36 accounts entered the top 100 this month; 35 dropped out. When the S&P 500 rose, FinTwit hit 43%; when it fell, 33%, a gap that looks more like market beta than skill.
What moved this month
Communication Services accuracy▼−24.6 ptsIndustrials accuracy▼−15.3 pts
01The Box Score30-Day Window
FinTwit Edge (30d)
—
How far the typical call finished from the market. Below zero, it lagged.
Beat S&P 500
—
Share of calls that actually beat the index.
Coin-flip margin
-7.4 pts▼−4.2 pts
How much the median account tops a 50/50 guess.
Bullish skew
—
Nearly every call is a buy. The crowd rarely calls a top.
Ranked field
226▲+8
Accounts with enough of a record to grade.
Top decile50.8%
Median42.6%
Bottom decile35.1%
A “decile” is just a tenth of the field: the best tenth, the middle, and the worst.
Crossing the wire: the names the crowd talked about most, and how often calls on them landed.
Volume state
Where the tape got loud
Resolved 30d tape
Resolved calls
—
Calls with a 30-day outcome in this edition.
Ranked field
226
Accounts with enough history to grade.
Buy pressure
—
Share of calls that leaned bullish.
Peak crowd
NVDA
264 distinct callers
Most crowded tickers
NVDA
264 callers
MU
241 callers
AMD
196 callers
NBIS
180 callers
03The CrowdConsensus → outcome
A call with a single account behind it wins 39% of the time, barely better than a coin toss. When 30+ separate accounts land on the same call inside a week, the win rate climbs to 30%. That cuts against the usual instinct to fade a crowded trade: when independent voices reach the same name and direction on their own, the agreement carries information, and here it is worth more than any single loud take.
The Agreement Curve
30d win % by conviction
Coin flip · 50%
39%
1 (solo) agreewin39.0%calls6,888med ret-1.2%
35%
2–3 agreewin34.6%calls7,028med ret-3.6%
37%
4–6 agreewin36.7%calls7,439med ret-3.4%
37%
7–15 agreewin37.0%calls12,918med ret-4.1%
38%
16–30 agreewin37.6%calls13,824med ret-1.9%
30%
30+ agreewin30.4%calls32,537med ret-5.6%
1 (solo)
2–3
4–6
7–15
16–30
30+
Win rate by number of agreeing accounts: 1 (solo) agree, 39.0% over 6888 calls; 2–3 agree, 34.6% over 7028 calls; 4–6 agree, 36.7% over 7439 calls; 7–15 agree, 37.0% over 12918 calls; 16–30 agree, 37.6% over 13824 calls; 30+ agree, 30.4% over 32537 calls.
Calls grouped by how many distinct tracked accounts made the same call (ticker + direction) in the same week. More agreement lines up with higher accuracy, not lower.
Crowded trades that paid
Each of these names drew twenty or more callers and still cleared a 60% win rate, the rare case where a crowded trade was also a correct one. When the consensus formed around these tickers, it formed for a reason, and the tape paid it out.
Cohort
n
Win %
Med. ret
AMD
196
68.9%
+6.5%
HOOD
115
65.4%
+11.5%
SNOW
94
82.1%
+5.2%
RDDT
73
71.4%
+8.3%
CRWD
71
64.1%
+8.5%
Most-piled-into tickers (≥20 distinct callers, last 90d) whose 30d win rate cleared 60%. The 'n' column is distinct callers.
Crowded trades that didn't pay
The flip side: the names just as many people piled into, where the win rate came in below a coin flip. Heavy attention pulls in momentum-chasers and latecomers as readily as it marks a real edge, and on these tickers the crowd mistook noise for conviction.
Cohort
n
Win %
Med. ret
NVDA
264
15.9%
-5.0%
MU
241
49.3%
+1.1%
NBIS
180
46.5%
-5.0%
GOOG
178
10.3%
-4.1%
META
175
37.8%
-3.2%
Most-piled-into tickers (≥20 distinct callers, last 90d) whose 30d win rate came in below a coin flip (under 50%). The 'n' column is distinct callers. The resolved window (Mar-May 2026) spans an April drawdown.
04Pros vs JoesWhen the specialists fight the crowd
When a ticker’s specialists took the opposite side from the crowd, they were right more often, winning 97 of 123 weekly standoffs. A deep read on one name beat a broad take on all of them.
Pros
Score ≥ 70 on the ticker
97
standoffs won
♕
VS
Joes
The crowd · everyone else
26
standoffs won
PROS 79%
21% JOES
The specialists win 79% when they break from the crowd.▲+14.0 pts
verdict at 7 days · 123 standoffs
Standoffs by ticker
Ticker
Standoffs
Pros won
Joes won
ETH
7
6
1
TSLA
6
4
2
BTC
4
4
0
NIO
4
3
1
ORCL
4
3
1
SPCX
4
3
1
ASTS
3
3
0
SOL
3
3
0
SPY
3
3
0
+ 64 more tickers
85
65
20
Total
123
97
26
A standoff is one week a ticker’s specialists (per-ticker score ≥ 70 that day) and the crowd took opposite sides; the winner is whichever side’s calls returned more over the next 7 days. These are aggregate figures, with no individual names. Newer standoffs land in the next edition as they resolve.
