Research notes, with the sample size attached
Every number on these pages carries the window it was measured over, how many observations it rests on, and what it was compared against. Where a control group exists it is stated; where the sample is too small to support a claim, that is stated instead.
New notes appear here first. Subscribe by RSS to get them without waiting for a search engine to notice.
Claims retail traders inherit — tested
Do overnight gaps get filled?
True in aggregate, but the gaps that do fill are a third the size of the ones that don't.
What a volume spike actually tells you
Three times the volatility next hour — and a 50.5% coin flip on direction.
Do volatile days offer more?
Yes, and for a specific reason: they are also more efficient, not just wider.
The calendar effects are real and still useless
Thursday is the widest median day on SPY and QQQ. The weekday spread is a fifth of the daily one.
Before you believe a reversal story
Stop hunts and VWAP magnets are testable claims — and both are usually told without the one thing that would test them: a control group.
What actually moved Bitcoin in August 2026
The 25% week ran on crowded shorts, a US policy week and ETF money — the 'stealth QE' story was its least supported part.
Why Bitcoin rallied at 0% rates and crashed at 9% inflation
Three dials hide inside "the Fed": real yields, liquidity and the dollar — and they do not always turn together.
Why did oil soar while gold and silver crashed?
The metals crashed on money news a month before the war — and the war, which sent oil through $100, failed to rescue them.
Are gold, silver and Bitcoin rallying for the same reason?
Mostly yes — one engine, three exhausts. Gold is the clean read; each spread above it is that asset's own story.
Silver’s squeeze was real. The price still halved.
A true physical deficit and a 47% drawdown in the same year — the two live on different clocks.
The stop-hunt pattern is real. The edge is not.
Level breaks re-cross the level 73.7% of the time — and every other move re-crosses what it passed at 74.3%.
Does price really snap back to VWAP?
Below chance at ordinary distances. The magnet is real only in the most stretched tenth, and it pays no drift.
Does VWAP hold in a trend? As a ceiling, barely. As a floor, no.
Downtrend rallies into VWAP hold 51–60% of the time; uptrend pullbacks hold 45%. Neither is an entry.
What a cluster of large trades tells you
1.21x the matched control's volatility over the next hour, scaling with how many funds join — and a coin flip on direction.
Before you read any indicator
When futures actually trade
One contract, five different closing times — and trading ends at 9:30 a.m. on expiry day, not at the close.
Best time of day to buy an ETF
SPY's busiest hour is 2.21x its quietest — and it is also where the price bands are widened.
How much does SPY move in a day?
Median 0.90%, mean 1.13% — and 66% of sessions come in below the average. Monthly medians span 3.65x, so no single number survives.
Trading ETFs premarket and after hours
64.3% of SPY's after-hours volume lands in fifteen minutes, and the circuit breakers are off.
The overnight gap is mostly over before you see it
A median 60% of SPY's gap was already done at 04:01 — and all six instruments agree.
How ETFs and futures track the same market
Same basket, two chains: index arbitrage holds the futures to it, creation/redemption holds the ETF — which is why the shapes agree and the clocks don't.
SPX, SPY, ES — three prices for one index
A measurement, an ownership claim and an agreement — separated by known arithmetic, and only one of them is awake overnight. NDX, QQQ and NQ repeat the triangle.
Do SPY and QQQ move together?
89.3% of hours agree on direction — and in SPY's biggest hours, 99.8%. Disagreement lives almost entirely in the noise-sized hours.
The account is a variable too
Why a winning strategy can still blow up the account
Five straight losses at 5% risk is a 22.6% hole that needs a 29.2% climb — expectancy and survival are different problems.
The market does not know your P&L
Drawdown rules and peak profits change the trader, not the market — and the change is measurable as behavior, not psychology.
How the levels are computed
GEX: open interest vs volume
Two ways to weight the same chain. They disagreed on the sign of gamma in 20.7% of ES snapshots.
Why two platforms show different GEX
The measured disagreement rate between weighting bases, and the other knobs that move the number.
How much the levels move
The gamma flip level moves all day
Median 96.8 points of travel per session on ES. Zero sessions out of 65 where it stayed still.
Call wall and put wall: what they are
Nothing sits at either level — the mechanism is hedging density, and it moves.
Whether the levels carry information
Testing the gamma wall: the hit rate asks the wrong question
The same control that works on S/R finds nothing on the walls — and why that is a finding about the metric, not the level.
What the executed trades add
One market, many tapes
A block print, an options position and a futures move are three kinds of evidence about one underlying — the chains license inference, not sequence.
Order flow + gamma confluence
Two measurements of different things agreeing says which price matters — never when.
Every note, in one list
The grouped view above is the reading order. This is the plain index — and the RSS feed and sitemap carry the same set for anything reading this page automatically.
- Do overnight gaps get filled?
- What a volume spike actually tells you
- Do volatile days offer more?
- The calendar effects are real and still useless
- Before you believe a reversal story
- What actually moved Bitcoin in August 2026
- Why Bitcoin rallied at 0% rates and crashed at 9% inflation
- Why did oil soar while gold and silver crashed?
- Are gold, silver and Bitcoin rallying for the same reason?
- Silver’s squeeze was real. The price still halved.
- The stop-hunt pattern is real. The edge is not.
- Does price really snap back to VWAP?
- Does VWAP hold in a trend? As a ceiling, barely. As a floor, no.
- What a cluster of large trades tells you
- When futures actually trade
- Best time of day to buy an ETF
- How much does SPY move in a day?
- Trading ETFs premarket and after hours
- The overnight gap is mostly over before you see it
- How ETFs and futures track the same market
- SPX, SPY, ES — three prices for one index
- Do SPY and QQQ move together?
- Why a winning strategy can still blow up the account
- The market does not know your P&L
- GEX: open interest vs volume
- Why two platforms show different GEX
- The gamma flip level moves all day
- Call wall and put wall: what they are
- Testing the gamma wall: the hit rate asks the wrong question
- One market, many tapes
- Order flow + gamma confluence
See these levels on a live chart
Whale-sized prints, XJER and key levels on one order-flow chart.