Volume Spread Analysis in a Trading Journal: How to Find Out Which VSA Signals Pay You

The Series: Eight Episodes, One Idea Each

Every episode teaches one VSA idea step by step, then on a dated chart built from real daily bars, with a volume panel colored by each bar's multiple of its 20-session median. They are free on YouTube, in order, in the Volume Spread Analysis playlist.

Why VSA Needs a Journal More Than Most Methods

A moving-average crossover is the same trade every time. A VSA signal is not. Episode 1 opens with two quiet bars that look alike and lead to opposite moves, because VSA is read in context: a narrow, light bar is no supply after a selling climax and no demand after a rally. The judgment is the method.

That makes VSA hard to evaluate from memory, for three reasons:

• You remember the textbook ones. The spring that ripped is vivid. The four that were invalidated a week later are not, and they are the ones that decide whether springs pay you.
• Your definition drifts. “Light volume” becomes 0.9× one week and 0.6× the next, and a bar that closed up gets counted as no supply because the chart looked right. Unless the numbers are written down at the time, you cannot tell later which version of the signal you were trading.
• The context is the variable. If the background decides the meaning, the background has to be in the record, or you can never check whether it actually did.

What to Record on Every VSA Trade

None of this is on a broker statement. It has to be written down as it happens, while you still know why you are watching the trade.

The split between signals and tags is deliberate. Signals hold what you knew at entry, so trades can be compared fairly. Tags hold what happened afterward. Mixing the two lets hindsight into the comparison: trades that ran in your direction will tend to look better, because those are the ones that got their confirming bar.

The definition itself belongs in an entry checklist, so the shape check happens before the order, not after. Hold your own trades to the standard the series' chart examples were held to: if a clause of the definition fails, it is a different trade, not the same trade with a softer description.

One Trade, Logged End to End: AAPL No Supply

This trade, like every screenshot in this article, is from a demo account. Its trades were generated mechanically by rules (described in the next section), entered after the fact, and given notes dated to the sessions they describe. In your own journal, each note is stamped when you write it. Every price and volume figure below is from Polygon daily bars. AAPL is one of only two no-supply trades the rules produced, and it was chosen because it shows every stage.

• The climax. On July 31, 2026, AAPL opened 8.6% below the prior close, at $304.81, on 2.71× its 20-day median volume, made a low of $300.00 and closed at $308.91, 83% of the way up its range. The usual VSA reading of heavy volume that closes well off the low is absorption. A gap that size usually follows news, though, and episode 1's caveat applies: check the headline before reading it as pure supply and demand.
• The bounce, and a warning. The $300.00 low held. The bounce topped on August 20 at $320.28, on a wide bar that closed near its low: a sign of selling, though on below-normal volume (0.85×).
• The no-supply bar. August 25 closed down $0.44 on 0.56× median volume: 75% of the previous session's volume and 55% of the one before. Its spread was 0.84× average, so narrow, but only just. A down bar, lighter than both bars before it.
• Not a trade yet. Episode 1 is explicit: don't buy the quiet bar, wait for the confirming close. August 26 closed up, but on only 0.75× volume, which is not confirmation.

Demand arrived on September 1: a wide up bar (2.07× the average spread) on 1.31× volume, closing at 83% of its range and above the $320.28 bounce high. The trade was entered at the next open, $326.87, with the idea wrong on a close below $300.00. That is $26.87 of risk, so 2R was $380.61. Price rose as high as $345.34 but never reached it, and the time stop (15 sessions after the confirming bar) closed the trade on September 23 at $337.02: +0.38R.

Here is the question the journal exists to answer. Buying the quiet bar at the August 26 open of $310.30, against the method, would have cost $16.57 a share less, and its own 2R target of $330.90 was reached on September 11. On this one trade, waiting cost most of the result. One trade cannot say whether that is the rule or the exception. A record of the confirmed price and the unconfirmed price on every no-supply trade can.

We graded the setup B rather than A in the post-mortem, because the no-supply bar was only just narrow and the bounce top had closed weak. The post-mortem records the question, not a conclusion.

What 49 Rule-Generated VSA Trades Showed

To show what a journal does with many trades, we filled the demo account mechanically, using the series' own definitions and entry rules on the same Polygon daily bars the episodes were built from:

• Universe and dates: the 50 US stocks and ETFs already in the series' data cache (today's large caps plus index, sector, commodity and bond ETFs, so survivorship bias applies), with signals from January 2025 to September 2026.
• Signals: every signal bar had to pass every clause of its definition, plus shape checks we added on top of the taught definitions (a spring opens above support with a real lower wick and closes back above it, inside a range; an upthrust opens below resistance with a real upper wick; a test makes a lower low; a no-supply bar closes down).
• Confirmation: no trade was taken on the signal bar. Each waited, for up to 20 sessions, for a wide bar (1.2× average spread or more) on at least normal volume (1× the 20-day median, the series' yardstick), closing in the top third and up on the day (bottom third and down for an upthrust). Springs on at least normal volume first needed a test: a later bar back near the spring low on under 0.8× volume.
• Execution: entry at the next open after the confirming bar; out at 2R, on a close beyond the invalidation line (at the next open), or 15 sessions after the confirming bar; one position per ticker at a time; $250 of risk per trade.

