Mirae Asset Securities Has Three Listed Tickers. Only One of Them Rallied in 2026.

Type “Mirae Asset Securities” into a screener and you get one price, one market capitalisation, one dividend yield. The Korean exchange lists the company three times. Between the last trading day of 2025 and 4 September 2026 those three lines produced returns of +47.5%, −18.8% and −3.1% — on the same earnings per share, filed in the same report, by the same company.

🏢 What Is Mirae Asset Securities, and What Does It Actually Sell?

Mirae Asset Securities is Korea’s largest securities firm by shareholders’ equity — $11.90B of owners’ capital sitting on $145.32B of total assets — and it sells four things: brokerage and wealth management to Korean retail savers, investment banking to Korean corporates, market-making and trading, and investments made with its own balance sheet.

The legal entity behind the ticker is a Seoul-headquartered financial investment company regulated under Korea’s Financial Investment Services and Capital Markets Act. It runs brokerage, wealth management, investment banking, trading and principal investment, and it does so through 151 consolidated subsidiaries and operations in 11 overseas regions (the filing counts Hong Kong, a Chinese special administrative region, among them) (Source: DART half-year report, as of 30 Jun 2026). The parent company employed 3,330 people at 30 June 2026, with average tenure of 15.63 years.

The scale that matters to a foreign reader is client money. At 31 December 2025 the parent held ₩518.1tn ($384.9B) of client assets, including ₩52.0tn ($38.6B) of foreign equities and ₩57.8tn ($42.9B) of pension assets (Source: DART FY2025 annual report, filed 16 Mar 2026). For an English-language anchor on the corporate entity itself, see Mirae Asset Securities on Wikipedia; every figure in this article comes from the filings, not from there.

📊 Where Does the Revenue Actually Come From?

Trading dominates the revenue line and produces almost none of the profit: Sales & Trading generated 85.3% of first-half operating revenue but 10.2% of operating profit, while principal investment generated 10.9% of revenue and 73.0% of profit.

Korean filers publish a segment table that no English press release reproduces. Here it is, converted.

Source: DART half-year report (반기보고서), filed 14 August 2026, note 6 (Segment Information) to the consolidated interim statements, which is the audited schedule. Read the comparatives with care: the same report also prints an H1 2025 segment table in its management-discussion section, and the two do not agree — that one shows Sales & Trading at ₩367,965M and Principal Investment at ₩116,752M where note 6 shows ₩350,026M and ₩134,668M. Both sum to the same ₩846,629M total, so no arithmetic check catches the difference. We use note 6 throughout. The current-half figures are identical in both tables. FX: ₩1,345.99/$1 as of 4 September 2026, applied to every figure in this article.

Three things follow. First, Sales & Trading revenue is a gross number, not a business size — a securities firm books the full proceeds of financial-instrument transactions as revenue and the offsetting costs as expense, so $22.84B of revenue produced $292M of profit. Do not read it as a market share.

Second, the retail-facing business is real and it is growing: wealth-management operating profit rose from $324M to $850M year on year, and it is the only segment whose profit contribution is close to what a foreign investor probably assumes a brokerage earns.

Third, investment banking went from $41M of operating profit to a $1.8M operating loss on essentially flat revenue. That is not a rounding error in a franchise that underwrote 21 Korean IPOs in 2025 (Source: DART FY2025 annual report, investment-banking review); it is a segment that stopped contributing while everyone was reading the headline profit number.

🗓️ How Did a 1970 Brokerage End Up Here?

The listed entity is not the company whose name it carries: it was founded in 1970 as Tong Yang Securities, became Daewoo Securities in 1983, and absorbed the former Mirae Asset Securities Co., Ltd. in December 2016 before taking that name in 2021.

Source: DART FY2025 annual report, “회사의 개요” and consolidated-statement note 1; corporate-history table, sections 2 and 9.

The 2016 direction of travel matters for anyone pulling historical data. Financial databases sometimes carry a pre-2017 “Mirae Asset Securities” that is a different, now-dissolved legal entity. The continuous price history under 006800 belongs to Daewoo Securities.

🌍 Is It Actually Korea’s Number One Broker?

