Daily realised PnL
After execution fees · UTC closing dates
Latest 14 dates with closes. Open positions are excluded. Missing dates are not treated as zero.
Paper performance
Win rate needs a sample size
One dot per wallet · Select a dot to inspect its evidence
Positive realised PnLZero / negative PnL
Fully closed games, not repeated fills. High win rate alone does not establish skill or positive expectancy.
01 / Compare the evidence
Wallets worth understanding
These are simulated follower results for a fixed, selected cohort. They are not the source wallets’ profits or an insider ranking.
| Wallet / sample | Paper PnL | Game wins | ROI |
|---|
02 / Know what is missing
Collection & paper execution
A fresh page can still contain an incomplete sample. Coverage and execution quality stay visible alongside the results.
Observed wallets by sport
Overlapping wallets across sports. Rolling 7-day sample, capped at 100,000 trades; no market-wide coverage percentage is known.
From observed BUY to paper fill
Includes ineligible and non-sports observations
What would strengthen the study?
- Let the frozen cohort accumulate complete weeks and independent games, including losing wallets.
- Record future closing-price comparisons and source exposure before judging repeatable edge. These are not measured here yet.
- Keep newly discovered wallets in a separate prospective cohort; never backfill hypothetical historical wins.
03 / Observations → research → content
Content leads, with receipts
Latest observed alerts, not recommendations. Check the original market, full selections and timestamps before writing a story.
Can we identify an insider wallet?
Public trades can reveal unusual behaviour. They cannot establish who knew what, or when. No insider probability is assigned here.
- Observe
Record size relative to a wallet’s own baseline, market and selection, source time, and first observation.
- Challenge
Check both sides, public news timing and later price movement. Hedging, copying and chance can look informed.
- Follow prospectively
Test repeated, fee-adjusted results on new events. Keep the hypothesis separate from a claim about the person.