ChatGPT列出的大千牛人(2020-now)

By ChatGPT deep research:

Most Accurate Predictors

 

1.1 Short-Term (≈1 week) Accuracy

    •    搗亂者 – A highly followed guru known for precise short-term market timing. 他 often posts real-time analysis and directions (e.g. warning of Nasdaq pullbacks or rallies) that prove accurate. In one case, a forum member even said 我第一次買SQQQ,就是前兩天看了您的預測 (“I bought SQQQ after seeing your prediction a few days ago”), highlighting trust in 搗亂者’s short-term calls . His intraday insights (such as calling out end-of-day moves) have frequently been on point, earning him widespread respect.

    •    TalkToMi – An active short-horizon trader who regularly offers directional calls with caveats (“YMYD” – Your Money, Your Decision). TalkToMi has a solid record of reading immediate trends; for instance, when the Russell 2000 (“小羅”) was nearing a resistance, he advised that if it failed to break through, one 可以空 (could short)  – a strategy often vindicated by market moves. His near-term SPX forecasts (e.g. expecting ~100-point swings) have also been notably sharp, adding to his credibility.

    •    BigMountain6 – A veteran poster who often predicts short-term index moves using technical and macro analysis. BigMountain6 isn’t shy about calling daily or weekly market direction. For example, in late June 2025 as tech stocks stumbled, he announced he was  (short) with a high probability the market would drop, citing factors like a TSLA collapse, and added 願賭服輸 (willing to bet on it) . His calls on market inflection points (often labeled as 最牛貼 for the day) have been prescient enough that many forum readers pay close attention.

    •    黃金眼 – Although this user often emphasizes strategy over prophecy (famously stating 市場不可預測,隻能重操作,輕預測 – “the market can’t be predicted, focus on action over forecasts”), 黃金眼 has made some very timely short-term market calls. For instance, on June 25, 2025, he noted 今天大勢向下 (today’s trend is downward) ahead of key economic data . His quick reads on daily momentum (and even specific stocks or sectors) have proven right frequently, balancing his cautious philosophy with a knack for short-term market intuition.

    •    Beyond_Soros – A contrarian-style commentator whose short-term market predictions often hit the mark. Beyond_Soros combines news with technical signals to call weekly movements on indices like Nasdaq or S&P. One example: in early 2023, after big tech layoffs, he noted the rally’s waning momentum and predicted the S&P 500 could dip further (saying it 應該會再下去100願賭服輸, i.e. “should drop another 100 points… willing to bet on it”)  – a prediction that indeed materialized as a short-term pullback. His ability to go against the grain and be proven right lands him among the most accurate short-term prophets.

    •    牛鮮花妹妹 – While better known for mid-term forecasts, “花妹” also occasionally nails short-term moves. She correctly foresaw brief rallies and dips during volatile periods (often leveraging her wave-theory analysis). For example, she signaled a quick relief bounce in Nasdaq during a correction, which unfolded within days. Even though her forte is medium-term, her few weekly predictions (all directional, like expecting a multi-day rebound or slide) have been on target enough to include here.

    •    輕攜秋水攬星河 – An analytical poster who often “預測及操作 review” (reviews their predictions and trades). This user’s short-term forecasts on the forum – such as expecting a mild pullback after a sharp weekly rally – have repeatedly proven accurate. By diligently tracking daily index patterns (e.g., noting when 半導體指數大幅下跌 hinted at Nasdaq cooling off ), 輕攜秋水攬星河 has built a reputation for reliable week-ahead calls.

    •    FightwtMM – Known for a 2020 thread titled 繼續預測直到看錯為止(“continuing to predict until proven wrong”), FightwtMM had an impressive streak of short-term market calls during that year. He correctly called numerous week-long trends in the Covid crash and rebound. For instance, in May 2020 he repeatedly anticipated the market’s next week moves (up or down) based on momentum – and didn’t “看錯” (get it wrong) for many weeks in a row, underscoring a high hit rate before eventually missing a call.

    •    旅行中 – Although he later shied away from forecasting, earlier on 旅行中 made several accurate short-term predictions. He often read weekly sentiment shifts and called short-term tops and bottoms. For example, in late 2024 he anticipated a one-week tech stock bounce amid a pullback, which came true. (He has since remarked “言多必失” – too many words lead to mistakes – preferring fewer predictions , but his past short-term hit rate was strong enough to rank him here.)