It was the largest equity-brokerage-commission earner in Korea in both 2024 and 2025 with a 10.2–10.3% share, but it slipped to second place in the first quarter of 2026, when KB Securities took 11.82% against its 10.00%.

Source: DART half-year report, reproducing the Financial Supervisory Service’s Financial Statistics Information System. Share of industry equity-brokerage commissions.

“Largest brokerage in Korea” is a claim usually made on equity capital, and on that measure it holds comfortably. On the business that most readers picture when they hear “brokerage” — taking commissions on share trades — the lead was four hundredths of a percentage point in FY2025 — 10.33% against KB Securities' 10.29% — and it changed hands in the most recent quarter reported.

📈 Is the Business Getting Better or Worse?

Consolidated profit has quadrupled since 2023 while the parent company’s own profit has gone sideways — parent-only net income was $442M in 2025 against $519M in 2024, even as consolidated profit attributable to owners rose 70.3%.

Source: DART FY2025 annual report and H1 2026 half-year report. *The half-year ROE is the company’s own figure, annualised by the company, and it is stated as such in the filing. Parent-only (“별도”) figures are the standalone accounts of the listed entity, excluding subsidiaries.

Read the last two rows together. In FY2025 the consolidated result attributable to owners rose 70.3% to $1.17B while the standalone company earned less than the year before. In the first half of 2026, parent-only profit of $635M represents 29.5% of the $2.15B consolidated result — meaning 70.5% of the group’s reported profit was generated outside the listed company itself.

What sits in that gap is a set of consolidated investment vehicles, and we traced them in a separate piece: a Korean broker’s Delaware fund structure and what it holds. This article does not repeat that work; the point here is narrower and mechanical. An earnings figure produced by revaluing assets is not the same kind of number as one produced by charging commissions, and Korean segment reporting tells you which is which if you open the table.

🏷️ Which Mirae Asset Ticker Are You Actually Buying?

There are three: 006800 (common, ₩34,450), 006805 (first preference, ₩11,930) and 00680K (second preference B, ₩10,900) — the filing prints the same earnings per share for the common and the second preference line, and that second line trades 68.4% below the common.

Prices, listed share counts and market values: KRX Open API settled closes, 4 September 2026; 2026-to-date measured from the 30 December 2025 settled close (₩23,350 / ₩14,700 / ₩11,250). Dividends per share and stock dividends for FY2025, and both preference-share charters: DART half-year report, dividend section — which restates the FY2025 table on the finally determined stock-dividend count and supersedes the provisional version printed in the March annual report. “0.0073206 com.” means 0.0073206 common shares received per share held: the stock dividend was paid in common shares to all three classes. Yields are our calculation — cash DPS, plus the stock dividend valued at the common’s 4 September 2026 close of ₩34,450, each divided by that line’s own 4 September close.

The company’s own income statement reports the same earnings per share for the common and the second preference line: ₩2,680 for FY2025, and ₩4,951 basic for the first half of 2026. That is not an oversight. It follows from the charter, and the charter is only published in Korean.

006805 (the 1983 line) is governed by the old articles, and the clause is more specific than it first looks. It reads: “우선주식에 대하여는 액면금액을 기준으로 보통주식에 대한 배당률의 100분의 10을 금전으로 추가배당한다” — an extra dividend of one tenth of the common’s dividend rate, applied to par value. It is not one tenth of par. The common’s ₩300 on ₩5,000 of par is a 6.0% rate; a tenth of that is 0.6%; 0.6% of ₩5,000 is ₩30. Hence ₩330. The rule reproduces itself across three years without exception: ₩150→₩165, ₩250→₩275, ₩300→₩330 — every one of them exactly 1.10×.

00680K (the 2018 line) is a “new-type” participating preference share. The charter guarantees a minimum preferential dividend of 2.4% of the ₩5,000 par value — ₩120 a share — and then states that where the common dividend rate exceeds the preferential rate, the preference shares participate in the excess at the same rate as the common. In FY2025 the common’s ₩300 exceeded the ₩120 floor, so the second preference received ₩300 as well.

