Realized edge +2.3 probability points across 451 scored resolved positions, 95% CI [-1.0, +5.3]. These figures describe the recent-trade window this read covered, not the wallet's whole resolved record, so they cannot be read as an edge either way. A Convexly read of the public record.
Diagnostics on public behavior, not investment advice. Demonstrated past skill is not a forecast. The figures describe the window this read covered, which omits redeemed positions and is not the full resolved record. This caveat travels with the badge in every view.
A methodology-cited read of this wallet's public Polymarket record, taken from the venue's recent-trade window. The statistics below describe that window, not the wallet's whole resolved record.
Sample quality: StrongLarge scored resolved sample. The rating is still historical and cross-sectional, but the noise floor is lower.
Diagnostics on public Polymarket behavior, not a trade-following recommendation or a per-wallet forward forecast. Not investment advice.
Data updated 2026-08-21
Read coverage
Covers the most recent 3,500 trades the public API returns for this wallet. Older history beyond that window is not included in this read.
Frozen research snapshot
On the frozen 2026-04-25 research snapshot (sub-daily crypto up/down micro-markets excluded), this wallet had 151 resolved positions with a win rate of 45.7% and a realized edge over entry prices of +4.2 probability points, 95% CI [-0.9, +9.9] (BCa bootstrap); the 95% interval includes zero, so on that snapshot the edge is not separable from chance.
Did not clear the Benjamini-Hochberg false-discovery screen at q = 0.1 on that snapshot (178 of 3871 wallets cleared). That means no corrected evidence of skill on the snapshot, which is not the same as evidence of no skill. In-sample separation on a frozen snapshot, not validated forward skill.
The frozen snapshot does not carry the single-event concentration screen that the live analyzer applies; the realized-edge read above (when present) is the concentration-screened read.
Brier gap (baseline minus Brier): +0.1122. Descriptive only: no interval is computed for this gap on this surface. The interval-backed skill read on this page is the realized-edge block. Brier is scored lower-is-better against resolved outcomes; the baseline is a constant base-rate forecast on the same positions.
Biggest event: Russia x Ukraine any diplomatic meeting by...? (54.9% of the net result; shares above 100% mean the remaining events netted negative)
Category breakdown
Descriptive per-category reads; none is an FDR-cleared skill claim unless labeled. Categories with fewer than 3 resolved positions are excluded by the analyzer; small per-category samples are noisier than the headline sample.
Category
Resolved
Win rate
Brier
Baseline Brier
other
165
52.7%
0.138
0.249
politics
152
55.9%
0.117
0.246
entertainment
95
53.7%
0.194
0.249
sports
24
25.0%
0.028
0.188
crypto
8
50.0%
0.060
0.250
science
7
57.1%
0.241
0.245
Resolved-position history by entry price
Realized win rate per entry-price band versus the band's implied probability. Bands with zero resolved positions are omitted; thin bands are noisy.
Entry-price band
Resolved
Realized win rate
Band implied
0-10%
54
1.8%
5.0%
10-20%
45
15.6%
15.0%
20-30%
41
26.8%
25.0%
30-40%
33
18.2%
35.0%
40-50%
36
33.3%
45.0%
50-60%
38
65.8%
55.0%
60-70%
52
75.0%
65.0%
70-80%
59
81.4%
75.0%
80-90%
61
91.8%
85.0%
90-100%
32
100.0%
95.0%
Frequently asked questions
What does the resolved Polymarket record show for wallet 0x38f5...6824?
As of 2026-08-21, Convexly's cached read shows 451 resolved positions, 237 wins and 214 losses (win rate 52.5%). Realized edge over entry prices: +2.3 probability points, 95% CI [-1.0, +5.3] (BCa bootstrap, n = 451 scored positions). These are descriptive statistics of the past record, retrospective and in-sample, not a forecast.
Is wallet 0x38f5...6824 skilled or just lucky on Polymarket?
Realized edge +2.3 probability points across 451 scored resolved positions, 95% CI [-1.0, +5.3]. These figures describe the recent-trade window this read covered, not the wallet's whole resolved record, so they cannot be read as an edge either way. A Convexly read of the public record. Demonstrated past skill is not a forecast, and this is not investment advice.
Is this page a forecast or a recommendation to trade like this wallet?
No. Convexly publishes diagnostics on public Polymarket behavior. Demonstrated past skill is not a forecast, the statistics on this page are retrospective and in-sample, and nothing here is investment advice or a recommendation to mirror this wallet's positions.
Want this for every wallet you watch?
Convexly Researcher tier ($49/mo) follows up to 10 wallets, ships a Monday digest of Edge Score deltas + concentration drift + new positions, plus a daily Coherence Signals feed.
Convexly