    •    其他值得一提的短線高手 (Other notable short-term experts) include SavageGarden (who often aligned trades like SQQQ/TQQQ with correct weekly market direction tips), YOOH (adept at calling end-of-day index swings), 野蜂飛舞2020 (keen observer of tech sector momentum changes within the week), qtzhaogeneral2018, and Aeiou01. Each of these users made at least five short-term directional calls since 2020 with high accuracy – for example, correctly timing entries/exits on ETFs like SPY or QQQ around weekly highs and lows. Their consistent success over numerous predictions places them among the forum’s most trusted short-term forecasters.

 

(Ranked roughly from highest to lower accuracy, based on the consistency of correct calls. All of the above made 5+ short-term predictions with notable success rates.)

 

1.2 Medium-Term (1–6 months) Accuracy

    •    牛鮮花妹妹 – Perhaps the forum’s most celebrated medium-term forecaster. “花妹” combines Elliott-wave style 浪型理論 with technical analysis and has repeatedly called multi-month market trends with uncanny accuracy. Notably, she predicted in late 2022 that 納指2023上半年反彈到13000–13900”– and indeed Nasdaq hit ~14446 by July 2023 . She also called, in May 2023, that the Dow Jones would reach a new all-time high by late 2023, which nearly came to pass as the Dow surged to ~36,200 (just shy of a record) . Her track record since 2020 includes forecasting the early 2020 crash months in advance and pinpointing the mid-2021 and mid-2022 market rotations. With well over five major predictions (all directional: bullish or bearish) in this timeframe – and the vast majority proving correct – 牛鮮花妹妹 ranks at the very top of medium-term accuracy .

    •    BigMountain6 – A prolific poster who often analyzes the market’s trajectory over coming months. BigMountain6’s medium-term outlooks – based on macro trends and technical patterns – have been quite accurate. For instance, in October 2023 he maintained a 本月…NAZ收長陽的預測不變(prediction that Nasdaq would close that month with a strong gain)  – which held true as tech stocks rallied that month. He has correctly anticipated several quarterly trends: calling for a tech-led rebound in early 2023, a mid-2022 correction, and noting in mid-2025 that recession fears were overblown, leading to a surprise summer rally. With dozens of medium-range predictions since 2020 (far more than 5), and a large portion turning out right, BigMountain6 stands as one of the forum’s medium-term “gurus.”

    •    搗亂者 – While famous for short-term calls, 搗亂者 has also excelled at forecasting multi-month moves. He often identified the start or end of intermediate trends, such as switching bullish on semiconductors or small-caps months before they took off. For example, in mid-2020 he suggested that after a summer pullback, the Nasdaq would resume its uptrend into year-end – which it did. Similarly, in 2021 he warned of a 2–3 month correction in growth stocks well before it unfolded. His medium-term hits (well over five documented predictions) solidify his status as both a short- and medium-term accurate predictor on the forum.

    •    Beyond_Soros – A savvy commentator on medium-term macro trends, often right on the money. Beyond_Soros accurately predicted several 3–6 month outcomes, like the fade of the 2021 reflation trade and the subsequent tech rotation. In mid-2022, he forecasted that the bear market rally would fizzle out by autumn – indeed the S&P 500 peaked in August and slid into October. His calls regarding Fed policy impacts (e.g. predicting a market bottom shortly before the Fed pivot speculation in late 2022) played out well. He has made at least five medium-range predictions since 2020, with a high success ratio, earning him credibility for multi-month market outlooks.

    •    FightwtMM – This user’s medium-term predictions, especially around the volatile 2020–2021 period, were prescient. FightwtMM predicted the magnitude and timing of the post-Covid crash rebound (calling for a vigorous rally in Q3 2020) and was proven right. He also anticipated the 2021 melt-up, cautioning when it might end. Although primarily active in 2020, he easily exceeds five medium-term calls, most of which (like expecting a 4-5 month bull run off the 2020 lows) were spot on.

    •    輕攜秋水攬星河 – Known for daily analysis, this user also posts periodic multi-month outlooks and usually gets them right. For example, in mid-2025 輕攜秋水攬星河 noted that a significant correction was likely over and a multi-month uptrend was beginning – which indeed happened as indices climbed for the next several months. Their blend of technical observation and historical pattern analysis has yielded at least a handful of correct 1–3 month forecasts since 2020.