There is a third distribution that a screener will not show you at all. Alongside the FY2025 cash dividend the company paid a stock dividend of 4,250,472 common shares, issued on 24 March 2026 at ₩5,000 par, ₩21,252,360,000 in aggregate — a 0.6% increase in the common share count (Source: DART half-year report, capital-increase schedule as of 30 June 2026). Every class received it: 0.0073206 shares for the common and the second preference, 0.0080527 for the first preference — that ratio, again, exactly 1.10×.

Because the new shares are common shares, the stock dividend is worth roughly three times as much to a preference holder as to a common holder. A holder of 00680K received 0.0073206 common shares for each ₩10,900 share held; valued at the common’s ₩34,450 that is ₩252 on ₩10,900, or 2.31%. The identical entitlement is worth 0.73% to a common holder. Adding cash and stock together at 4 September 2026 prices, the FY2025 distribution was worth 1.60% on the common, 5.09% on the first preference and 5.07% on the second — a gap that the cash dividend alone understates rather than overstates.

Neither line votes, except that both regain voting rights for the period between a general meeting that resolves not to pay the preferential dividend and the meeting that resolves to resume it. Neither is convertible into common and neither is redeemable; the filing records “전환권자 없음” and “상환권자 없음” — no conversion holder, no redemption holder.

Liquidity is the part of the discount you can actually measure, and the filings put a number on it. At 30 June 2026 the company held 32.40% of the first preference line in treasury (4,560,260 of 14,075,750 shares) against 22.01% of the common (123,156,809 of 559,566,880) and 4.43% of the second preference (Source: DART half-year report, treasury-share schedule as of 30 June 2026). What reaches the tape follows: on 4 September 2026 the common turned over $27.2M, the second preference $1.42M and the first preference $0.41M (Source: KRX Open API, same session). A line that trades four hundred thousand dollars a day is not a line an institution can build a position in, and that is a real reason for a discount rather than a mispricing.

So a reader is looking at three claims on the same earnings stream, differing in voting rights and liquidity, priced 65–68% apart. Korean preference shares normally trade at a discount — that is a well-worn feature of the Korea Discount. What is unusual here is the size, and the direction of travel.

💰 What Do You Pay Today, and How Does That Compare?

Counting all listed classes, Mirae Asset Securities trades at 1.27× book against a Korean brokerage peer group at 0.43× to 0.95× — the only name in the group above book, and the only one whose preference shares trade at a discount wider than 30%.

Market values are the sum of every listed line of each issuer at the settled close of 4 September 2026 (KRX Open API). Owners’ equity is the equity attributable to owners of the parent at 30 June 2026, per each issuer’s DART half-year report (consolidated). P/B is our calculation. “Widest pref. discount” is the largest discount of any preference line to that issuer’s common on the same date; “n/a” means we found no listed preference line for that issuer in the KRX daily file.

Two cautions before anyone reads this table as a valuation verdict. First, the P/B numerators here include the preference lines and most screeners’ do not — on common shares alone Mirae Asset would read 1.17×, and a peer with a large preference float would read lower still. Second, book value at 30 June 2026 already contains the first-half revaluation gains. A book multiple computed on a marked-up book is not the same conservative measure it is for a bank.

📉 Is a Korean Brokerage a Leveraged Bet on the KOSPI?

It was not in 2026: between 31 March and 4 September the KOSPI rose 32.4% while the KRX securities-sector index fell 29.7% and all six listed brokers we measured lost value, Mirae Asset most of all at −44.1%.

Our calculation from KRX settled closes and KRX index levels. Two windows are shown deliberately: the signs disagree, and which window you pick changes the conclusion. “Not measured” means the peer’s settled daily series in our database begins on 31 March 2026, so we do not publish a full-year figure for it.

The two windows contradict each other and that is the finding, not a problem with the data. Measured from the start of 2026 the stock is up 47.5% and the sector is up 22.3%; measured from the end of the first quarter both are deeply negative while the index is up. Anyone quoting one window without the other is telling half the story.

On daily co-movement over the 107 trading-day returns we hold, the stock’s beta to the KOSPI is 0.88 with a correlation of 0.75 (R² 0.57); its beta to the securities-sector index is 1.14 with a correlation of 0.96 (R² 0.93). The second figure is partly self-referential — Mirae Asset is itself a large constituent of that index — but the contrast is still informative. Roughly 43% of this stock’s daily variance is not explained by the Korean market index at all. Whatever this security is, it is not an index proxy. For the mechanics of how flows move Korean share prices in general, see our primer on how foreign and institutional flows move Korean stocks, and on the 2026 volatility backdrop, why the KOSPI has been so volatile this year.

