Video summary

We Talked To An Ex-Wall Street Quant & He Told Us EVERYTHING | Basis Points

Main summary

Key takeaways

Finance

Quantitative Trading: What It Is and Why It Matters

  • Quantitative trading is presented as systematic, model-driven trading, especially on short time horizons (from days down to microseconds/seconds), rather than relying primarily on long-term fundamentals.
  • A major quant activity is market making: posting two-sided quotes to capture the bid-ask spread by buying low and selling high.
  • The host argues that when retail trades via apps, the counterparty is often a quant market maker. In this framing, quant firms help set market “rates” and improve liquidity.

Biggest Market Inefficiencies Quants Exploit (vs. Retail)

1) Exchange and venue microstructure (plus fragmentation)

  • The US market is split across multiple trading venues:
    • exchanges,
    • dark pools,
    • ATSs (alternative trading systems).
  • Different venues have different order matching and execution rules.
  • Latency and order sequencing matter (e.g., where and in what order you place orders).

2) Payment for Order Flow (PFOF)

  • Retail orders can be routed away from directly matching on an exchange and instead sent to major market makers (example mentioned: routing to Citadel Securities).

3) Flow toxicity / adversarial selection

  • Retail flow is described as “softer” / less informed.
  • Quants model expected profitability versus risk when trading against different counterparties.

Dark Pools: Definition, Benefits, and Downsides

What they are

  • Dark pools are alternative venues where order placement is more opaque—other participants can’t see order location as easily as on “lit” venues like NYSE or Nasdaq.

Why it matters

  • Lit venues propagate visible liquidity and order/trade information, which can leak “alpha” to competing quants.

Upside

  • Helps large orders execute without spooking the market (less price impact).
  • Sometimes retail may get better pricing via dark-pool routing.

Downside

  • Less transparency makes it harder to measure true execution quality and market quality, since displayed liquidity may be only a small portion of total liquidity.

Market Making at Scale: Economics and Performance

  • The guest describes early experience trading with roughly $20–$30B/day volume (for a team of ~3).
  • Profit is claimed to come from fractions of a basis point (or fractions of a cent) per trade, requiring very high volume.
  • Competition tends to tighten spreads; the host claims US large-cap bid-ask spreads have “never been tighter.”
  • This is used to justify the idea that systematic players face smaller opportunities, but also limited ability to lose systematically (in this model).

How the Event-Driven Quant Trading Pipeline Works

Data ingestion

  • Connect directly to exchanges with minimal latency by co-locating infrastructure near matching engines:
    • New Jersey (example for NYSE matching),
    • Chicago (example for CME),
    • other regions mentioned for ICE.
  • Consume market data events (e.g., trades in Tesla).

Feature construction (indicators)

  • Examples:
    • momentum (more buyers than sellers),
    • order book imbalance (bid depth vs offer depth).
  • Cross-venue / cross-market relationships:
    • e.g., WTI in Chicago vs Chinese crude,
    • “returns correlation.”

Modeling

  • Feed features into a model (could be a neural net or other model types).
  • Output is a very short-term price prediction, often expecting the market to be mostly flat, sometimes moving by around 1–2 basis points (examples given).

Quoting and execution

  • Place quotes around the model’s implied fair value:
    • buy at mid price minus a few cents,
    • sell at mid price plus a few cents.
  • As trades happen, accumulate expected value by effectively:
    • buying below fair price,
    • selling above fair price.

Risk management controls

  • Validate input data quality.
  • Prevent model outputs from “going crazy.”
  • Limit exposure so the book doesn’t become too one-sided before hedging is possible.

Crypto, Tokenization, and Exchange Design (Industry Direction)

  • The guest’s view: crypto’s core value is better financial systems, not “just blockchain.”
  • Claimed advantages of crypto/tokenized trading:
    • Radical transparency (public verification on-chain),
    • potential 24/7 trading,
    • instant settlement and reduced/eliminated delays (example narrative: tokenized assets appear immediately upon purchase),
    • faster transfers using stablecoins (“wire… in minutes” vs days).
  • Revenue model contrasts:
    • Traditional exchanges earn via data/fees,
    • brokers may earn via fees or spread (example: Robinhood with “zero commission” but spread-based revenue).
  • Future structure claim:
    • trading becomes more continuous/24-7 (example: options start times like 7:15 a.m. vs 9:30 a.m.).

