LIL222 doesn’t look like the wallets we usually report on. They don’t scalp. They don’t flash thousands of orders a day. They sit on a Polymarket account, take maybe four positions in a week, and walk away with returns that put nine out of ten quant funds to shame.
This report is the result of three weeks following the wallet on-chain, mapping every fill, and trying to understand - with the resolution of the markets they traded in hand - what they actually saw. Some of it is reproducible. Some of it isn’t. Both are interesting.
The portfolio shape
If you sort LIL222’s twelve-month history by P&L contribution, the top six positions account for 78% of net profit. That isn’t a tail; that’s a strategy. They are explicitly hunting for high-conviction directional asymmetry, sizing into it, and accepting that everything else they touch in between is noise.
That asymmetric profile shows up in the trade-level chart in section 002. We highlighted the top six in blue.
Where the edge appears to come from
Reading wallet activity blind, two patterns emerge. The first is a strong preference for markets where the resolution criteria are unusually clean. LIL222 isn’t betting on whether a politician’s mood will improve; they’re betting on whether a number prints above a specific threshold by a specific date.
The second is a positioning rhythm that lines up almost too well with announcement-driven volatility. Across the top six, the wallet entered between four and twelve hours before a scheduled data release or court ruling and exited within ninety minutes of resolution. That isn’t edge in pricing. That’s edge in structuring participation.
What you can copy
Three things from this wallet are immediately portable to a bot:
1. The clean-criteria filter. Before a market is even considered for sizing, it has to pass a resolution-clarity check. We coded a small heuristic version of this in the mean-reversion guide - word counts on the resolution text, presence of explicit numeric thresholds, and a blacklist for subjective phrasing.
2. The pre-event positioning window. You can pull the same announcement calendars LIL222 appears to be reading from public sources. The Discord #calendar channel maintains a member-curated version with API access for paying members.
3. The exit clock. The wallet almost never sits in a position past resolution-minus-30-minutes. That’s a hard rule, and it’s the easiest part of the strategy to encode - a single timestamp comparison, gates on every order placement.
What you probably can’t copy
The sizing. We tried.
LIL222 sizes on a curve that doesn’t map cleanly to volatility, Kelly, or implied probability. After a lot of squinting at the wallet, our best guess is that the sizing reflects an off-platform conviction model - either a private set of priors or a private model the trader is reading. We have no way to reconstruct it from on-chain data alone, and the linear approximations we tried gave back ~40% of the wallet’s twelve-month return.
That gap - the part you can’t copy - is what makes LIL222 a good case study rather than a strategy you can clone wholesale. The reproducible parts still give you a reasonable bot. The non-reproducible parts tell you what to keep working on.
