"Down 2.1 percent on 3.4 million shares against a 10-day average of 2.4 million, no trading pause" is an observation. "The Fed did it" is an inference, and "it bounces tomorrow" is a forecast. A journal that keeps those three in separate columns can be audited later; one that blends them cannot, because there is no way to tell afterwards which part was measured. The market supplies the thresholds that make the first column exact, and they are published numbers.

Write the number, not the adjective

"Volatile" is a conclusion. The measurable version is a price change over a stated window, a volume figure against a stated average, and whether either crossed a published threshold. Relative movement has a published measure too: FINRA's page on volatility states that a stock with a beta of 1.2 has historically moved 120 percent for every 100 percent move in a benchmark index, that a beta of 0.85 means smaller moves than the index, and that the benchmark itself is assigned a beta of 1. That same page notes a stock can carry high volatility and a low beta when its moves do not correlate with the market's, which is why the benchmark comparison belongs in the observation column and the explanation does not.

The thresholds that decide what counts as a big move

Two sets of numbers turn "big" into a yes or no. The SEC's Investor Bulletin: New Measures to Address Market Volatility gives the market-wide circuit breakers: a 7 percent decline in the S&P 500 is Level 1, 13 percent is Level 2 and 20 percent is Level 3, thresholds that replaced the older 10, 20 and 30 percent triggers on 4 February 2013. Levels 1 and 2 halt trading for 15 minutes when they hit before 3:25 p.m. and do not halt trading at or after that time; Level 3 closes the market for the remainder of the trading day.

Single stocks run on their own bands. The SEC's Investor Bulletin: New Measures to Address Market Volatility sets limit up-limit down price bands of 5, 10 or 20 percent, or the lesser of $0.15 or 75 percent, depending on the price of the stock, and a price has to stay outside its band for 15 seconds before a five-minute pause begins. Those bands phased in on 8 April 2013 for S&P 500, Russell 1000 and selected exchange-traded products, and on 8 October 2013 for the remaining National Market System securities. A journal entry that records "no pause" against a 2 percent move has recorded something checkable; a comparison on the same basis is what makes the rest of the entry mean anything.

One session, written twice

An example, using stated figures rather than a recommendation. A sector exchange-traded fund closed at $48.30 against a prior close of $49.34. Volume was 3.4 million shares against a 10-day average of 2.4 million. Its benchmark index closed 0.9 percent lower. The fund's published beta is 1.2.

The raw note: "dropped a lot on the rate news, looks bearish, might bounce."

The three-column version:

  • Observation. Close $48.30 from $49.34, a change of $1.04, or 2.1 percent. Volume 3.4 million against a 10-day average of 2.4 million, 41.7 percent above it. Both figures platform-reported. No limit up-limit down pause; 2.1 percent sits inside the 5 percent band, the narrowest percentage band the SEC bulletin lists.
  • Interpretation. At a beta of 1.2, the benchmark's 0.9 percent decline accounts for roughly 1.1 percent of the move. About 1.0 percentage point is unexplained by that relationship. A policy statement was released the same afternoon. Co-movement, not cause.
  • Uncertainty. Whether the volume came from one large order or many small ones. Whether the gap from the beta-implied move persists over the next five sessions or reverses.

The third row is the one that pays. Six weeks later the question "did the unexplained point close?" has an answer, and the answer costs nothing to look up.

Tag where the information came from

Source quality belongs in the observation column as a label, because the difference between a filing and an anonymous post is a fact about the input. The CFTC's advisory There's Nothing to Like About Scammers on Social Media sets out what the low end looks like: screenshots of winning trades, inflated account balances, new profiles with few connections, and the line that does the work — fear of losing out is a bigger motivator than winning big, so fraudsters fake urgency to get a commitment. An entry that records "screenshot, anonymous account, no filing" has priced the input honestly.

The regulated side of the market draws the same line in writing. FINRA Rule 2210(d)(1)(F), last amended on 16 August 2019, states that member communications may not predict or project performance or imply that past performance will recur, with narrow carve-outs for a hypothetical illustration of mathematical principles, an investment analysis tool meeting Rule 2214 and a research-report price target whose valuation method and risks are disclosed. A brokerage cannot put a forecast in the observation column either.

Read the journal back against the same thresholds

The retrospective pass is a count, not a grade. Take entries at least four weeks old and sort the interpretation column three ways: consistent with what followed, contradicted by it, still open. Then check the observation column against the thresholds it recorded — the 7, 13 and 20 percent market-wide levels and the 5, 10 and 20 percent single-stock bands — and see how many entries called a move large that never came within a third of any of them. That ratio is the habit's real output, and it is the same discipline that decides what a paper-trading result can support.

Source references

Primary sources used for this guide. Follow the links for the original scope, definitions and conditions.