Skip to main content
Home
News Repo 2026 (v05)
since 2026-01-01

Breadcrumb

  1. Home
  2. 散戶投資者失敗十大原因

散戶投資者失敗十大原因

KF Cheng
By KF Cheng on Fri, 2026-03-13 - 17:57
Short Title
散戶投資者失敗十大原因
Private
Public
Image
蕭若元說
2026-03-13

Post in Patreon by 小god — 8:18 AM Mar 13, 2026

📉 散戶投資失敗十大原因(Evidence-Based)

過度交易(Overtrading)
Data:

Barber & Odean (2000)研究60,000個散戶帳戶
交易最頻繁的20% → 年回報11.4%
交易最少的20% → 年回報18.5%
Trading cost + bad timing = -7.1% underperformance

Why:

每次交易有cost(佣金、價差、稅)
散戶timing通常差(買high賣low)
Market noise被當成signal

 

你可能未諗過的點:
即使你100% timing正確,overtrading仍然hurt你。
Example calculation:
假設你年回報20%(極好)
但你每月rebalance一次(12 trades/year)
每次交易cost 0.5%(價差+滑點+稅)
→ Total drag = 12 × 0.5% = 6%
→ Net return = 14%

如果你只是buy and hold:
→ Return = 20%

Difference = 6% annual underperformance
Over 10 years = 79% vs 619% return
你的case:

你做semi-annual review(一年2次)
這是optimal range
但很多人做不到,忍不住constantly調整


追逐過去表現(Chasing Past Performance)
Data:

Morningstar研究:散戶資金流向
5-star基金流入最多
但這些基金未來表現通常underperform
Past performance ≠ future results(但散戶不信)

Dalbar QAIB研究(2023):

S&P 500過去20年年化回報:9.8%
平均散戶實際回報:6.8%
Behavior gap = -3% annually

Why散戶chase:
看到某股票過去1年 +200%
→ FOMO
→ 買在peak
→ 之後回調-50%
→ Panic賣
→ 然後它又升
你可能未諗過的點:
"Buy the dip"和"chasing performance"的分別在於你的research depth。
Chasing performance:

Stock升了很多 → 我買
不明白business model
Pure momentum gambling

Buy quality on panic:

Stock跌了很多
但fundamentals仍solid
你研究過先買

Example from你的watchlist:
Stock情況Chasing vs Smart buyNVDA 2022Peak $346 → 跌到$108Chasing = 買$346,Smart = 買$108-150META 2022Peak $384 → 跌到$88Chasing = 買$384,Smart = 買$90-120PLTR 2021Peak $39 → 跌到$6Chasing = 買$30+,Smart = 買$6-10
你的"buy quality on panic"strategy explicitly避開這個。
但要小心:區分真panic(temporary)vs structural decline(permanent)。

過度自信(Overconfidence)
Data:

82% drivers認為自己above average(statistically impossible)
74% mutual fund managers認為自己above average
散戶overconfidence更嚴重

Manifestations:
A. 過度集中(Under-diversification)

相信自己能pick winners
All-in 1-3隻股票
High risk without proportional return

B. 過高槓桿

相信自己timing準確
用margin/options放大
一次錯誤就爆倉

C. 忽視風險

"這次不同"
"我研究過,不會錯"

Famous last words

你可能未諗過的點:
Overconfidence和conviction的分別:
Overconfidence:

基於limited data
忽視contrary evidence
Emotional certainty

Rational conviction:

基於comprehensive research
考慮了bear case
Position sizing反映uncertainty
Intellectual humility despite strong view

Test:如果你的thesis wrong,你會lose多少?
Overconfident investor:

50-100% portfolio in one bet
All-or-nothing

Rational investor:

即使high conviction,cap at 20-30%
Survive to fight another day

你的portfolio:

你提過持有PLTR和NVDA
但沒說position size
如果combined >50%,可能overconfident
如果<30%,reasonable conviction


錨定效應(Anchoring Bias)
What:

過度依賴first piece of information
難以adjust from initial reference point

Investment manifestation:
A. 買入價錨定
你$100買入stock
現在worth $50
你想:"等它返回$100先賣"

問題:Market不care你買入價
如果fundamentals變差,它可能去$20
B. 歷史高位錨定
Stock曾經$200
現在$100
你想:"好平,50% discount!"

