A plain-English breakdown of the institutional math powering our market making engine — and why it works where speed-based strategies fail.
Most retail traders try to guess which way markets will go. We take the other approach — we sit in the middle and collect the spread from both sides.
Not all prediction markets are worth trading. We filter for markets where retail traders — not institutional quants — are the primary participants.
Our scanner looks for: volume between $2K-$150K (too small for hedge funds), prices between 10¢-90¢ (maximum uncertainty = maximum spread opportunity), and bid-ask spreads wider than 4¢ (enough room to profit after fees).
Once we select a market, we place two orders simultaneously: one to buy YES at a low price, and one to sell YES at a higher price. The gap between them is our profit.
The Avellaneda-Stoikov framework (used by institutional market makers since 2008) calculates exactly where to set these prices based on our current inventory, market volatility, and time remaining to resolution.
The biggest risk for market makers is "adverse selection" — when someone with inside information trades against us before we can react.
VPIN (Volume-synchronized Probability of Informed Trading) measures the imbalance between buy and sell orders. When it spikes, informed traders are present. We automatically widen spreads or cancel orders entirely, protecting your capital.
For the quant-curious. Plain English explanations of the institutional mathematics powering our engine.
What it means: When we hold too much inventory on one side, we skew our prices to attract the other side. If we hold lots of YES contracts, we lower our YES ask price slightly to sell them faster, while keeping our bid lower too. This prevents dangerous one-sided positions from building up.
What it means: Two sources of profit in one formula. The first term compensates us for the risk of holding inventory. The second term is pure liquidity provision profit — the spread we earn just for being there to trade against. Wider spreads = more profit per trade but fewer fills. We tune γ to find the sweet spot.
What it means: We never bet a fixed dollar amount. Kelly tells us the mathematically optimal fraction of bankroll for each trade given our estimated edge. We then apply a 75% haircut (fractional Kelly) to account for uncertainty in our own estimates. This maximizes long-run growth while preventing ruin.
Everything you need to know before getting started.