🌏 Is Foreign Money Buying or Selling?

Foreign investors were net buyers of Mirae Asset Securities across all three windows we measure, adding $328.8M over 60 sessions, while domestic institutions sold $100.5M and retail investors sold $273.9M.

Our own flow database, sourced from Korea Investment & Securities’ Open API. Method: each session’s net share count is valued at that session’s settled close and the results summed — an approximation, not execution prices. FX ₩1,345.99/$1.

All three windows carry the same sign, so there is no reversal to report here — but the counterparty is worth naming. Over 60 sessions, foreign buying of Mirae Asset Securities worth $328.8M was absorbed almost entirely by Korean retail selling of the same stock worth $273.9M. For scale against the sector: over the same 60 sessions foreigners added $228.6M of Korea Investment Holdings, $150.0M of Kiwoom, $52.8M of Samsung Securities and $48.7M of NH Investment. Mirae Asset took more foreign money than any other listed Korean broker over the window, while falling further than any of them. We track this weekly in the Foreign Flow Watch series.

On access: Mirae Asset Securities does not appear in our directory of Korean stocks with US-listed ADRs, which covers the names we have verified as carrying a US line; readers wanting exposure should confirm availability with their own broker and see our guide to buying Korean stocks as a foreign investor.

🏛️ Who Controls It, and What Do Minority Holders Get?

Mirae Asset Capital and its related parties hold 35.17% of the voting shares, up from 33.23% — and roughly three quarters of that 1.94-point rise came from the company cancelling its own stock rather than from buying, with the remaining quarter coming from 2,586,314 shares the group did add.

The chain is short and it is worth stating plainly. Mirae Asset Capital, an unlisted consumer-finance company with $4.71B of assets and $1.16B of equity on its second-quarter 2026 parent-only accounts, holds 190,256,629 common shares directly; with specially related persons the group reported 191,004,401 shares, or 35.17% of the 543,021,051 voting shares outstanding, on 7 August 2026. The largest shareholder of Mirae Asset Capital itself is founder Park Hyeon-joo, with 34.32% (Source: DART major-shareholding report, filed 7 Aug 2026).

The filing states the method of change as “share cancellation by the issuer and changes in the holdings of specially related persons” — both, not one. Its own change row reads +2,586,314 shares and +1.94 percentage points, against 24,064,683 fewer shares in issue. Splitting those two effects is arithmetic anyone can repeat: had the group’s holding not changed at all, the smaller denominator alone would have carried it from 33.23% to 34.70%, or +1.47 points — about 76% of the move. The remaining +0.47 points came from the extra shares.

Where those extra shares came from is worth knowing, because only some of them were bought. The filing’s detailed change schedule shows Mirae Asset Capital’s own holding rising by 2,376,984 shares in two distinct events: 1,419,313 shares received on 24 March 2026 as a stock dividend, which cost it nothing, and 957,671 shares purchased on-market across three sessions from 15 to 17 June 2026 at ₩50,620–₩52,105. The number of specially related persons in the filing also fell from 21 to 17 over the period, so the +2,586,314 is a net figure across a changing group, not a single buying programme.

The point survives the correction, in a sharper form. A buy-and-cancel programme raises the controller’s percentage without the controller spending anything, and a stock dividend hands the controller more shares without the controller spending anything either. Both ran here in the same six months, and together they account for the larger part of a 1.94-point increase in control.

One more number belongs next to the 35.17%. Treasury shares do not vote in Korea, and this company holds a great many of them: 123,156,809 common shares at 30 June 2026, or 22.01% of the common then in issue, of which 107,495,035 is the block created in the 2016 merger (Source: DART half-year report, treasury-share schedule). Strip non-voting treasury out of the base and the controlling group's effective voting weight rises from the headline 35.17% into the low-to-mid 40s. We do not print a single decimal for that figure deliberately: the holding is reported at 7 August, the treasury count at 30 June, and the two filings use different statutory scopes for what counts as treasury — combining them to one decimal place would be false precision. The direction is not in doubt; the second decimal is.

