Mid-band mega prints: conviction money at coin-flip prices
Across 112,000 resolved whale bets, six-figure buys lose at nearly every price — except between 45¢ and 70¢, where they ran +23%. The prints stopped for a month, then came back: six markets, six wins so far.
The hypothesisWhale size alone is noise, but size × price might not be. A full grid of resolved large bets by entry price and dollar size shows almost every cell red — favorites at every size, longshots at every size — except one: $100k+ buys at 45–70¢. Nobody parks six figures on a coin flip for fun; mid-band mega prints are conviction, not hedging or market-making flow. Copy exactly that cell.
The test
The grid came first: every resolved whale buy, honestly scored, bucketed by price band and dollar size. The 45–70¢ × mega-size cell was the only one positive in consecutive months, and cross-checks at our own fills agreed (copies triggered by $100k+ mid-band prints went 6-for-6; big triggers below the band went 0-for-4). A live probe then watches for new prints in the cell, logs the executable ask within seconds, and settles at resolution — with a pre-registered concern written into the code: the cell was found by slicing a grid, so multiple-comparisons risk is real and only forward performance counts.
The result
The market stopped producing the input for a while: in-band mega prints had clustered in a major sports-tournament month, and after it ended every six-figure print arrived in the ≥70¢ favorites band the grid measured negative. Then they came back. Since mid-August the probe has logged 10 qualifying prints on 6 markets, and all 6 markets won — +67% ROI at the logged ask, at an average price of 59¢ (3 markets in August, 1 in September, 2 in October).
At the prices paid, six straight wins would happen by chance about one time in twenty-three. That is encouraging and nowhere near proof: six results cannot tell an edge from a good streak, and the cell was picked out of a grid, which makes a lucky start more likely, not less.
Conclusion
Open. The tripwire is doing its job: it costs one database query every ten minutes and catches the prints when they come. The verdict waits for 30 markets, which at the current pace is months away.