Reading the Hedging Picture: A Trader's Field Guide

This is a hands-on guide to using SOAP2's GEX / VEX output to form a view on what forces are acting on the market, where it might be drawn or pushed, and which levels deserve attention, with particular attention to what changed since the previous snapshot. It needs no mathematics and no knowledge of the Greeks. Where a term is needed it is explained in plain words the first time.

It is written for two readers: a trader at the screen, and an AI assistant answering a trader's question through the SOAP2 AI bridge. The last section (§9) is the playbook for the second reader; every field name it uses is the real one.

Read this first. Everything below rests on one assumption: that dealers (market makers) are, on average, positioned a certain way, and hedge accordingly. SOAP2 estimates that position from open interest; it cannot see it. The levels you will derive are zones where a hedging effect is plausible, not barriers and not predictions. Nothing here is a trade signal. If you want the "why" behind the numbers, the theory primer explains it from first principles.

Related: GEX / VEX user guide (buttons and filters) · Theory primer · Dealer-positioning tutorial · AI assistant guide

1. The model in sixty seconds

When you buy an option, the firm that sold it to you usually does not want to bet on direction. It protects itself by trading the underlying (futures, ETF or shares) in the opposite direction, and it has to adjust that protection as the market moves. Those adjustments are real orders. SOAP2 estimates, from the open interest on every strike, how big they are and which way they point. Two effects matter:

• The "brake or accelerator" effect (GEX). When the net position makes dealers lean against moves (sell as price rises, buy as it falls), the market tends to be pinned and mean-reverting: a brake. When it makes them chase moves (sell as price falls, buy as it rises), moves feed on themselves: an accelerator. SOAP2 calls the first a positive-GEX regime and the second a negative-GEX regime. The line between them is the Flip level.
• The "volatility pushes the hedge" effect (VEX). Even if the price does not move, a change in implied volatility changes how much protection dealers need. A fall in volatility after a scheduled event, for example, can force them to buy back hedges. VEX measures where that effect is concentrated.

Put together: GEX tells you how the market is likely to behave when it moves; VEX tells you what a change in volatility alone could do; the strike-by-strike map tells you where.

2. The vocabulary: what each output means for a trader

The same quantities appear in the OA GEX/VEX tabs, the CBOE GEX/VEX window and the AI bridge. Field names in code are those the AI bridge returns.

Index points, not futures points. Levels are for the cash index (SPX, RUT, NDX). If you trade the futures (ES, RTY, NQ), add the current futures-to-cash difference to every level before using it.

3. The five-step reading routine

Always work in this order. Skipping step 1 is how people trade yesterday's picture.

Step 1: Is the picture fresh, and for which expiry?

• Scope. A single-day scope (0DTE, 1DTE) describes that day's hedging. The month-end, monthly OPEX and Aggregate scopes describe pressure that builds over several sessions. They can disagree, and both can be right: the near-dated one governs the next few hours, the longer-dated one the days around expiry.
• Age. Judge freshness by data_as_of (the newest trade inside the CBOE file), not by computed_at or the file timestamp. The CBOE feed is about 15 minutes delayed and its open interest is the previous session's.
• Live tab only: wait for the status line to say OI complete (all strikes) before trusting totals.
• Spot source. The picture is anchored to a spot; check whether it came from your watchlist price or CBOE's delayed quote (spot_source). A stale spot shifts every distance.

Step 2: What regime are you in?

1. Sign of Net GEX: brake or accelerator overall.
2. Where is spot relative to the Flip? Above (brake side) or below (accelerator side), and how many points away?
3. Is the Flip steep or shallow? A shallow gradient means the regime boundary is a zone, so give it a wide berth.

"Close" has no calibrated threshold in SOAP2. A practical yardstick is the day's expected move (for example from the market read's expected-move block, or the option-implied range on your own chain).

Step 3: Where are the walls?

List the top strikes and mark the call peak and put peak. Note which are above and which below spot, and their size relative to the others. One dominant strike matters far more than a flat cluster. (How to turn this into levels is in §5.)