The buyback record is genuinely strong

In February 2024 the company committed to returning at least 35% of adjusted net profit over 2024–2026, including a minimum annual cancellation of 15,000,000 common and 1,000,000 second-preference shares. It has published Corporate Value-Up plans in August 2024 and June 2025 — see our explainer on Korea’s Value-Up programme. The delivery record is verifiable in the filings: 10 million common shares cancelled in March 2021, 20 million in January 2022, 10 million in February 2023, 10 million in February 2024 and again in November 2024, 3,230,674 common plus 821,518 second-preference in December 2025, 11,769,326 common plus 178,482 preference in February 2026. Running the other way over the same window, the company issued 4,250,472 new common shares as a stock dividend on 24 March 2026.

Then, in a single quarter: a ₩300bn ($222.9M) buyback decided on 17 June 2026 and executed between 18 June and 6 July, purchasing 4,637,645 common shares at an average of ₩43,125, 621,617 first-preference at ₩16,087 and 6,338,072 second-preference at ₩14,200. On 23 July the board resolved to cancel 16,545,829 common shares and 9,850,467 preference shares (621,617 first-preference and 9,228,850 second-preference), at a stated cancellation amount of $373.2M, effective 31 July 2026. The filing reports the common and preference totals separately and states no combined figure; we have not added one. It was made with no reduction of stated capital, under the proviso to Article 343(1) of the Commercial Act (Sources: buyback decision, execution report, cancellation decision). Compare the pattern with our study of Celltrion’s cancellation record, where the share count moved the other way.

🔍 The constraint English coverage misses: the profit is legally undistributable

The same buyback filing carries the arithmetic that caps how much of this company’s record profit can ever reach shareholders, and the binding line is a deduction for unrealised gains.

Korean companies may only buy back or pay dividends out of distributable profit, a statutory figure computed on the parent-only accounts. The 17 June filing publishes the calculation in full:

Source: DART material-event report (자기주식 취득 결정), filed 17 June 2026, “자기주식 취득금액 한도” schedule. Converted at ₩1,345.99/$1.

Read the bolded line against the company’s own results. The deduction for unrealised gains, $2.17B, exceeds the whole of the $746.4M distributable pool that survives it. The group reported $2.15B of consolidated profit attributable to owners in the first half of 2026, and its legal capacity to buy back stock at that moment was $512.5M.

This is a structural point, not a criticism. Korean company law deliberately keeps marks out of the pot that can be paid out; the same rule that lets a revaluation flow through the income statement stops it from flowing through to a dividend. For an investor, it means a headline profit driven by fair-value gains does not translate into shareholder returns on anything like the same scale. The FY2025 consolidated payout ratio was 11.10% for exactly this kind of reason — total cash dividends of $129.4M on $1.17B of attributable profit.

📚 Lingo Check

🎯 Why It Matters for K-Export Stars

Three of the load-bearing facts in this article — the identical earnings per share across share classes, the charter language that produces it, and the unrealised-gain deduction that caps the buyback — exist only in Korean, in documents filed with DART, and none of them appears in any English summary of this company that we could find. That is the whole thesis of this site in one company: the information is not hidden, it is just not translated.

In practice the takeaway is a narrow one. A foreign investor screening Korean financials sees one ticker per company and a market capitalisation that omits, in this case, 7.6% of the equity value — and a quoted cash yield of 0.87% where the same company’s second preference line distributed 5.07% on identical earnings per share. Before deciding whether a Korean broker is cheap, it is worth knowing how many securities the company has issued, which of them you are being quoted, and whether the profit that makes it look cheap is legally allowed to leave the building. For the wider structural version of that question, start with the Korea Discount, and for a comparable financial-sector case where the reported earnings sit largely outside the operating business, our study of Samsung Life.

Disclaimer: This article is for informational and educational purposes only and is not investment advice, a recommendation, or an offer to buy or sell any security. All figures are drawn from public filings and exchange data as of the dates stated and will change. Prices, market values and flow figures are as of the settled close of 4 September 2026 unless otherwise noted. Do your own research and consult a licensed adviser before making any investment decision.