幾塊你要買？｜ ETF 折溢價

台股、美股兩版都是個股的估價公式。ETF 是一籃子證券、沒有財報，個股那十一種公式大多完全不適用，所以獨立成這一頁，看的是 ETF 真正該看的指標。 The Taiwan and US editions apply valuation formulas to individual stocks. An ETF is a basket of securities with no financial statements, so most of those eleven formulas simply do not apply — hence this separate page, which shows what actually matters for an ETF.

為什麼 ETF 不能套用個股的估價公式？

ETF 是一籃子證券，沒有財報 —— 沒有每股盈餘、營收、淨利、股東權益報酬率，也沒有盈餘成長率。所以個股那些公式：

本益比法（年化 EPS）	✗ 沒有季報
本益比法（近四季）	✗ 交易所的本益比資料只涵蓋個股
本益成長比 PEG	✗ 沒有盈餘成長率
月營收動能法	✗ 沒有營收
葛拉漢公式	✗ 需要 EPS
ROE 法	✗ 沒有股東權益報酬率
股價營收比 P/S	✗ 沒有營收
股價淨值比 P/B	△ 算得出來，但意義完全不同
股利法	✓ 配息型 ETF 適用

股價淨值比那個「△」是重點。個股的股價淨值比是估值指標， 1.2 倍和 5 倍代表不同的貴賤；但 ETF 的淨值就是它持有的那籃子證券的市值總和，所以理論上市價應該恆等於淨值。偏離不叫「便宜」或「貴」，那叫折溢價 —— 是造市、流動性與申購贖回機制的問題，用估值的邏輯去解讀會完全會錯意。

折溢價率

折溢價率 = (市價 − 淨值) ÷ 淨值 × 100%

溢價（正值）代表你付的價格高於它實際持有的資產價值，等於多付錢；折價（負值）則相反。正常情況下會貼近 0，由造市商透過申購贖回機制拉回。持續偏離通常出現在流動性差、連結海外市場（有時差）、或市場情緒過熱的時候。

折價就是便宜嗎？

單就算術而言：折價代表你用低於淨值的價格買到同一籃資產，溢價則相反。但把它當成買賣訊號之前，有四件事會讓這個直覺失準：

幅度通常很小。台股 ETF 的折溢價多半在 ±1% 以內，由造市商透過申購贖回機制拉回。相對於長期報酬，這個幅度往往不是主要變數。
連結海外市場的，折價常是「時差假象」。 美股、日股或海外債券 ETF 的淨值反映的是前一個海外交易日收盤的持股價值，但台股盤中的市價已經反映了最新的海外期貨走勢。海外市場今天大跌時，這類 ETF 會先跌、淨值卻還停在昨天 —— 畫面上出現「折價」，實際上只是淨值還沒更新，不是打折。債券與跨國型 ETF 尤其要小心。
折價可能反映流動性差。成交量低的 ETF 買賣價差大，省下的折價可能還不夠付來回的價差成本。
內扣費用的影響通常更大。經理費與保管費逐年扣抵，年費差 0.5% 持有十年的落差，遠大於一次進場的 1% 折溢價。

換句話說，折溢價比較像是「同一檔 ETF 挑進場時機」的參考 —— 同樣要買它，溢價 2% 時買不如貼近淨值時買 —— 而不是用來 比較不同 ETF 的好壞。跨類型比較（債券 vs 原型 vs 槓桿）更沒有意義。

ETF 還該看什麼

本頁沒有涵蓋、但同樣重要的：內扣費用（經理費＋保管費，長期報酬的最大變數）、追蹤誤差、規模與流動性、以及成分股的集中度。這些資料交易所沒有提供，需要到各投信官網或投信投顧公會查詢。

An ETF is a basket of securities and has no financial statements — no earnings per share, no revenue, no net income, no return on equity, no earnings growth. So of the formulas used for individual stocks:

P/E (forward and trailing)	✗ No quarterly filings
P/E (trailing)	✗ The exchange's P/E data covers stocks only
PEG ratio	✗ No earnings growth rate
Monthly revenue momentum	✗ No revenue
Graham formula	✗ Needs EPS
Return on equity	✗ No return on equity
Price / sales	✗ No revenue
Price / book	△ Computable, but it means something entirely different
Dividend method	✓ Works for income ETFs

That "△" on price-to-book is the important one. For an individual company, P/B is a valuation measure: 1.2x and 5x say different things about how expensive it is. But an ETF's net asset value simply is the market value of the basket it holds, so in theory price should always equal NAV. A deviation is not "cheap" or "expensive" — it is a premium or discount, a question of market-making, liquidity and the creation/redemption mechanism. Reading it as a valuation signal misses the point entirely.

Premium / discount

premium = (market price − NAV) ÷ NAV × 100%

A premium (positive) means you are paying more than the assets it actually holds are worth; a discount (negative) means the opposite. Normally it sits close to zero, pulled back by authorised participants through creation and redemption. A persistent gap usually shows up where liquidity is thin, where the fund tracks an overseas market in a different time zone, or when sentiment runs hot.

Is a discount the same as cheap?

Arithmetically, yes: a discount means you are buying the same basket for less than NAV, and a premium the reverse. But before treating it as a signal, four things break that intuition:

The magnitude is usually tiny. Most Taiwan-listed ETFs stay within ±1%, pulled back by the creation/redemption mechanism. Against long-run returns, that is rarely the variable that matters.
For funds tracking overseas markets, a discount is often a time-zone illusion. The NAV of a US, Japanese or overseas bond ETF reflects holdings as of the previous overseas close, while the Taipei price already reflects the latest overseas futures. When those markets fall hard, this kind of ETF drops first while NAV is still stuck at yesterday — the screen shows a "discount" that is really just a stale NAV, not a markdown. Bond and cross-border ETFs especially.
A discount can be a symptom of poor liquidity. Thinly traded ETFs have wide bid-ask spreads, and the discount you save may not cover the round-trip spread.
Expense ratios usually matter more. Management and custody fees are deducted year after year; half a percentage point of annual fees compounded over ten years dwarfs a one-time 1% premium.

In other words, the premium is closer to a way of timing your entry into one particular ETF — if you are buying it anyway, buying near NAV beats buying at a 2% premium — than a way of comparing different ETFs. Comparing across types (bond vs plain-vanilla vs leveraged) is less meaningful still.

What else to look at

Not covered on this page but equally important: the expense ratio (management plus custody fees — the single biggest variable in long-run returns), tracking error, size and liquidity, and how concentrated the holdings are. The exchange does not publish these; check each fund manager's site or the SITCA.

幾塊你要買？