Step 4: What changed since the previous snapshot?

This is the most informative step, and the easiest to misread. See §4.

Step 5: Add context, then decide what you would need to see to change your mind

• VEX: is a volatility event coming (data release, central-bank meeting, earnings season, OPEX)? Does the VEX sign reinforce or fight the GEX regime?
• Build-up: use TRACE to see whether a wall intensifies as expiry approaches (see Reading TRACE: A Trader's Field Guide).
• Is it happening? Hedging Flow shows whether the exposure is currently turning into real buying or selling.
• Your own trade: compare the levels with your breakevens and strikes in the Risk Graph.
• Write down the level or event that would invalidate your reading before you act on it.

4. Reading the change since the previous snapshot

Where it comes from:

• CBOE window / bridge (get_cboe_gex): every scope keeps a Latest and a Previous run. Previous is only kept when it is a distinct earlier run, that is, 30 or more minutes before the latest. A second call within half an hour overwrites Latest and keeps the older Previous. The change block gives the net changes, spot_change, flip_change, flip_appeared / flip_disappeared, elapsed_s, same_session and the list changes_in_evidence.
• OA GEX/VEX tabs: the change subchart under the main chart shows per strike how much the Call/Put exposure moved, since the previous recompute (about every 5 seconds) or, with Δ Since Last Save on, since the last saved snapshot. Auto-Recompute (History) writes the saved snapshots to disk.
• Bridge, live IB: get_market_read returns stored GEX per expiry (positioning.gex) plus a short gex_history of reduced snapshots (net GEX, Flip, Max/Min level over time).

The key distinction: mechanical change versus positioning change

Open interest, the count of contracts outstanding, is only updated once per session. So:

• Within a session, differences between snapshots are almost entirely mechanical: the price moved, implied volatility moved, or time passed. The dealers' positions in the model did not change. A Flip that "moved up" because spot rose is the same picture seen from a new price, not new positioning.
• Across sessions, differences also include positioning: open interest was reset, expiries rolled, new walls were built, old ones expired.

Always read same_session and changes_in_evidence first. If both runs used the same CBOE timestamp, the evidence list says so: nothing new came from the market, only the spot or settings differ.

Patterns and what they suggest

Rule of thumb: decompose any change into price moved (spot change), time and volatility moved (same session) and positions changed (new session). Only the last is a genuine change in who holds what.

5. From the picture to levels and targets

A level's role depends on the regime. The same strike can attract price in one regime and speed it up in another.

Turning that into targets

• In a brake regime, targets are magnets. The nearest large positive-exposure strike (or the Max GEX level) in the direction of drift is the natural pin or fade target. Fading excursions back toward it is the usual style; chasing breakouts against the brake is not.
• In an accelerator regime, targets are the next big strike beyond the trigger. If price loses the Flip, the next large strike (or the Min GEX level) in that direction is where speed is likely to continue or pause. Size smaller and widen stops: this is the regime where moves overshoot.
• The Flip is a trigger, not a target. Use it for "if price closes beyond X, the picture is different" decisions and for invalidation.
• Invalidation. Typical ones: price accepted beyond the Flip against your reading; the dominant wall's open interest disappearing at the next session's roll; a volatility event changing the VEX picture.
• Cross-check with other tools. Where a GEX level coincides with an expected-move boundary, a large open-interest strike, a volatility-surface kink or a TRACE ridge, the level has more independent support. A level supported only by one thing deserves less weight.

Test it yourself

SOAP2 stores the results (Results\{Symbol}\, and with Auto-Recompute the GEX history file) so you can check how price behaved around your levels on past days before you put money on them. Do this for your instrument and your scope; do not assume that a pattern seen on SPX 0DTE holds for monthly RUT.

6. Five situations and how to read them

A. Quiet brake day (positive GEX, spot comfortably above the Flip)

Read the nearest large positive strikes as magnets and the Flip as the line you do not want to be on the wrong side of. Expect compression; premium-selling and fade styles have the wind behind them in this regime, and breakout styles do not. Risk: the cushion is only as good as the open interest behind it; read the change panel for erosion.