Perpetual Futures vs Options (Leverage + Efficiency Argument)

Perpetual futures (“perps”) framing

  • Perps are described as simpler leverage than options:
    • A spot/options example: buying a 2-year at-the-money call has uncertain payout without fully knowing Greeks and path dependency.
    • A perp example: 3x leverage on Tesla with a +20% move is claimed to yield ~60% gain with a clearer downside path.

Why market makers prefer perps (as claimed)

  • Perps concentrate liquidity in one order book (single product/expiry).
  • Options fragment liquidity across many strikes/expiries (example scale: ~2,000 options in different forms for the S&P 500), which allegedly creates inefficiency and tighter per-option spread opportunities.
  • Named options-related market makers include:
    • Susquehanna,
    • IMC,
    • Optiver.
  • Elsewhere in the discussion, Citadel and others are referenced in the broader market making context.

Explicit Recommendations / Cautions

  • The guest does not provide a retail “buy/sell” recommendation.
  • Cautions include:
    • performance depends heavily on the time scale (market making requires engineering + low latency; longer-horizon approaches are less microstructure-dependent),
    • trading against sophisticated market makers is difficult; retail faces low odds if competing on the same microsecond edge.

Disclosures / Disclaimers

  • No explicit “not financial advice” disclaimer is visible in the provided subtitles/summary text.

Tickers, Assets, Sectors, and Instruments Mentioned

Stocks / names

  • Tesla (TSLA) (repeated)
  • Nebius (ticker not explicitly provided; mentioned as a “hot stock” and a purchase price example appears)
  • Porsche (used as an analogy; not a ticker)
  • CoreWeave, Rocket Lab (mentioned in retail chatter; tickers not explicitly listed)
  • GameStop, Palantir (“Palenter” in text), Rocket Lab (retail example list; tickers not clearly specified)

Commodities / futures

  • WTI (West Texas Intermediate) crude oil futures
  • Brent (ICE mentioned)

Index / derivatives context

  • S&P 500
  • Options
  • Perpetual futures (“perps”)

Exchanges / venues / institutions

  • NASDAQ, NYSE
  • CME, ICE
  • dark pools, ATSs, off-exchange venues
  • Robinhood, Interactive Brokers, WeBull
  • Citadel Securities (PFOF routing example)
  • Ken Griffin / Citadel (dark pool / ATS discussion)
  • XTX Markets (dark pool discussion; London firm)

Key Numbers and Concrete Figures

  • ~80% chance: claimed estimate that retail buying a US stock is matched against a quant.
  • $20–$30B/day: approximate trading volume in early role; “one person” responsible (within a ~3-person team, as described).
  • Fractions of a penny / basis point: per-trade profit magnitudes cited.
  • Up to ~10 exchanges: venue fragmentation example.
  • Order/price example:
    • $250.73 (Nebius purchase price example mentioned).
  • Prediction scale:
    • 1–2 basis points move examples.
  • Options timing:
    • options start 7:15 a.m. instead of 9:30 a.m. (CME/CBOE context).
  • Compensation / hiring rumors (as stated):
    • HRT offering up to $1,000,000 for a new grad (algo dev team) (claim).
    • Jane Street mentioned: ~$800 (cut off in subtitles; likely $800k, but not fully explicit).
  • Non-compete examples mentioned:
    • 18-month, 2-year, and rumors of 5-year.

Mentioned Presenters / Sources

Hosts / interviewers

  • “Steve and I” (last name not provided)

Guest

  • Anan (ex-quant building his own exchange)

Referenced individuals / entities

  • Amit (example name)
  • Stanley Druckenmiller (dark pool example)
  • Ken Griffin / Citadel
  • XTX Markets
  • Citadel Securities
  • Susquehanna, IMC, Optiver
  • CME, CBOE
  • NASDAQ, NYSE
  • CFTC / Mike Celig

Original video