問題:可能business已經變差
"便宜"的東西可能變更便宜
Real example:
TSLA case(可能relevant to你):

2021 peak: ~$410(split-adjusted)
很多人在$300-350買入
2022跌到$100
很多人hold住想:"等返$300先賣"
2023-2024升返$200-250
但很多$300買入的人仍在等
機會成本:2年沒賺錢

你可能未諗過的點:
每天都是新的投資決定。
不應該問:"我買入價$X,現在應該hold還是賣?"
應該問:"如果我今天手上有現金,我會買這隻股票嗎?"
如果答案是no → 賣
如果答案是yes → hold
買入價irrelevant(除非tax considerations)。

確認偏誤(Confirmation Bias)
What:

只看supporting evidence
忽視contrary evidence
自己騙自己

Investment context:
你買了PLTR:
Confirmation bias做法:

✅ 讀所有bullish articles
✅ Join PLTR bull討論區
✅ 記住所有positive news
❌ 忽視negative news
❌ Dismiss bear arguments
❌ 不看short reports

Result:

你convinced PLTR只升不跌
當真正風險出現,你blind sided
太遲react

Real example:
Cathie Wood / ARK Invest:

2020-2021大升時:Market genius
Bullish on所有disruptive tech
只看confirming evidence
2022-2023:ARKK -75% peak to trough
因為confirmation bias miss了rising rates風險

你可能未諗過的點:
Best investors actively seek disconfirming evidence。
Charlie Munger:

"I never allow myself to have an opinion on anything that I don't know the other side's argument better than they do."

Practical habit:

For every stock you own,read最少一篇credible bear case
能夠articulate為什麼bears錯
如果你不能,maybe they're right

Exercise for你:
PLTR bull case(你肯定知):

AI platform leadership
Government contracts moat
等等

PLTR bear case(你能articulate嗎?):

Stock-based compensation極高(~30% revenue)
Government revenue growth放緩
Commercial sector adoption uncertain
Valuation expensive(P/S still >20)
Karp隨時可能賣股
這些你認真考慮過嗎?

如果你能反駁這些bear points → conviction justified
如果你沒認真想過 → confirmation bias

損失厭惡(Loss Aversion)
Kahneman & Tversky:

損失的痛苦 = 2.5x 同等收益的快樂
People hate losing more than they love winning

Investment errors:
A. Hold losers too long
你$100買入
跌到$80:-20%,心想"等升返先賣"
跌到$60:-40%,心想"已經跌咁多,應該會反彈"
跌到$40:-60%,心想"都輸咁多,不如hold住"
跌到$20:-80%
每個階段你都在rationalize不賣,因為實現loss太painful。
B. Sell winners too early
你$100買入
升到$120:+20%,心想"賺咗啲,take profit保平安"
之後升到$300
Loss aversion讓你太早lock in small gains。
你可能未諗過的點:
"Cut losses short, let winners run"是最難執行的策略。
因為它違反human nature:

Cutting losses = 承認錯誤(painful)
Letting winners run = 冒風險失去profit(scary)
Instinct是相反:hold losers, sell winners

Data from你的PIOF framework可能幫到:
如果你有systematic sell criteria:

PIOF score >15 = 賣
不管profit/loss
Remove emotion from decision

但問題:你真的會execute嗎?
Example:

假設PLTR有天PIOF score升到16
但你仍有20% unrealized gain
你會賣嗎?

如果答案是"我會再研究下"→ 你可能有loss aversion(想keep winner)
如果答案是"根據rules就賣" → 你有discipline

羊群效應(Herding Behavior)
Data:

IPO首日平均+18%,但1年後average return 0%
SPAC mania 2020-2021:平均-50% from peak
When everyone買,通常是頂部

Mechanism:
媒體報導某sector很hot
→ 朋友開始買
→ 你FOMO
→ 大量散戶湧入
→ Price peak
→ Smart money exit
→ Crash
Recent examples:
HerdPeakTroughLessonCrypto 2021BTC $69k$15k-78%SPAC 2021Avg $15Avg $7-53%Meme stocks 2021GME $483$20-96%AI hype 2023C3.AI $40$20-50%
你可能未諗過的點:
Herding和conviction的分別:
Herding indicators:

❌ "Everyone都買緊AI stocks"
❌ "朋友話PLTR好"
❌ "Reddit都話bullish"
❌ "不買會miss out"

Conviction indicators:

✅ "我研究了fundamentals"
✅ "我understand business model"
✅ "我做了PIOF analysis"
✅ "即使別人不同意我都confident"

Test:
如果PLTR突然不再popular(媒體不報導,Reddit不討論),你仍會hold嗎?
If yes → conviction
If no → you were herding
你的case:

你有systematic framework(PIOF)
你做independent research
Likely不是herding

但watch out:AI sector整體可能是herd。
你的portfolio heavily tilted to AI(NVDA, PLTR)

是因為你genuinely相信AI transformation?
還是因為AI是hot topic?
Be honest with yourself


忽視費用和稅務(Ignoring Costs & Taxes)
你可能未諗過的點:
很多散戶focus on gross return,ignore net return。
Example calculation:
Investor A(頻繁交易):
Gross return: 20%
Trading costs: -2%(佣金+價差)
Taxes: -5%(short-term capital gains)
Net return: 13%
Investor B(buy and hold):
Gross return: 15%
Trading costs: -0.1%(很少trade)
Taxes: -2%(long-term capital gains + deferred)
Net return: 12.9%
看似A更好(20% vs 15%),但net幾乎一樣。
Over 30 years:

A: 13% compounded = 39.1x
B: 12.9% compounded = 37.4x
差不多

但B的risk可能更低(less turnover)。
Hong Kong context:
好消息:Hong Kong沒capital gains tax

這是huge advantage
你可以頻繁trade without tax drag
但仍有trading costs

但如果你將來移民:
CountryCapital gains taxUSA15-20%(long-term),up to 37%(short-term)UK10-20%Canada50% of gain taxed at marginal rateAustralia50% discount if hold >1yr
所以你的"semi-annual review"strategy:

在Hong Kong:optimal
如果移民美國:可能要調整成更long-term
Tax jurisdiction matters


時機選擇妄想(Market Timing Delusion)
Data:

Morningstar:散戶trying to time market → underperform by 1.7% annually
"Time in market > timing the market"

Why timing fails:
Miss best days problem:

如果你miss S&P 500最好的10天(over 20 years)
Return從9.5% → 5.6%
但這10天無法預測

你可能未諗過的點:
Timing和allocation的分別:
Market timing(通常fail):

"市場太貴,我全部賣出等跌"
"市場會crash,我等$20再買"
All-or-nothing approach

Strategic allocation(可行):

"估值高時降低equity allocation 70% → 50%"
"估值低時提高到80%"
Gradual adjustment

你的approach(semi-annual review):

不是predict short-term moves
是reassess fundamentals
如果fundamentals變差,減持(不是predict crash)
如果fundamentals improve,加持

這是allocation,不是timing。Good。
但要小心一個陷阱:
"等跌先買"syndrome:
你研究完某股,覺得好
但心想:"等跌多10-15%先買"
結果它升了30%
你miss了
Better approach:

如果fundamentals justify current price → 買一部分(e.g. 50% position)
如果跌 → 買剩餘50%
如果升 → 你至少有50%
不會完全miss


缺乏系統(No System/Process)
這是最fundamental的failure。
Successful investing需要:

Clear investment philosophy
Systematic research process
Position sizing rules
Entry/exit criteria
Review schedule
Written down and followed

大部分散戶:

❌ No clear philosophy(今日value,聽日growth)
❌ Random research(睇下新聞,聽下朋友)
❌ No position sizing(想買幾多就幾多)
❌ No exit plan("升到某個價"但無具體數字)
❌ No review(buy and forget)

Result:Emotional decisions。
你可能未諗過的點:
System不只是防止bad decisions,也是learning tool。
With system:
Decision → Document reasoning → Outcome → Review
→ Learn what works → Improve system
Without system:
Decision → Outcome → "運氣好/壞" → No learning
你的PIOF framework是excellent start:

✅ Clear criteria(5 categories, 25 sub-indicators)
✅ Scoring system(0-5 per category)
✅ Threshold(e.g. >15 = avoid)

但要ensure:
A. 你actually follow它

不是做完analysis然後ignore
不是selective application
Discipline matters

B. 你document決定

Why你買/賣
當時PIOF score
Expected outcome
這樣才能後續review

C. 你定期review system本身

PIOF criteria是否仍relevant?
Weighting是否要adjust?
System evolve with market


🎯 Summary:

十大失敗原因#Failure modeCore issue

  1. 你的vulnerability
  2. OvertradingTransaction costs✅ Low(semi-annual)
  3. Chasing performanceBuy high sell low✅ Low(buy on panic)
  4. OverconfidenceExcessive risk⚠️ Medium(取決於concentration)
  5. AnchoringCan't adapt⚠️ Medium(需要monitor)
  6. Confirmation biasIgnore risks⚠️ Medium(seek bear case?)
  7. Loss aversionHold losers✅ Low(有PIOF exit rules)
  8. HerdingFollow crowd⚠️ Medium(AI sector popular)
  9. Ignoring costsHidden drag✅ Low(HK no cap gains tax)
  10. Market timingMiss opportunities✅ Low(focus on fundamentals)
  11. No systemRandom decisions✅ Low(have PIOF framework)

 

投資理財
About BYOS

Build-Your-Own-System is a practice to build digital solutions for business and enterprise based on integration of open source technology and web resources, an excellent approach for digital transformation by practice and experience.  It does not require IT background or programming, because business analysis, experiment and digital innovation is what it is about.

BAYGO

Build-as-you-GO is the strategy behind building a site, and the flexibility of BYOS architecture supports this strategy, which makes everybody the best designer for their own system over time.

More about BayGO

BYOS Workshops

It is the skillset for the digital era.  No IT background is needed.  No programming is needed. It is just a matter of hands-on practice.  You'll be surprised you can do it too.

Our Services

We provide services in:

  • Agile System Building
  • Digital Transformation Consultancy
  • Knowledge Management
  • Combined Consultancy
  • BYOS Workshops
  • BYOS Mentorship
A self-hosted platform for knowledge management
Built by KF Cheng    ©2026
Self-hosted Platform: KHub v4.00.62
Running on Ubuntu 22.04 PHP 8.1 MariaDB 10.10 Nginx 1.18
Clear keys input element