B. Accelerator day (negative GEX, spot below the Flip)

Expect larger and faster moves in both directions. Do not use big strikes as places to take profit blindly; they may be where speed picks up. Smaller size, wider stops, and respect for the Flip as the level that would bring stability back.

C. Near the Flip

The regime itself is in doubt. Prefer to wait for a decisive move through the Flip, or trade with explicit invalidation on the Flip. A shallow gradient makes this more ambiguous: both regimes are weak near the boundary.

D. Expiry day and OPEX (strong pinning candidates)

Gamma concentrates near the strike on the last day, so single-day scopes (0DTE) and the monthly-OPEX scope matter most, and large strikes become the likeliest pins. Compare 0dte with opex: if a big strike coincides in both, it deserves the most weight. After an OPEX, a lot of exposure disappears at once; the next session's picture can look very different (same_session false), and that is normal.

E. Volatility-event day (data release, central bank, earnings season)

Look at VEX before and after the event. A volatility drop with positive VEX is the textbook support for the market after an event; a volatility rise with the same sign works the other way. Whether the effect shows up depends on the size of the volatility move and on GEX at the time. Treat VEX as a supporting argument, not a lead.

7. A worked example (illustrative numbers)

These numbers are made up to show the method. They are not a forecast and not from a real session.

1. Freshness and scope. Same session, one hour apart, 0DTE scope. Fine. The change is mechanical, not new positioning.
2. Regime. Net GEX positive, spot 32 points above the Flip (5,000 versus 4,968). Brake regime, but the cushion is thinner than an hour ago (it was 57 points).
3. Why it changed. Spot fell 12 points and the Flip moved up 13. The two moves compound: the cushion shrank by 25 points. With the same open interest, this is the picture re-evaluated at a lower price plus time passing.
4. Walls. A large positive strike at 5,050 (above, a magnet and soft ceiling in a brake regime) and a large negative one at 4,900 (below, an acceleration zone if reached). The Flip at 4,968 lies between spot and 4,900.
5. Reading. "Still a brake regime, weakening as price slides toward the Flip. A close below about 4,968 would put the market in accelerator territory, where the next large strike down is 4,900. Upside is capped in the brake regime by the 5,050 magnet."
6. Levels. Trigger: 4,968. Downside "air pocket" target if the trigger fails: 4,900. Upside pin/target while the brake holds: 5,050. Invalidation of the brake view: acceptance below 4,968.
7. Add context. VEX positive but fading with Net GEX; no volatility event scheduled, so it adds little. Check TRACE to see if the 5,050 wall builds into the close.

Note the sentence "weakening as price slides toward the Flip" describes the geometry, not a belief that price will fall.

8. What you cannot conclude

• Who is long or short. Open interest has no direction. The brake/accelerator labels are conventions about average dealer behaviour, and some days the convention is wrong (for instance after large institutional trades the other way).
• That price will stop at, or be pulled to, a level. Hedging pressure is one force among news, flow and positioning outside the options market.
• Absolute dollar flows from VEX. Its units are a relative scale.
• Intraday positioning changes. Open interest moves once per session, so within a day only spot, volatility and time move the picture (§4).
• Precision on the Flip. A comparison against an independent reference once put it a couple of percent of spot away. Treat it as a zone (see Limitations).
• Equal quality across feeds. The CBOE chain is delayed about 15 minutes with the previous session's open interest; the live OA tab reads open interest from IB. Expect small differences between them.
• Anything about instruments other than SPX, RUT and NDX. The OA GEX/VEX tabs cover SPX and RUT; the CBOE window and get_cboe_gex cover SPX, RUT and NDX.

9. For the AI assistant: how to answer from SOAP2's calls

This section is the knowledge base for an assistant answering "what is the market doing, and where to?" from SOAP2 data. Follow it; it also encodes the bridge's rules (facts, not verdicts; a null is never 0).