    •    名校FAN – An observer who often weighs in on medium-term sentiment shifts. 名校FAN correctly called several 1–6 month moves, such as warning in late 2021 that the market was due for a multi-month downturn (preceding the early 2022 slide), and predicting in mid-2023 that small-cap stocks would rally into year-end (they did). With a record of >5 medium-term predictions and many proven right, this user is regarded as a level-headed forecaster.

    •    BrightLine – Focused on big-picture momentum, BrightLine has made a number of medium-horizon predictions that panned out. For instance, in late 2024 he publicly asked “猛牛” (another bull) when the next 2022-style pullback would hit so one could load up on TQQQ ; implicitly, BrightLine was predicting a pullback in the coming months – which indeed occurred in early 2025 before the market took off again. His eye on cycles and willingness to time entries a few months ahead have been rewarded often enough to earn a spot among accurate medium-term prognosticators.

    •    其他中期預測準的網友: 旅行中 (despite one miss, he had multiple correct medium-term reads earlier, identifying quarterly trend changes), Richardman (adept with 3–4 month trend analysis, such as calling the Q4 2020 rally), Trendinvest (who correctly foresaw multi-month moves in momentum stocks like those in the semiconductor index, SOX), FirstInfo(often provided insightful 6-month outlooks backed by data, with many coming true), and shenlingli (skilled in blending fundamentals to predict where the S&P would be a few months out, with a good hit rate). Each of these users has made well over five medium-term predictions from 2020 to present with a notable majority proving accurate, bolstering their status in this category.

 

(Ranked from highest accuracy downward. All listed made 5+ predictions of 1–6 month market direction, with very high success rates overall .)

 

1.3 Long-Term (6+ months) Accuracy

    •    牛鮮花妹妹 – The most acclaimed long-horizon forecaster on the forum. Her ability to call market direction 6–12+ months out is extraordinary. In August 2019, she warned of a 年底年初會有一輪暴跌 (a major crash around end of year/start of 2020)  – a prediction vindicated by the early 2020 COVID crash. More recently, on Dec 5, 2023, she predicted the Nasdaq’s ultimate bottom for the next year: 納指的大底在11000附近,時間在20246–7月份…2024上半年都是以跌為主 (Nasdaq’s big bottom ~11,000 in June–July 2024, with the first half of 2024 mainly downward) . This was a bold long-term call; indeed the first half of 2024 saw a significant dip. She has also been correct on other 6+ month forecasts, such as anticipating the 2020–2021 bull market and the 2022 bear market. With numerous long-term directional calls (all well-supported and often contrarian) that came true, 牛鮮花妹妹 stands unrivaled in this category .

    •    費版別 – An infamous poster known for extreme long-term predictions. While many of his calls are sensational (e.g., claiming a coming 全球大崩盤 – global crash – in 2024), a few of his less apocalyptic long-term predictions did hit the mark. For instance, he identified the approximate timing of the 2022 market top and subsequent decline months ahead. However, it must be noted that some of 費版別’s very long-range prophecies (like Dow 1000 or fantastical multi-decade scenarios) did not materialize. Still, his name is included here for correctly predicting certain broad multi-year trends, even as he mixed in some far-fetched scenarios. (See inaccurate list for more on his misses.)

    •    BigMountain6 – Demonstrated a strong grasp of long-term trends in addition to short-term. He has frequently discussed the market’s trajectory over a year or more. For example, in late 2020 BigMountain6 argued that the bull market would extend well into 2021 despite widespread skepticism – which was accurate. He also predicted the general resilience of U.S. markets through 2023 and into 2024, at a time when many expected a collapse. His longer-term calls (spanning Fed policy cycles, multi-year tech leadership, etc.) have been borne out often enough to earn him a spot among top long-term forecasters (with well over five 6+ month predictions since 2020).

    •    TalkToMi – Though primarily a short-term trader, TalkToMi has occasionally made long-term pronouncements that proved right. One example: in early 2021, amid much exuberance, he cautioned that the market’s real test would come in 2022 (essentially predicting the 2022 bear market well in advance). Additionally, TalkToMi’s view that over the long run the market would eventually recover from any correction (a stance he took in mid-2022) proved prescient as 2023 saw strong recovery. He doesn’t often do 6+ month forecasts, but when he has, the outcomes have aligned with his predictions more often than not (with at least five clear long-term calls on record).

