SPY at 15:30 ET on March 16, 2020 - Replaying Dealer Positioning During the COVID Crash

The standard explanation for why the COVID crash moved the way it did goes something like: "dealers were deeply short gamma, so every leg down forced more selling, so the move amplified into the close." It's a reasonable story. It's also one almost nobody had the data to verify at the minute. You can back it out of daily EOD chains if you're willing to do the BSM and aggregation work, but the minute-level picture - what did positioning look like at exactly 15:30 ET, thirty minutes before the close on the single worst day of the crash - has been out of reach.

FlashAlpha's Historical API changes that. One call, one timestamp, one real response:

curl -H "X-Api-Key: YOUR_API_KEY" \
  "https://historical.flashalpha.com/v1/exposure/summary/SPY?at=2020-03-16T15:30:00"

The Response

Here's what comes back (abbreviated - the full response includes interpretation text and hedging estimates):

{
  "symbol": "SPY",
  "underlying_price": 246.01,
  "as_of": "2020-03-16T15:30:00",
  "gamma_flip": 270.92,
  "regime": "negative_gamma",
  "exposures": {
    "net_gex": -2633970601,
    "net_dex": -169419489077,
    "net_vex": 152461756844,
    "net_chex": -13122349
  },
  "interpretation": {
    "gamma": "Dealers short gamma - moves amplified, trend following likely",
    "vanna": "Vol up = dealers buy delta - downside dampened if vol spikes",
    "charm": "Time decay pushing dealers to sell - pressure into close"
  }
}

Let's walk through what this actually means, line by line.

Spot at 246.01, Gamma Flip at 270.92

SPY traded at $246.01 at 15:30 ET. The gamma flip - the strike at which net dealer gamma crosses zero - sat at $270.92. SPY was trading 9% below the flip. That is a big gap. In normal markets the flip is usually within 1-2% of spot. On this day, spot had crashed so far below the structurally-positioned strikes that it was deep in the left tail of the gamma surface.

This matters because dealer hedging behavior flips at the flip strike. Above the flip, dealers are net long gamma and stabilize the market (they sell into rallies, buy into dips). Below the flip, dealers are net short gamma and destabilize it (they sell into dips, buy into rallies). With spot 9% below, we were deep in the destabilizing regime.

Net GEX: -$2.63B

Negative, and large in absolute terms relative to the shrunken notional of a -12% SPY. The interpretation field makes the implication explicit: "Dealers short gamma - moves amplified, trend following likely."

For calibration: on a normal positive-gamma day in 2020, net GEX for SPY typically ran +$3B to +$8B. A -$2.6B reading is a full regime inversion. The textbook claim is that moves in this regime get bigger and run longer. The closing print - SPY finished down 12% with the worst of the decline concentrated in the last hour - is exactly what that theory predicts.

Net DEX: -$169B

Delta exposure. Negative and huge. Dealers were carrying a massive short delta book - meaning for every $1 SPY fell, dealers' books accrued more short delta and they needed to sell equities to stay hedged. This is the mechanical feedback loop that makes crashes worse in negative gamma. The -$169B number isn't abstract; it's the notional of forced selling that any further downside would trigger, to first order.

Net VEX: +$152B (Vanna)

Vanna is the cross-greek between delta and vol. Positive net VEX means that when implied vol rises, dealers' delta increases, so they buy equities. On a crash day with VIX exploding from ~60 to ~83 intraday, positive VEX is actually a partial offset to the negative-gamma feedback: some of the vol spike mechanically forced dealers to buy. This is the "downside dampened if vol spikes" line in the interpretation - and it's a real, measurable counterforce that textbook "short gamma amplifies everything" descriptions usually omit.

Net CHEX: -$13M (Charm)

Charm is delta decay with time. Negative net CHEX in a negative-gamma environment adds pressure into the close - dealers' deltas decay in a direction that forces additional selling as the clock runs. Small number on this day in dollar terms (the big story is the gamma and vanna), but real, and pointed at the close. Which is when the worst of the move happened.

Putting It Together

At 15:30 ET on March 16, 2020:

• SPY was 9% below the gamma flip - deep in the destabilizing regime.
• Net GEX was -$2.6B - dealers short gamma, amplifying moves.
• Net DEX was -$169B - every $1 lower meant more forced selling.
• Net VEX was +$152B - the vol spike provided a partial offset via vanna.
• Net CHEX was -$13M - time decay tilted flows into the close.

The 30 minutes after this snapshot - 15:30 to 16:00 ET - saw SPY slide another ~2.5% into the close, finishing near $239.85. The mechanism the textbook describes is exactly the mechanism the data measures. The feedback loop was real, the VIX-driven vanna offset was real, the charm-driven into-the-close pressure was real. You can see all three in one JSON response.

Why This Matters Beyond a Cool Artifact

This isn't a trivia piece. Three things follow from the fact that this data now exists and is queryable:

1. Training Set for Crash-Regime Models

If you're training any model to detect crash-regime dealer positioning - as a risk signal, a trade entry, a macro overlay - you need labelled examples. The COVID crash is the best single training example available. Before the Historical API, you either didn't have the labels or you paid someone else's interpretation of them. Now you can pull March 9 through March 20, 2020 at minute resolution and label the features yourself.

2. Stress-Testing Strategies

Every premium-selling or GEX-conditioned strategy claims to "turn off" in extreme regimes. That claim is always tested against simulated or EOD-aggregated drawdowns. Now it's testable against the real intraday mechanics. Replay March 12-20, 2020 against your strategy's gating logic. Does it flip off when the regime inverts? Does it flip back on too quickly during the Vanna-driven bounces? Real data, no simulations.

3. Content and Teaching

The most effective way to teach dealer-positioning concepts is to replay a day the student already has intuition about. Everyone remembers the COVID crash. Few people have seen the per-minute dealer positioning during it. This is now a one-line API call away.

Try It Yourself - Other Days Worth Looking At

Each of these is one API call. The response shape is identical to live - which means you can diff today's dealer positioning against any historical moment using the same parsing code.

Related Articles

• Historical Options Analytics API - full overview
• Backtesting GEX-Regime Strategies
• Gamma Flip, Call Wall, Put Wall - Explained
• GEX Trading Guide

Upgrade to Alpha API Spec