Session-wide divergence: the trap
Everything above works on a short horizon: a few bars, one swing, the first minutes of a session. On the scale of a full day the rules change, and this difference breaks many traders.
ES has a trend day up. Price moves from 5500 to 5530 during the session. CVD falls all day: bar after bar has negative delta, aggressive selling keeps hitting the bids. The temptation is obvious: "Short. Buyers are exhausted. This divergence must reverse."
It does not have to reverse.
On the scale of a session, the current is not a stream but a river. Session divergence is a fact about institutional flow, not a reversal signal.
On a session scale the current can be a fund, a TWAP algorithm, or a large passive participant buying thousands of contracts through the day. Every sell market order meets a passive bid. Retail stops fire, short-term traders sell in panic, scalpers take profit, and the passive buyer takes the other side.
CVD falls because aggression is on the sell side. Price rises because every sold piece is immediately absorbed. On the 5-minute chart each bar can have red delta, while the daily candle closes green and strong. CVD shows effort. Price shows result. Result wins.
The metaphor is the same, but the scale is different. On a few bars, the current is a stream: one iceberg or one large limit order can absorb the rowers for a while, then stop. On the session scale, the current can be a river with patience and capital. You do not out-row the river.
If CVD and price diverge for hours, do not read it as "the market must reverse". Read it as: there is large passive activity on one side of the book.
What to do instead
Zoom in. Switch to a shorter timeframe and search for local divergences inside the session context. Short-term absorption and lack of participation still work on the scale of one swing. The session context only tells you what not to fight.
The scale of action must match the scale of the divergence. A five-bar divergence can produce a tactical entry. A full-session divergence describes the background flow.
Practical workflow: session start
ES, 09:30 ET, market open. On the screen: a 1-minute chart, footprint, and a CVD line under price.
The first minutes are noise. Overnight orders hit the market, spreads widen, delta jumps, and CVD draws a messy zigzag. Do not hurry. Let 3-5 candles form so the market can show its first direction.
After that, compare slopes. Where is price leaning? Where is CVD leaning?
If price and CVD both move down, that is convergence: confirmed selling pressure, nothing unusual. If price moves down while CVD flattens or turns up, the slopes separated and deserve attention.
Now classify the separation.
CVD did not confirm a new price low? That is lack of participation. Price broke, but sellers did not press.
CVD broke its own low and price did not? That is absorption. Sellers pressed harder, but result did not appear.
Both broke? Open footprint and check whether this was a real initiative move or a stop run into passive liquidity.
Next filter: delta percent of the diverging candle. Below 5% means balance and supports lack of participation. Above 10% while price stalls means strong aggression is being absorbed. Between 5% and 10% is a gray zone where footprint decides.
Last step: open the footprint of the diverging candle. Where is the volume? Where is delta? Is POC at the high, middle, or low? Are there imbalances? With practice this chain becomes quick: slopes, classification, delta percent, footprint.
Example. Three candles after the open: price loses four ticks and CVD falls with it. Convergence. Wait.
Fourth candle. Price makes a new low by one tick, but CVD makes a higher low. Price broke; CVD did not. Lack of participation.
Check it. The candle delta is only -2%, balanced. In footprint, the low has 40 contracts, no sell imbalance, and POC sits in the upper third of the candle. Main trading happened higher; the new low is empty. Nobody really wanted to sell there.
Price bounces eight ticks during the next three candles. The divergence did not guarantee a reversal, but it warned that selling into the new low was weak.
If footprint showed the opposite, trust footprint. If the low had 400 contracts, 300% sell imbalances, and POC on the minimum, that would be real initiative selling despite the CVD hint.
Do not scan for tiny divergences across the whole session. By bar 150, CVD has accumulated too much history and small separations drown in the curve. Focus on fresh areas: the first 30-60 minutes, the start of a new swing, or the approach to an important level.
Anchor CVD to the session. Each trading day starts from zero. If you anchor CVD to a week, small intraday divergences become buried under old history.
Cumulative Delta
In ATAS, add Cumulative Delta below the price chart, reset it by session, and compare its slope with 1-minute price action. When slopes separate, open the footprint of the diverging candles.
- Short-term divergence is tactical; session-wide divergence is context
- Do not fight a full-session river just because CVD and price disagree
- Workflow: slopes, classification, delta percent, footprint
- Footprint overrules CVD when it shows strong initiative at the extreme