Which call for which question

Live-state calls need a Premium (or trial) licence and the AI bridge enabled; a denied call answers NOT_LICENSED. Say so rather than guessing.

The answering routine

1. Fetch with refresh=true (the default) for a current picture. If the user asked for a comparison, make sure Previous exists; if not, report CBOE_GEX_NO_PREVIOUS and say that a diff needs a run 30+ minutes after the last.
2. State freshness first: data_as_of and its age, cboe_timestamp, spot used and spot_source. After the close, data_as_of sits near 20:15 UTC for the index options; say so.
3. Regime: net_gex_sign, net_gex_bn, and the Flip versus spot using from_spot_now.flip (fall back to from_spot_at_compute and say so). If flip is null, say the range contains no sign change.
4. Map: top_strikes (sort by distance from spot), call_peak_strike, put_peak_strike, max_gex_level, min_gex_level. Give each level's role in the current regime (§5), not a fixed "support" or "resistance" label.
5. Change: report change only with elapsed_s, same_session and every item of changes_in_evidence. Decompose into spot move, time/volatility and (only if a new session) positioning (§4). Never present an intraday change as new positioning.
6. VEX: report net_vex_sign and whether it reinforces or opposes GEX. Say the VEX sign convention is the less settled one.
7. Derive, labelled as derived: trigger, magnet or acceleration targets and invalidation, as in the example (§7). Mark them as the assistant's reading of SOAP2's facts.
8. Close with limits: assumed dealer positioning, delayed data, zone-not-line; no trade recommendation.

Rules that must not be broken

• An absent or null field has no value in SOAP2: never read it as 0, never fill it in.
• Do not mix scopes (0dte, opex, aggregate), symbols, or the CBOE chain with the live-IB tab in one comparison. They are different datasets.
• Do not call refresh=true repeatedly to "get a diff": a second call inside 30 minutes does not create a Previous.
• Report fetch problems (CBOE_FETCH_FAILED, CBOE_REFRESH_BUSY) and say that the stored state is not fresh.
• For a user with little background, give a one- or two-sentence plain-language primer on gamma and the Flip before the numbers, and link the theory primer or this guide.
• Never describe a level as guaranteed support or resistance. Say what role it plays in the current regime and what would invalidate that.

A response template

Data: SPX, scope 0dte, CBOE chain as of 15:44 UTC (age 2 min), spot 5,000 (watchlist price).
Regime: Net GEX +2.4 bn (positive: brake under the usual convention). Spot is 32 points above the Flip (4,968), gradient moderate.
Map: Largest strikes: 5,050 (+), 4,900 (−). Max GEX level 5,040, min GEX level 4,880.
Change vs previous (60 min, same session): Net GEX −0.7 bn, Flip +13, spot −12. Mechanical: same open interest, so this is price and time, not new positioning.
VEX: positive (+95 mm), fading with GEX; no scheduled volatility event noted.
Reading (derived): brake regime weakening toward the Flip. Trigger 4,968; if lost, next large strike down is 4,900 (acceleration zone). While the brake holds, 5,050 is the pin/fade target. Invalidation: acceptance below 4,968.
Caveats: dealer positions are assumed, data ~15 min delayed, levels are zones, not a recommendation.

10. One-page checklist

1. Right symbol, right scope, fresh data (data_as_of), sensible spot.
2. Regime: sign of Net GEX; spot versus Flip in points; steep or shallow.
3. Walls: biggest strikes above and below, call peak, put peak, Max/Min level; their role in this regime.
4. Change: same session? what is in changes_in_evidence? price move, time/volatility, or positioning?
5. VEX and events: does volatility add to or fight the picture?
6. Context: TRACE build-up, Hedging Flow, your own breakevens.
7. Write the trigger, the target and the invalidation. Then size for the regime.
8. Remember the assumption underneath. If the day contradicts the picture, believe the day.

Go back to the main help page: SOAP2 Help Guide

⬅ Back