I just built my own read-only collector and scanned all 14 MLB moneyline markets on Polymarket and Kalshi. All 14/14 matched—the dates, the teams, and exactly which game in each case all correspond perfectly.

The mid prices between the two platform’s corresponding results differ by an absolute amount of 0 to 2.0 percentage points. If you buy the opposite sides on the two platforms, the lowest all-in cost before fees is only 1.00—so there’s still no profit.
After factoring in both sides’ costs using the publicly disclosed taker fee estimates, the best net edge is around -2.92%. After checking all 14 games, there is no simple executable arbitrage opportunity.
Plain cross-platform arbitrage depends on the two sides showing a large enough, executable deviation that still remains positive after fees—not just prices being close.
In actual trading, the event identity must match completely; you must be able to truly execute at those prices. Fees can’t wipe out the profit. The order book depth must be sufficient for you to fill both legs, and the price gap must persist for a while rather than disappearing.
This time I only took a snapshot of data at a single point in time. That doesn’t mean extreme mispricing windows won’t appear later, but at least I personally verified it: whether an opportunity with a spread can actually work after deducting fees and accounting for execution conditions.
If you want to check these markets yourself on Polymarket, you can use this link：