Long-form research from the fxcryptobots desk: Twitter-influencer backtests, on-chain edge studies, sector vs single-name alpha, and prediction-market smart-money work. Every report ships with the raw data, the methodology, and a downloadable PDF.
60-day quantitative backtest of @Ren_aramb (24,432 followers, AI / Head of AI / 10y PM). 376 original tweets, 255 ticker-mention tweets, BULL 139 / BEAR 30 / NEUTRAL 86. BULL signals returned +25.22% mean at the 14-day horizon (n=125, WR 87%) with +13.67% alpha vs XLK and +21.16% vs SPY. Concentrated AI-infrastructure / photonics / NeoClouds / semiconductor universe ($SIVE, $LITE, $NBIS, $POET, $AAOI). Includes sentiment-by-horizon table, most-mentioned tickers, best/worst 30-day calls, methodology and a calibrated FOLLOW verdict bounded by regime caveats.
70-day quantitative backtest of @ParadisLabs (45,631 followers, AI / Semiconductor Analyst & Trader). 353 original tweets, 231 ticker-mention tweets, BULL 103 / BEAR 27 / NEUTRAL 101. BULL signals returned +18.66% mean at 14 days (n=122, WR 79%, +6.19% alpha vs XLK) and +55.58% mean at 30 days (n=37, WR 97%, +28.21% alpha vs XLK). High-conviction supply-chain mapping across $NVDA, $TSM, $AAOI, $SIVE, $IQE, $SNDK, $LITE, $NBIS. Account is <3 months old so no out-of-sample data; calibrated FOLLOW verdict bounded by regime + persistence caveats.
Quantitative backtest of @pennycheck (61,226 followers, 50,091 lifetime tweets, asymmetric micro-cap / semi style). 165 original tweets sampled across 585 calendar days, but only 36 contain extractable $TICKER cashtags - the author refers to own positions with a penguin emoji rather than $TICKER, so the regex sentiment classifier under-counts heavily. BULL n=14 / 5d +5.98% (WR 79%), BULL n=5 / 14d +11.37% (WR 100%) - directional but underpowered. Recent (Apr-May 2026) $POET BULL -> BEAR pivot was well-documented and was a meaningful inflection invisible to the regex. Verdict: market-color feed, do not follow mechanically.
42-day quantitative backtest of @labubu_trader (3X Long Labubu, 55,358 followers, 16-year account). 232 original tweets, only 36 ticker-mention tweets, BULL 13 / BEAR 5 / NEUTRAL 18. BULL 14d returned +6.78% mean (n=7, WR 71%) but alpha vs the tech sector ETF (XLK) is NEGATIVE -3.52% - the equity picks underperform the tech tape. BEAR n=3-4 produced positive returns but the sample is too small to interpret. Real strength is in macro overlays ($TLT, oil hedges, US10Y / SPX timing, FOMC / NFP framing) and event context - mostly cashtag-free, so invisible to the equity backtest. Verdict: macro color, do not follow for equity picks.
24-month quantitative backtest of @legen_eth (InsiderWave founder, 169K verified followers) and his 'copy politicians, beat the market' thesis. Pelosi-style 10-name mega-cap basket (GOOGL/NVDA/AAPL/AMZN/AVGO/PANW/TEM/CRM/MSFT/META) returned +61.82% over 24m vs SPY +40.72% (+15% excess) but trailed QQQ by 9.5% on the 1-year window. Tweet-follow alpha on 20 BULL mentions: 5-day mean alpha -0.57%, win rate 45% (coin-flip floor). Edge decomposes to mega-cap tech beta, not politician selection. Same promoter-personality cluster as @kevinxu / @vulturetrades / @BoraOzkent.
Beta-control audit of @BoraOzkent (155K verified followers, founder of paid Skool community 'Haddini As Kulubu', Istanbul-based US tech investor) and his two public performance claims: +99.10% since Mar 7, 2025 and +33.30% YTD 2026. Pure NVDA buy & hold over the same 15-month window returned +94.85%, explaining 95.7% of the +99% claim. Implied NVDA-equivalent weight to match exactly: 108.5% (NVDA + small option leverage). The +33% YTD claim does not reconcile with any pure-equity mega-cap allocation tested (max +23% on GOOGL alone) - either option-heavy or post-Q1-dip entry. Stock-selection edge after beta-control: statistically zero.
Rationalized consolidated analysis of @jukan05 (120K verified followers, supply-chain narrative analyst). Two parts: (1) 16-month sector-controlled backtest -- +5.82% mean alpha vs SMH at 60-day holds (WR 57%, n=294), +31.85% at 180-day holds (WR 68%, fat-tailed). MEMORY_BULL is the only topic with strong sector-controlled alpha. 7-name basket returned +294.86% over 16m, but the fair benchmark is SMH (+120.33%) -- honest excess is ~2.5x sector, not 8x NASDAQ. Counterfactual analysis: SNDK+STX or INTC+MU would have produced similar/better outcomes -- the skill is theme identification, not stock picking. (2) Forward-looking 12-position niche portfolio with EVIDENCE labels: 2 DIRECT cashtag positions (DELL, INTC = 33%), 8 inference-mapped (64%), 1 analyst-report share (CWAVE 3%), 2% cash. Realistic forward expectation: base case tracks SMH; we explicitly do NOT project a repeat of the historical 8x NASDAQ outcome.
33-day intraday backtest of every option-entry tweet from @vulturetrades (164.7K verified followers). 152 deduped calls, all calls 0 puts. Mirroring every entry compounds $1 → $0 (−100%); 87 of 101 expirations went to zero (win rate at expiration 10.9%). Tweets do move the underlying briefly (+1.28% mean intraday MFE) but EOD alpha is statistically zero. 73% of his most-viral tweets are DM-list-building / affiliate posts. Verdict: do not follow, but a contrarian mega-cap fade is worth validating.
12-month quantitative backtest of @kevinxu (CEO @alpha_ai, founder @AfterHour_HQ, 139K followers). 832 tweets, 209 ticker mentions, 14 first-mention BULL signals. Two clean asymmetries: mega-cap FOLLOW (NVDA, AAPL, GOOG, INTC, ARM) returns +3.93% alpha at 90d with 80% win rate; small/mid-cap FADE (OKLO, IREN, HIMS, BOT, QS) returns +9.69% alpha at 90d with 89% win rate (or +16.64% at 14d with 100% win rate). Same trader, opposite edges, separated cleanly by market cap.
10-month, 5,126-mention quantitative backtest of Serenity's (@aleabitoreddit) BULL stock picks held to today. Net alpha -37.32% vs XSD semi ETF, 79% of picks underperformed the sector. Includes top 25 winners (AXTI +711%, SNDK +516%), top 25 losers, high-conviction trap (HIMS, CRCL, MSTR), and a pragmatic cherry-pick + counter-trade strategy.
2.4-year, 136-mention quantitative backtest of Zephyr’s (@zephyr_z9) BULL stock picks held to today. Net alpha +13.17% vs XSD semi ETF, +63.59% vs SPY, carried by a memory/storage thesis (SNDK +1,145%, MU +283%, WDC +210%, LITE +154%). Mega-cap and China ADR picks dragged ≥100 alpha points each. Includes top 15 winners, top 15 losers, high-conviction names, portfolio simulation and account comparison.