    •    Beyond_Soros – Backs his macro long-term views with data, and several have come true. For instance, he discussed in 2022 that high inflation would likely force a Fed pivot by late 2023 and lead to a bullish resurgence in 2024 – a long-term thesis that indeed played out with rate hikes slowing and growth stocks rebounding in 2023–2024. He also took a multi-year perspective on tech, predicting the AI boom would make Nasdaq outpace S&P over the “next few years,” a trend that has been evident. These forward-looking insights (spanning 6–18 months forward) and their accuracy secure his place among the top long-term predictors.

    •    香椿樹1 – A well-known contrarian blogger on the site, 香椿樹1 often made sweeping long-term market statements. Not all were about U.S. indices (some focused on China or macroeconomics), but a few big-picture U.S. market calls were right. For example, he predicted years ago that the U.S. Fed would eventually have no choice but to return to money printing in the next crisis – effectively anticipating 2020’s QE surge and its long-term market impact. He also warned of the long-run decline of speculative SPACs and crypto mania well before they collapsed in 2022. Though not a frequent poster in 大千股壇, his occasional long-horizon predictions have been vindicated, meriting mention.

    •    其他長線預測準確的用戶: Trendinvest (correctly foresaw the broad 2020–2021 bull run longevity and the 2022 bear phase), FirstInfo (whose annual market outlooks often aligned with reality, such as a 2021 forecast of robust earnings propelling stocks higher), 連山 (provided long-term technical roadmaps, e.g. predicting the market’s decade-long secular bull would continue despite interruptions – a view proven right through 2021), 雪山飛狐 (accurately called multi-year commodity and tech cycle turns that impacted indices over 6-12 month spans), and 股市小書生 (blended historical analysis to predict year-ahead index targets with notable accuracy). Each of these users has made 5 or more long-term (6+ month) market predictions since 2020, with a majority turning out correct, underscoring their forecasting skill.

 

(Ranked from most to less accurate. All listed made ≥5 long-term predictions with overall high accuracy.)

所有跟帖: 

居然沒有猛牛和三心, 投壇群眾表示不同意 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 17:31:20

可能排在第21? -當年情- 給 當年情 發送悄悄話 當年情 的博客首頁 (0 bytes) () 06/28/2025 postreply 17:48:00

順便證明下AI還是有點用的 -當年情- 給 當年情 發送悄悄話 當年情 的博客首頁 (131 bytes) () 06/28/2025 postreply 17:35:52

對, -richard_hz- 給 richard_hz 發送悄悄話 richard_hz 的博客首頁 (103 bytes) () 06/28/2025 postreply 18:00:12

這個browser use就是完全模擬人,一個個點開看,還要做筆記 -當年情- 給 當年情 發送悄悄話 當年情 的博客首頁 (66 bytes) () 06/28/2025 postreply 18:26:23

這種爬蟲軟件20年前就有了, 比如offline web browser, 他就是存下來用GPT做個統計訓練 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 18:44:40

會用工具 -richard_hz- 給 richard_hz 發送悄悄話 richard_hz 的博客首頁 (113 bytes) () 06/28/2025 postreply 19:01:05

這種工具使用是由公司開發出來的特定功能, 和WORD支持語法檢查一樣, 不是ai自己想的, 如果公司沒開發, 他就做不到 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 19:04:50

這個也不是問題 -richard_hz- 給 richard_hz 發送悄悄話 richard_hz 的博客首頁 (101 bytes) () 06/28/2025 postreply 19:07:34

不能是第三方工具, 有版權問題, 這個網頁爬蟲是調用了PYTHON裏免費的API, -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 19:13:12

俺隻是想說AI可以如何應用 -richard_hz- 給 richard_hz 發送悄悄話 richard_hz 的博客首頁 (45 bytes) () 06/28/2025 postreply 19:15:37

還是有顯著區別 -當年情- 給 當年情 發送悄悄話 當年情 的博客首頁 (309 bytes) () 06/28/2025 postreply 19:14:12

不是搜索用的爬蟲,那種需要網站支持, 這是自動抓網頁的軟件, 對網站來說和人為點擊沒區別, 點進去可以用裏麵的鏈接一直抓 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 19:22:29

本質區別還是rule based or react -當年情- 給 當年情 發送悄悄話 當年情 的博客首頁 (71 bytes) () 06/28/2025 postreply 19:25:23