05The AnatomyWhere it works
A single headline win rate hides as much as it reveals. The same record looks very different once it is cut by how long a call is held, whether it is a buy or a sell, the size and sector of the company, and the size of the account making it. Slice it those ways and the real strengths start to separate from the blind spots.
By audience size
Mega · 500k+ accounts grade 43%; Small · <5k accounts, 40%. A bigger audience does not buy a better record. Reach rewards confidence and posting volume more than it rewards being right, so the loudest accounts are not reliably the sharpest. Follower count measures how many people are listening, and almost nothing about whether they should be.
Cohort
n
Win %
Mega · 500k+
15
42.9%
Large · 50–500k
100
43.4%
Mid · 5–50k
87
42.4%
Small · <5k
24
39.9%
Win % column here is each band's median caller accuracy.
By sector
Healthcare leads at 52%; Utilities trails at 23%. The crowd knows some terrain and guesses at the rest. Accuracy clusters in the sectors people actually live in, where the average poster has a feel for the products and the headlines. The harder corners reward specialist knowledge most of the field does not have, and the win rate shows it.
Win rate by sector, ranked: Healthcare, 51.9% over 4788 calls; Financial Services, 44.5% over 5733 calls; Technology, 42.8% over 49426 calls; Consumer Cyclical, 35.4% over 5367 calls; Communication Services, 32.6% over 7173 calls; Industrials, 31.6% over 8209 calls; Consumer Defensive, 30.8% over 852 calls; Energy, 30.0% over 1074 calls; Basic Materials, 24.3% over 883 calls; Real Estate, 23.3% over 305 calls; Utilities, 22.6% over 699 calls.
Stock calls only; covers 99.6% of resolved stock-call volume (the major-cap names).
06The RegimeSkill vs. the tide
In the months the S&P 500 rose, FinTwit’s calls won 43% of the time. In the months it fell, 33%. A gap that wide says the tide is doing most of the lifting: when nearly every call is a buy, a rising market flatters the entire field and a falling one drags it down, no matter who made the call.
Riding the Tide
Win rate vs. market direction
Coin flip · 50%
43%
Market rosewin42.8%calls72,899med ret-0.6%
33%
Market fellwin32.6%calls28,848med ret-5.7%
Market rose
-0.6% med
Market fell
-5.7% med
Win rate by market regime: Market rose, 42.8% over 72899 calls; Market fell, 32.6% over 28848 calls.
30d call win-rate split by whether the S&P 500 rose or fell over the same holding window.
By volatility (beta), in each regime
High-beta names amplify whatever the market is already doing: thrilling on the way up, brutal on the way down. The same aggressive call can read as genius in a rally and recklessness in a sell-off, even when the underlying thesis never changed. Low-beta, defensive names move less either way, which is why they hold up when the tape turns.
Beta × Regime
High / mid / low, each way
Market roseMarket fellCoin flip · 50%
44%
High beta (≥1.5) · Market rosewin43.6%calls33,877med ret-1.6%
34%
High beta (≥1.5) · Market fellwin34.4%calls13,778med ret-7.3%
Win rate by beta tier and market regime: High beta (≥1.5) Market rose, 43.6% over 33877 calls; High beta (≥1.5) Market fell, 34.4% over 13778 calls; Mid (1.0–1.5) Market rose, 44.9% over 13728 calls; Mid (1.0–1.5) Market fell, 31.0% over 5353 calls; Low (<1.0) Market rose, 42.5% over 6130 calls; Low (<1.0) Market fell, 33.7% over 2495 calls.
High-beta/momentum names outperform when the market rises and crater when it falls; low-beta names hold up better in the sell-off. Covers 74.1% of resolved-call volume.
— of calls are buys; the crowd rarely calls a top.
Strength in numbers
At peak agreement, the win rate climbs to 30.4%.
Top sector
Healthcare leads every sector at 51.9%.
Ranked field
226 graded accounts this edition.
Cite this report
APA
SignalSnitch. (2026). The Accountability Desk: FinTwit vs. the Market (July 2026 edition). Retrieved from https://signalsnitch.io/state-of-fintwit/m/2026-07
Inline
Per SignalSnitch’s Accountability Desk (signalsnitch.io/state-of-fintwit/m/2026-07), median accuracy across 226 ranked finfluencers is 42.6%.
Free to republish with attribution under CC BY 4.0.
SignalSnitch grades the public market calls of social-media traders against actual price action and publishes the results as a leaderboard. The system is independent of brokerages and runs continuously; every figure on this page comes straight from the live dataset, with no hand-picking. Poise Through the Noise.
Aggregate figures only, across the ranked field. Educational, not investment advice. Frozen edition, computed from the live dataset at month close: July 2026.