A list of contracts is not automatically a diversified portfolio. Two markets can have different titles while both depend on the same announcement, game or price move. If that shared event breaks against your thesis, several positions may lose together. The practical question is not “How many rows do I hold?” but “Which distinct facts must be true for these positions to pay?”

Map the common driver

Imagine illustrative contracts asking whether an asset exceeds a first threshold by Friday, exceeds a higher threshold by Friday and records a large weekly rise. Their payoffs differ, but one adverse move could hurt all three. They are not three independent tests of a model’s forecasting skill. Similarly, a team-to-win market and a player-award market may share dependence on the same final game. Correlation need not be perfect to matter.

For each contemplated position, write its market URL, selected side, expiry, resolution source and a one-line causal driver. Group rows by that driver before adding exposure. Then sketch two scenarios: an outcome that helps all positions, and an outcome that harms all. If the losing scenario is essentially the same event repeated, set one combined exposure budget rather than treating each row as a fresh limit.

The rules still differ

Overlapping events are not identical contracts. Cutoff times, wording, data sources and exceptional conditions can separate two seemingly equivalent bets. Polymarket says the full rules define resolution; read both sets before assuming the outcomes are logically linked. If one contract requires an official announcement and another requires a completed event, do not net them as an exact hedge.

Likewise, a quick difference between two displayed percentages is not necessarily tradeable. The figures may be midpoint or last-trade displays rather than executable bids and asks; order-book depth can be insufficient. Check how displayed prices are calculated and the limits of executable size. A hypothetical logical relationship can remain difficult or costly to express.

Record concentration explicitly

A useful journal has one row per contract and an additional shared-driver field. Sum the maximum loss across positions tied to a driver as a conservative gross-exposure upper bound, before reliable offsets—not a forecast that every leg loses together. Note any offsets that rely on a particular resolution interpretation. Review any other holdings affected by the same macro event, even if they sit in different categories. Diversification can reduce some concentration risks but cannot eliminate loss.

No universal count of markets makes a strategy safe or profitable. The right answer is a risk budget informed by common causes, contract wording, liquidity and uncertainty in your estimate. A model gap in five dependent markets can represent one hypothesis expressed five ways, not five independent advantages.