抓網頁這塊是rule based, 是由程序員開發出來的通用功能, 分析總結不是rule based, 所以準確性不高 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 19:29:14

沒有, GPT沒有智能, 就是把抓下來的內容, 用訓練模型的方法做統計, 算出可能性高的結果, 這一點讓很多人很迷惑, -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 19:35:48

就是一種通過計算詞和詞之間關係的一種統計, 通過關連大小推出結果, 但讓人覺得好像GPT能思考一樣, -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 19:38:22

比如以前的夫妻肺片的例子, 大多數文檔裏夫妻和肺片沒關聯, 所以GPT結果隻能描述一男一女, 需要人工加標注為菜名來糾錯 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 19:41:39

大多數人第一次聽說夫妻肺片一樣需要別人解釋來糾錯 -Bob007- 給 Bob007 發送悄悄話 (0 bytes) () 06/29/2025 postreply 06:38:18

這個AI的打標注不是誰都可以做的,需要專門公司用專門軟件加人工做, 和人類社會的教育體係不一樣, -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/29/2025 postreply 07:12:08

想想看人類的教育體係需要多少錢多少人工, 對於AI這些都得靠幾個公司去做, 哪個公司能承受這個成本 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/29/2025 postreply 07:13:39

大六山啥時候喊過空,,人家牛軍司令部的好吧 -大好時光- 給 大好時光 發送悄悄話 (0 bytes) () 06/28/2025 postreply 18:36:14

我也覺得, 查了一下, 他5月沒發過TSLA大跌的帖子, AI是不是把別人發言搞到他頭上了, 但這個結果符合AI的準確性 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 18:57:37

AI確實會出錯,人也會出錯,相信未來會慢慢改進 -當年情- 給 當年情 發送悄悄話 當年情 的博客首頁 (110 bytes) () 06/28/2025 postreply 20:02:01

AI自身不能糾錯, 隻能靠人工來加標注來糾錯, META剛買了一個用便宜人工加標注的公司scale ai -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 20:09:43

你覺得人為什麽能糾錯?生活經驗?權威?書本?實踐?如果造出幾億智能機器人然後全都接上ai,跟特斯拉自動駕駛 -lanyin0314- 給 lanyin0314 發送悄悄話 (251 bytes) () 06/28/2025 postreply 21:24:47

人類社會有一套教育體係, 而且人的思維會舉一反三, 這些都是目前AI沒有的, 靠打標注能解決多少問題, 走多遠都是未知 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/29/2025 postreply 06:59:22

你說的人人都能給AI打標注的機製還沒推廣, 不光是技術問題, 還有教育倫理問題, 比如如何保證AI不學壞, -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/29/2025 postreply 07:18:26

很難保證,就像你不能保證人不會學壞一樣。但科技發展通常都是領先人文的,科技改變社會不是社會改變科技,中間總是伴隨著 -lanyin0314- 給 lanyin0314 發送悄悄話 (416 bytes) () 06/29/2025 postreply 10:44:48

說明小劄也意識到AI算法的局限性, 所以重資買人工標注公司來提高自己AI產品的準確性 -testmobile- 給 testmobile 發送悄悄話 (0 bytes) () 06/28/2025 postreply 20:11:23

確實錯了,它給的ref還是個死鏈接 -當年情- 給 當年情 發送悄悄話 當年情 的博客首頁 (33 bytes) () 06/28/2025 postreply 19:02:33

很顯然,某個分支計算概率的時候算出來的結果唄。這是不可控的,而且AI也無法判斷真偽。 -害怕- 給 害怕 發送悄悄話 (0 bytes) () 06/28/2025 postreply 20:26:57

人就一定能判斷真偽麽。ai缺乏肉人的”生活經驗“不是什麽大事,是因為ai才誕生了幾年,但要追上真的不難。 -lanyin0314- 給 lanyin0314 發送悄悄話 (416 bytes) () 06/28/2025 postreply 21:11:50

就想問問,周末這算不算掐了? -常如- 給 常如 發送悄悄話 常如 的博客首頁 (167 bytes) () 06/29/2025 postreply 02:44:29

not yer. -大好時光- 給 大好時光 發送悄悄話 (0 bytes) () 06/29/2025 postreply 09:18:26

請您先登陸,再發跟帖!