Psychology & Behavior
Base Rate Forecasting
基準率參照 · Source: Philip Tetlock / Daniel Kahneman
Decisions requiring outcome forecasts — from startup success rates to investment returns, start by asking "how does this type of thing typically go"
Core Concept
People naturally reason from their specific situation rather than statistical baselines. Base rate forecasting requires you to first find a "reference class" — the category of events most similar to yours — and start from that class's historical outcomes before adjusting for your specifics.
✓ When to use this
When predicting probabilities, estimating success rates, evaluating "can I pull this off." Startup success rates, drug trial pass rates, job-change salary lifts — anywhere historical statistics exist.
✗ When not to use this
When your situation truly diverges from the reference class — e.g. you have advantages absent from the statistical sample (insider resources, unique skill stack). But "I am special" itself needs proof first.
Questions you will be asked
Using this framework, you will work through —
- 1.What are you trying to forecast or evaluate?
- 2.Find your reference class: what type of event does this decision belong to?
- 3.What are the historical outcomes for this reference class?
- …and 3 more
Worked example
Expand to see what a filled-in run looks like
›
Worked example
Expand to see what a filled-in run looks like
Situation
預估自己出版第一本商管書能賣多少本、是否值得花一年寫。
1. What are you trying to forecast or evaluate?
第一本書能賣多少本、會帶來多少版稅？
2. Find your reference class: what type of event does this decision belong to?
台灣商管類首版書，沒名氣作者第一本書。不是「商管書」這麼大類別。
3. What are the historical outcomes for this reference class?
中位數約 1500–2500 本，前 25% 到 5000 本，前 5% 到 15000 本以上；版稅約 8–10%。
4. What specific factors in your situation deviate from the base rate? In which direction — better or worse?
+：我有 5 萬粉絲的電子報；-：我從沒寫過長篇；+：題目跟我本業強相關；-：出版社是中型不是大型。淨值：略高於中位數，估 3000 本上下。
5. Considering the base rate and your adjustments, what is your honest estimate?
預期銷量 3000 本、預期版稅約 NT$45k。
6. Has this honest estimate changed your willingness to make this decision?
改變了。原本以為「不錯的書應該能賣 1 萬本」，看完基準率知道機率很小。寫不寫的決定不該基於版稅，要基於品牌與內容沉澱的價值。
Use it inside ChatGPT / Claude
Paste the prompt below and the AI will walk you through this framework, one question at a time.
你現在是引導使用者做「基準率參照」的決策教練（Tetlock / Kahneman）。 人們預測結果時習慣從自己的具體情況推論，忽略統計基準。 依序問： 1) 你正在預測或評估什麼？ 2) 找到參考類別：這個決策屬於哪一類事件？（要夠具體，不能只說「新創公司」） 3) 這個參考類別的歷史結果是什麼？要查實際資料。 4) 你的具體因素如何讓你偏離基準？哪個方向？ 5) 結合基準率和調整後，你的誠實估計是多少？（要數字或範圍） 6) 這個誠實估計改變了你做決策的意願嗎？ 當使用者用模糊的「應該還可以」回答時，逼他給出機率或數字範圍。 互動規則： 1. 一次只問一題，等使用者回答後再進入下一題。 2. 使用者答完所有題目前，不要做總結或下結論。 3. 若答案太抽象、太籠統，請追問一次具體例子或數字後再繼續。 4. 全部答完後，輸出三段：(a) 摘要使用者的關鍵判斷；(b) 你看到的盲點或張力；(c) 一個具體下一步行動建議。 5. 不要替使用者做決定，只把判斷攤開讓他自己決定。
Related Frameworks
Psychology & Behavior
Inside View vs Outside View
Planning and forecasting — especially when you're overly optimistic about your own project, goals, or capabilities
Investment & Finance
Expected Value Analysis
Decisions with quantifiable outcomes — investments, business decisions, choices with probability and payoff structures
FAQ
Is base rate forecasting the same as inside/outside view?
Not quite. The inside/outside view is the mindset (outside first, then inside); base rate forecasting is its concrete method: find a reference class → take that class's historical results → adjust for your situation. The outside view tells you to look outward; base rate forecasting tells you how.
What if I can't find precise base rate data?
You don't need precision. A rough but honest range ("this kind of startup survives three years about 30–40% of the time") is enough to correct your optimism. The point of a base rate isn't an exact forecast — it's pulling you from the "mine will definitely work" inside narrative back to the right order of magnitude. Better roughly right than precisely wrong.
My situation is genuinely special — doesn't the base rate not apply?
"I'm special" is optimism bias's favorite line. The base rate's default is that you're like the reference class; unless you can name a specific, verifiable way you clearly beat that class, the rational expectation is the base-rate outcome. A difference needs evidence, not just a feeling.
Related studies