into the Max Payout and current Cash-Out Value

If GSW wins and the price never falls to his Knockout level (0.5299), he gets his $100 margin back plus net winnings of 1,729 × (1 − 0.5776) − 1.7  = $728.6. On a plain spot market that same $100 would buy only 173 contracts, for at most 173 × (1 − 0.5776) = $73 of upside.

**Example 2** — one minute left, same 42% / 58% odds. Now a 10× long on GSW carries a much larger adjustment, because there's almost no time for the price to drift and trigger a Knockout before the game resolves in a single jump:

* Posts $100 margin
* Opens 1,729 contracts at 0.5776 — a $1,000 notional position
* A $380 stability adjustment is baked into the Max Payout and current Cash-Out Value

If GSW wins (and the price never hits his Knockout level), he gets his margin back plus net winnings of 1,729 × (1 − 0.5776) − 380 = $350. That's roughly what he'd make putting the same total cash to work on spot: $100 + $380 = $480 buys \~830 contracts at 0.5776, worth 830 × (1 − 0.5776) = $350 if GSW wins.

#### Performance adjustments

Performance Adjustments are the second lever that enables leveraged trading on PredictEX. They are charged on the gains of winning traders, are dynamic, and are baked into the Max Payout and Cash-Out Value.