![[flow_rotc_vs_vix_verified.png]] ### Market maker 证券市场做市商 我一直对做市商这门生意很感兴趣，也觉得用它们来给投资组合做对冲会很酷。 简单说，做市商就是同时报出买价和卖价，从中赚取价差。市场越动荡，交易通常越活跃、价差也越大，它们反而更容易赚钱。因此，与长期持有、不断承受换仓损耗的VIX ETF相比 (VXX -5%/月)，做市商有点像一种“自带利息的波动率对冲”：平时赚钱，市场大乱时可能赚得更多。 可惜实际效果没有这么完美。 公开市场上比较纯粹的做市商，主要是美国的Virtu Financial（VIRT）和欧洲起家的Flow Traders（FLOW）。更强的Citadel Securities和Jane Street都是私人公司，外人很难参与。VIRT和FLOW虽然通常受益于市场波动，但它们毕竟是公司，不是VIX。监管、管理层和资本配置都可能影响股价。比如FLOW在2024年突然暂停分红，把利润留下来扩大交易规模；VIRT也曾因客户交易信息的隔离问题被SEC起诉。所以市场大跌时，它们可能赚得更多，股价却未必刚好上涨，对冲效果并不可靠。 不过，单看这门生意本身，回报确实惊人。普通年份，FLOW和VIRT投入交易资本的回报大约在50%–70%；机会特别好的年份可以超过100%。2020年市场剧烈动荡，FLOW达到164%，VIRT也接近140%。 Citadel Securities和Jane Street没有公布可比较的资本回报率，但盈利规模更夸张。Citadel Securities在2024年取得约97亿美元交易收入和52亿美元EBITDA；Jane Street的交易收入则从2023年的约105亿美元，上升到2024年的205亿美元、2025年的396亿美元。由于它们没有公布实际投入了多少交易资本，我们不能硬算回报率；只能说，它们大概率代表了这个行业的最高水平，回报率应该超过FLOW和VIRT。 当然，高交易回报并不等于股东把钱全部拿走，因为金融业人力成本非常之昂贵。 2020年，FLOW赚了€933M交易收入，约534名员工的总人工成本达到€274M，平均每人超过€500K，其中绝大部分是奖金。一般年份没那么夸张：2025年平均每人约€281K。作为对比，2026年NIH给博士后的起薪标准是$63,480；即使把医保、雇主税费等全部算上，一个博士后的总成本大概也只有$80K–$100K，相当于金融公司好年景平均员工成本的15%–20%，一般年景的30%左右。 --- 我给自己两年时间，像博士生刚入学一样在不同实验室做Rotation，尝试不同的资产、组合和策略。 做市商拥有顶级人才、技术系统、廉价融资和跨市场杠杆，我当然不可能在基础设施上与它们相比。但如果把50%的年回报看作“Go Anywhere、Flexible Leverage”策略的一根高标尺，我希望两年以后，至少能够证明有能力接近它。 当然，我的成本低得多：不需要百万美元的年终奖，老婆可以养我两年；也不需要曼哈顿顶楼的豪华办公室，只需要占用车库里的一个边角。 --- ### Market Maker I've long been fascinated by the market-making business — and by the idea of using market makers as a portfolio hedge. The concept is simple: market makers continuously post a bid and an offer, earning the spread in between. The more volatile the market, the heavier the trading volume and the wider the spreads — so market makers often make *more* money precisely when things are in chaos. Compared to a VIX ETF like VXX, which bleeds roughly 5% a month in decay, a market maker is something like a "volatility hedge that pays you to hold it": it earns in normal times and can earn even more when markets go haywire. In practice, it's not quite that clean. The main pure-play public market makers are Virtu Financial (VIRT) in the U.S. and Flow Traders (FLOW) out of Europe. The heavier hitters — Citadel Securities and Jane Street — are private and closed off to outside investors. VIRT and FLOW do generally benefit from market volatility, but they're companies, not VIX contracts. Regulation, management decisions, and capital allocation all affect the stock price regardless of what markets are doing. FLOW abruptly suspended its dividend in 2024 to retain earnings and scale up its trading book; VIRT was sued by the SEC over concerns about information barriers around client order flow. So while their *businesses* may thrive in a selloff, their *stock prices* don't necessarily follow — which limits their usefulness as a hedge. That said, the underlying economics are remarkable. In a typical year, FLOW and VIRT earn returns of roughly 50–70% on deployed trading capital; in particularly favorable years, that can exceed 100%. During the turbulence of 2020, FLOW hit 164% and VIRT came close to 140%. Citadel Securities and Jane Street don't publish comparable return-on-capital figures, but the raw numbers are staggering. Citadel Securities reported roughly $9.7 billion in trading revenue and $5.2 billion in EBITDA in 2024. Jane Street's trading revenue went from around $10.5 billion in 2023 to $20.5 billion in 2024 and $39.6 billion in 2025. Without knowing how much trading capital they actually deploy, you can't calculate a clean return figure — but they almost certainly represent the ceiling of this industry, well above what FLOW and VIRT post. Of course, high trading returns don't mean shareholders pocket it all. Labor costs in this business are eye-watering. In 2020, FLOW earned €933 million in trading revenue. Its roughly 534 employees cost €274 million in total compensation — an average of over €500,000 per person, overwhelmingly in bonuses. A more typical year looks considerably more modest: around €281,000 per person in 2025. For context, the NIH's 2026 starting salary for postdoctoral researchers is $63,480. Factor in health insurance and employer taxes, and the all-in cost of a postdoc runs roughly $80,000–100,000 — about 15–20% of a finance firm's average cost per employee in a good year, and around 30% in a normal one. --- I've given myself two years — like a first-year PhD student rotating through different labs — to try different assets, portfolios, and strategies. Market makers have access to top talent, sophisticated systems, cheap financing, and cross-market leverage. I can't compete on infrastructure. But if a 50% annual return on trading capital is the high-water mark for a "go anywhere, flexible leverage" approach, then two years from now I'd at least like to prove I can come within striking distance. My cost structure helps: no seven-figure bonus to fund, a wife willing to carry us for two years, and no need for a glass tower in Midtown Manhattan — just a corner of the garage.