LEARN / START HERE

Start Here: Dealer Positioning in Ten Sessions

SquawkFlow's Start Here path, published 2026-08-29, is a ten-session sequence through the site's free SPX positioning tools and explainers. Each session names a few reads, one task to do on a live page, and one check to run the next trading day. It teaches how to read dealer positioning, volatility and flow, and how to keep a log that measures whether any of it predicted anything. It is not a strategy and not advice. Nothing on this site places an order.

HOW THIS WORKS

  • Ten sessions, at your own pace. A session is a trading day, because every check needs the next day's data.
  • Every task runs on a page of this site. No account, no broker, nothing to install.
  • Every session ends with a check you can only run the next trading day. That is the feature.
  • Nothing here places an order or tells you what to trade. The path teaches you to keep a log and read it.

YOUR PROGRESS

0 of 10 sessions done

Kept in this browser only. Clearing site data clears it.

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10

1. How a chain is built: volume, open interest and the multiplier

An options chain is a table of standing positions, not a price feed. Before any positioning tool means anything you need to know which column is a position and which is today's noise.

READ

On the terminal: the SPX gamma heatmap. On the heatmap, read two tiles. 0DTE volume magnets are the strikes where today's same-day volume concentrated. Overnight OI change is the net contracts opened or closed across the chain since the prior settlement, with the three strikes that moved most. Write down the date, the magnet strikes, and the three biggest open interest movers with their contract counts.

Next session: Read the overnight OI change tile again. Did any of yesterday's volume magnets show up among the strikes that gained open interest? Volume told you what traded; the change in open interest tells you what stayed open, and a position opened and closed inside one session never appears in it at all.

After this session: You can tell volume from open interest on sight, and you know that only one of them is a position.

2. Delta and gamma, the two Greeks that describe hedging

Delta says how much an option moves with the underlying. Gamma says how fast that delta changes. Together they say what the person who sold the option has to do when price moves, which is the whole reason positioning data exists.

READ

On the terminal: the SPX gamma heatmap. On the heatmap, find the strike nearest to spot with the largest gamma. Write one line explaining why a dealer who is short gamma at that strike has to buy as price rises through it and sell as it falls.

Next session: Look again. Is that strike still the largest, or did the concentration move with spot? Note which.

After this session: Gamma stops being a number about an option and becomes a statement about what the hedger has to do.

3. Who is on the other side: dealers and delta hedging

Most listed options are sold by market makers who do not want a directional bet. They hedge, and the hedging is mechanical. Which way it pushes depends on whether they are long or short gamma in aggregate.

READ

On the terminal: the SPX gamma page. Read today's regime on the gamma page. Write what dealers have to do into a one percent rally and into a one percent selloff under that regime.

Next session: After the close, compare today's high-to-low range with yesterday's. Was it wider or narrower, and is that consistent with the regime you wrote down? One day proves nothing, which is exactly why the next session starts a log.

After this session: You can state the mechanical reason positive gamma damps moves and negative gamma amplifies them, without appealing to sentiment.

4. GEX: the map of where the hedging is

Gamma exposure adds up the hedging pressure strike by strike. The result is a map with three landmarks: the call wall, the put wall and the flip. This session starts the log that every later session adds a column to.

READ

On the terminal: the SPX gamma page. Start a levels log. Today's row: the date, spot, the call wall, the put wall, the flip, and the distance in percent from spot to each of the three.

Next session: After the close, note which level price came within half a percent of, if any, and what happened in the hour after it got there.

After this session: The log exists. From here on every session is a row, and the rows are what will eventually tell you whether any of this predicts anything.

5. Walls, and what "held" actually means

A wall is the strike where the hedging is heaviest. Whether it held sounds like a simple yes or no, and it is not. A wall that price never reached held by default, so any rate that mixes those days with real tests is a number about distance rather than strength. That is why the gamma page publishes the walls and no hold rate, and why this session starts you counting your own.

READ

On the terminal: the SPX gamma page. Read what the gamma page actually publishes: the call wall, the put wall, the vol trigger and the flip, the time the snapshot was captured, the regime, the net overnight change in open interest with the strike that moved most, and a link through to the per-strike dealer gamma behind those levels. There is no hold rate beside any of them, on purpose. Write today's call wall and put wall into your log with the distance in percent from spot to each, then add one column: did price come within half a percent of the wall today, yes or no.

Next session: Over the next five sessions, count outcomes on touched days separately from days the wall was never reached. Keep both counts and the sample size beside each. On any day you are unsure what the structure did, read the session log on the same page: it is a timestamped row per snapshot describing the change, with no directional call attached.

After this session: You never again take a hold rate on trust, ours or anyone else's, without asking how far the level was, what counted as a touch, and how many sessions are in the sample. And you have started the only count whose answers to those three you know.

