Live strategy authoring, explained

The claim

Minmax frames sports and prediction-market betting as currently split between two modes:

• Pre-game — research the event for days or weeks, place a position before kickoff, wait for resolution. Static, considered, slow.
• In-play — react clickbyclick to events as they unfold. Reactive, emotional, fast.

Minmax adds a third:

• Live strategy authoring — watch the market, build or adjust a quantitative strategy in real time as conditions change (a goal scored, an injury announced, the tempo of play shifting), and an agent executes the strategy against the live order book on your behalf.

The framing is: closer to running a portfolio than placing a bet. You're not predicting an outcome. You're managing exposure to a live, evolving market.

How it works concretely

The user starts with a baseline strategy authored before the market opens — say, a fixed probability threshold for entering positions on Polymarket BTC 5-minute contracts. Once the market is live, the user can push rule changes to the running agent without redeploying or restarting:

• "Increase exposure to the underdog if the spread crosses 75 basis points"
• "Halve position sizes if 5-minute realised volatility exceeds 2x rolling average"
• "Stop trading entirely if drawdown exceeds 15% of bankroll"

The agent adopts the new rule on the next decision tick. No downtime, no missed window. Polymarket trades fire from a per-user Safe wallet, gasless, with 210ms signal-to-fill. The rule set is the authoring artifact; the agent is just the executor.

Is this actually a new category?

Honest answer: no, but it's an unusually clean implementation of an existing concept.

Discretionary algo trading — where a human authors and updates a rule set in real time while a system executes against it — has been around in equities for decades. Two Sigma, Renaissance, and every other quant fund's research desk has been doing some version of this since the 1990s. What's new about Minmax's framing for Polymarket and sports markets is mostly the accessibility: a no-code authoring UX, gasless execution, and a data context (50B rows of order-book history) that makes the rules more meaningful than they'd be in a thin-data market.

Calling it a "third mode" is marketing. The category exists in institutional trading; what Minmax claims is to be the first commercial implementation that brings it to retail Polymarket and (via the NDA'd sportsbook partner) sports traders. Reasonable claim. Worth tracking.

Where it should work

Sports markets. A football match has 90 minutes of continuously evolving information — goals, injuries, substitutions, momentum shifts. Neither pre-game research nor click-by-click reaction works for the trader trying to capture the value of the match as a portfolio. Live strategy authoring is the natural format. This is also the highest-stakes part of Minmax's roadmap because sports markets are larger by orders of magnitude than Polymarket's current event surface.

High-event-frequency prediction markets. Polymarket BTC 5-minute markets, Kalshi short-window contracts, election-night moment-by-moment repricings. Anywhere the rule set needs to update faster than a human can manually retype the order.

Where it won't help

Slow, low-event markets. An election resolving in three weeks, a regulatory decision resolving in two months. Nothing's happening minute-to-minute that requires a live rule update. A static strategy authored once at market open does the same job at lower complexity.

Thin liquidity markets. Live authoring requires an agent that can fill orders quickly enough that your rule updates translate to executed trades. If the market doesn't have depth, the rules don't matter — you can't get filled at the price your rule set assumes.

Markets without recorded history. The rules are only useful if you can backtest them against past order-book states. Without depth of data, you're authoring rules blind.

What Minmax has actually shipped vs roadmap

As of 2026-05-25:

• Shipped (closed beta): Polymarket live strategy authoring via the Strategy Builder, with 26+ drag-and-drop blocks plus rule logic. Per-bot Safe wallets, gasless execution, hot-reload of rule updates without restart. Backtester with 20,000+ historical windows.
• Roadmap: Kalshi (data recorder running ahead of execution support), sports markets (depends on the NDA'd regulated sportsbook partner integration completing), event derivatives (later).

The Polymarket implementation is the proof. The sports rollout is where the category claim either pays off or doesn't.

The honest take

Live strategy authoring is real, not entirely new, and best understood as "discretionary algo, made accessible." Minmax's Polymarket implementation works today; the sports leg — where the claim either becomes a category-defining product or fades into the existing discretionary-algo universe — is roadmap-dependent. The 50-billion-row order-book archive underneath is the unique-and-defensible piece. Live authoring is the user-facing interface to that data. Trust the data; verify the sports rollout when it ships.