The rules were not tuned to the results. The first version of this demo did break episode 1's main rule: it bought the signal bar itself. A review against the course caught it, and rebuilding it to wait for confirmation changed every number below That is a reason to write your rules down. This is not a backtest of VSA and not a recommendation. Forty-nine trades is a small sample, and VSA traded by a person uses judgment that rules cannot.

Four things a VSA trader would want to know from this, none of which a chart shows:

1. Waiting for confirmation makes the clean setups rare

In 20 months across 50 tickers, the full no-supply sequence ending in confirmation appeared twice, and the test never confirmed. Most spring bars either broke their line or went 20 sessions without a confirming bar. A journal tells you how often your signal actually occurs, which tells you how long it will take to learn anything about it. At two no-supply trades in 20 months on this universe, that is years.

2. The stop is a close, so losses run past 1R

The series defines invalidation as a close beyond the line (for the upthrust, a strong close on good volume), and this demo acted on each one at the next open. Fifteen trades were invalidated, and they averaged −1.50R, not −1R: 13 of the 15 lost more than the risk they planned. Had each cost exactly 1R, the whole sample would have finished at −0.30R instead of −7.78R, so the overshoot accounts for nearly all of the loss. That is not an achievable result, though: a hard stop at the line would also have cut five other trades that traded through it during the day and recovered (one of them a 2R winner), and that version finishes near −8.6R, worse than the close-based line.

JPM shows how. A spring confirmed on September 15, 2026, and the trade was entered at $353.38 the next morning, with the spring low of $348.69 as the line: $4.69 of risk. On September 16 it closed at $348.92, 23 cents above the line. On September 22 a wide down bar on 2.08× volume closed at $340.00, and the exit at the next open, $337.61, was −3.36R. XLY lost −3.25R for the other reason a close-based stop overshoots: its line was only 75 cents below a $114.61 entry, so a modest miss was several times the planned risk.

The journal exposed something else about those two trades. On both, the confirming bar had itself traded below the line during the day before closing back above it: JPM's September 15 low was $343.07, under its $348.69 line, and XLY's March 9 low was $110.99, under $113.86. The rule kept a line that the market had already broken intraday. A trader looking at the chart would probably have redrawn it, or passed. That is a flaw in the rule, found by reading the worst trades one by one.

None of this says a close-based line is wrong; it is what the method teaches, and as the hard-stop comparison shows, it kept trades that an intraday stop would have cut. It says the planned R is not the R you will realize, and that only a record of both shows by how much. What you do about it (a minimum distance to the line, redrawing the line when the confirming bar breaks it, or a hard limit some way beyond it) is a rule you can then test.

3. Springs by volume: no conclusion yet

Each trade carried a signal for its signal bar's volume. Low-volume springs (under the 20-day median), which only had to wait for confirmation, made −0.03R over 20 trades. Higher-volume springs, which waited for a test and then confirmation, lost −7.17R over 16, or −0.45R a trade. Episode 5 says a low-volume spring is the stronger form, and the totals lean that way. But the gap is JPM and XLY, the two stale-line trades above: without them, the other 14 tested springs made −0.56R, close to the low-volume springs' −0.03R. Sixteen and twenty trades cannot separate either from noise. It is a question to keep logging, not a finding.

4. Most trades never resolved

Twenty-eight of the 49 trades neither reached 2R nor broke their line within 15 sessions, and ended on the time stop, for +2.7R between them. That makes the time stop, a rule the series never sets, one of the most important rules in the sample. If your own journal shows the same, it is the rule to examine first.

Setting Up a VSA Journal in SignalDeck

1. One strategy per setup (Setup → Strategies, in the profile menu), with the definition, the confirmation rule and the exits in its notes, so the definition you trade is the definition on file.
2. Signals for what you knew at entry (Setup → Signals): the signal bar's volume class, whether a test came, where the bar sat in its range. Keep them to things you knew before the order.
3. Tags for the exit: target hit, invalidated, time exit.
4. An entry checklist with the shape checks and the confirmation, set as a hard gate in Trade Preferences if you want it enforced.
5. On each trade: the stop at the invalidation line, a note with the signal bar's numbers, a note when confirmation comes (with the price you could have had without it), and your chart as the entry and exit screenshots.

Then read the results by strategy on the Runs page, and by signal and tag under Analytics → Quant, all in R. The journal does not tag VSA bars for you, and none of the backtester's built-in strategies is a VSA strategy, so the signal on your own trades has to come from you. What it adds is a running count of how each signal you trade has actually done, without you building or maintaining the formulas. You can also connect an AI assistant such as Claude or ChatGPT to the journal and ask about your trades in plain English.

What to Do With This

Watch the series in order, starting with episode 1. Then pick the one or two signals you actually trade, write their definitions and your confirmation rule down as a checklist, and log every trade with the signal bar's numbers, the price you could have had without waiting, and the R you realized. Twenty trades is where a signal stops being an anecdote, not where it is proven; read it by background only once each background has twenty of its own. The chart teaches you the bar. The journal tells you whether it pays you.