6. The flip and the regime line

The flip is the price where aggregate dealer gamma changes sign. Above it, hedging leans against moves; below it, hedging leans into them. This session turns the regime into a forecast you can score.

READ

On the terminal: the SPX gamma page. Before the open, write down spot relative to the flip and predict one word for the day: wider or narrower than yesterday's range. After the close, log the outcome next to the prediction.

Next session: Five sessions of predictions against outcomes. Count the hits.

After this session: A regime call is a forecast, a forecast has a hit rate, and your log now measures it instead of remembering the days it worked.

7. Expirations: 0DTE, OPEX and pinning

Positioning is not one number. Gamma that expires today behaves differently from gamma that expires in three weeks, and expiration day concentrates the difference. The heatmap splits gamma by expiry so the same-day layer is visible on its own.

READ

On the terminal: the SPX gamma heatmap. On an expiration day, note the strike with the largest same-day gamma on the heatmap and the max pain strike for that expiry from the SPX max pain page. After the close, log where SPX settled relative to each, in points. Also uses the SPX max pain page.

Next session: Repeat on the next expiration. Two data points do not make a pattern; ten start to.

After this session: You can separate the same-day gamma layer from the weeklies, and your log now measures how far the close landed from max pain, which is the only way to test the pinning claim rather than believe it.

8. Volatility: implied, realized and the rule of 16

Implied volatility is the move the options market has priced. Realized volatility is the move that happened. The gap between them is the volatility risk premium, and the rule of 16 turns an annual number into a daily one you can check against the tape.

READ

On the terminal: the VIX term structure page. Note whether the curve is in contango or backwardation, and write down the front-month and back-month levels with the date they settled; the page shows futures settlements, which arrive the next trading day. The spot VIX index is not on the page, and the front-month future runs close to it outside of stress, so divide the front month by 16 and write down the implied one-day move in percent.

Next session: After the close, compare the implied move with the close-to-close change, not the high-to-low range, which runs about twice as large. A one standard deviation band is expected to break about one day in three, so over five sessions a count near two is the arithmetic working, not a finding. The premium is a different measurement: implied running above realized over a month, which the reads above quantify.

After this session: You understand why "implied volatility is high" is a description of price rather than a trade, and you can convert a VIX print into a one standard deviation daily move in your head.

9. Flow: what the tape can and cannot tell you

Options flow is the record of what traded. Heavy volume in one contract is a fact; its direction is an inference, and for most of it two opposite inferences fit equally well. This session makes you write both before you pick one.

READ

On the terminal: the options flow page. Take the contract with the largest estimated premium on the flow page. That row is a day of aggregated volume in one contract, not a single trade, and the page says so beneath the table. Write two explanations that fit it equally well: one where it is a hedge against an existing position, one where it is a directional bet.

Next session: Note whether the underlying moved in the direction the "bet" explanation implied. Log it, and keep going until you have twenty rows.

After this session: You know that direction is not readable from aggregated contract activity, and your log will show you the hit rate of assuming it is.

10. Base rates, sample size and the Lab

The log now has at least twenty rows. This session turns it into the one thing most sources never show: a conditional rate reported next to its own base rate, with the sample size beside both. The Lab is this site doing the same thing to its own research, losers included.

READ

On the terminal: the Lab. From your log, compute three pairs: your own wall hold rate on days it was touched against the rate on all days; regime prediction hits against a coin flip; days realized exceeded implied against the count of days. Write each number with its sample size next to it. Then read how the Lab publishes every simulated trade from every engine, with the same discipline.

Next session: Does any conditional rate beat its base rate by more than the noise of twenty observations, which is roughly plus or minus twenty-two points around each rate and wider around their difference? If not, that is the finding, and it is a better one than a story about the days it worked.

After this session: You hold a touch-conditioned sample of your own. If there is a next step, it is paper trades, at least fifty logged before sizing anything, and nothing on this site will place them for you.

Limits

Dealer positioning is an assumption, not an observable. Open interest shows that a contract exists, never which side a dealer holds, so every level in this path inherits that convention. The gamma page publishes no hold rate for its walls, because the one it showed until 2026-08-31 was measured on a third-party definition of a wall and was dominated by how far that wall sat from spot. Session 5 asks you to count your own instead.

The block print tape on the dark pool page is modeled, not licensed data, and this path deliberately never sends you there. The DIX figure on that page is real.

Twenty observations put a confidence interval of roughly plus or minus twenty-two percentage points around a single rate, and wider around the difference between two rates. A log that size can describe what happened; it cannot yet establish an edge.

Nothing on this site is investment advice, a price target or a recommendation, and SquawkFlow has no order execution.