WBTrading Labs

Independent research on US markets: what popular ideas are worth, and what the data actually shows. Tested against history and against what happens anyway, with the misses published too.

US stocks · options · macrostudiesdata notesexplainerstoolsin English and Polish

Today's markets against their own history

    Data: FRED, multpl.com (including Robert Shiller's CAPE), Yahoo Finance. On US market holidays the readings don't change. Each reading carries its own date, as some series are published days or a month later. The shaded band in each chart is the middle half of that reading's history. pp: percentage points. Baa: Moody's lowest investment-grade rating.

    Latest · Study · October 2026

    Does buying options before earnings pay? Across 459 US earnings reports, a call lost 42.8% on average at real prices.

    459 US earnings reports, 24 June to 9 October 2026. A call with its strike as far above the share price as the move the stock's options priced in lost 42.8% of what was paid, on average. In my test, each call is bought at the ask at the close before the report and sold at the bid at the next close. At the midpoint price, halfway between bid and ask, the same trades lost 9.0%, too little to tell apart from no loss at all: most of the loss is the spread.

    Wojciech Bernacki · independent research
    Read the study Follow on X
    Average return per trade on the call: at real prices −42.8% (margin of error −66.6% to −19.1%), at the midpoint price −9.0% (−42.2% to +24.3%). They are the same trades: the gap between them, about 34 percentage points, is what the bid/ask spread cost.
    WB Trading Labs · Fig. 1Average return per trade on calls bought at the ask at the close before US earnings reports and sold at the bid at the next close, with the margin of error; 459 reports, 24 June to 9 October 2026. Data: my own record of US option prices, my calculations.

    WB Trading Labs research

    Studies, data notes, explainers and tools

    Studies check a popular claim and end in a verdict, misses included. Data notes describe what the market is pricing or doing, without a verdict. Explainers cover the method. A tool lets you check a result yourself.

    Every claim tested so far, in one list →

    1. Study12 Oct 2026 Does buying options before earnings pay? Across 459 US earnings reports, a call lost 42.8% on average at real pricesCalls bought at the ask before the report and sold at the bid at the next close: −42.8% on average, −9.0% at the midpoint. Selling in the morning did no better than holding to the close. Does not hold
    2. Study8 Oct 2026 In the year after hitting their own all-time high, small US stocks lagged on average - the median stock didn't8,496 US stocks that started trading in 1998 or later, failed companies included. Small stocks a year after a record: +6.8% on average against +10.9% after any day; the median small stock +3.7% against +3.4%. On average only
    3. Study7 Oct 2026 After an all-time high, the S&P 500's next year looked like any other - for small caps it didn'tEvery all-time-high close since 1988 in four US indexes. In the large-cap indexes the year after a record looked like any other; in small caps it was weaker on average, on the edge of chance. Same, except small caps
    4. Tool7 Oct 2026 What happened after the S&P 500 hit an all-time high? Check it yourselfFour indexes, five horizons, price or with dividends; every run of records since 1988, against what followed all days.
    5. Tool6 Oct 2026 What the S&P 500 did after a new high in the 10-year Treasury yield - check it yourselfChoose one of four definitions of a high and a horizon (3 months or a year); every case since 1962, against what followed all days.
    6. Study5 Oct 2026 The 30-year Treasury yield is near 6%, as in 2000 - but heading the other wayIn 2000 yields fell to 6%; today they're climbing to it. After a new high in the 10-year yield, the next year for stocks looked like any other. CAPE, at 41, is where today does look like 2000. Only CAPE matches
    7. Explainer5 Oct 2026 Compared to what?Why a 62% win rate can be worth four points, and how every claim here is measured.
    8. Study4 Oct 2026 Analyst upgrades move stocks on the day, then stop+1.7% on the day of the note, then nothing for 45 sessions. Four claims tested on 295,000 rating changes. No lasting edge
    9. Data notenext Earnings catalysts: what options price in, and what happensImplied moves around earnings against the moves that followed, updated through the season. In preparation

    Method

    Compared to what?

    A setup that "works 62% of the time" is worth four points if the market goes up 58% of the time without it. Every claim here is measured against what happens anyway: the market, a typical company, a random day. A study sees only the data that existed on the day it is about, and then it is checked on years and companies it has not seen.

    I run every study in my own research pipeline, which uses AI. I check every number before it is published.

    WB Trading Labs · Fig. 2A simulation: 72 random price paths and the spread they end in. This is what happens anyway; an idea has to beat it.
    Wojciech Bernacki

    Who is behind it

    Wojciech Bernacki

    Founder, WB Trading Labs

    I'm an engineer and former software developer with an Executive MBA (University of Illinois and University of Warsaw), and I have spent more than 20 years in financial services: banking, leasing, pricing and credit, including as a member of a bank's management board. Today I lead data- and AI-driven sales and retention solutions at a global carmaker. Away from work, I'm a happy husband and father, and I currently live in Germany.

    I have invested in US stocks since 2019. I built this lab's research pipeline myself, with machine learning and AI. WB Trading Labs is my independent project: nothing here is for sale, and the views are my own. I invest my own money; when a post features a company I hold, the post says so.

    My method: I combine many data sources, often reaching far back in time, and use my data and AI skills and years in finance to put popular claims to the test and show what the data really says. AI is part of my toolkit: it helps me pull data, run the numbers and draft charts and text. The questions, the checks and the final word are always mine.

    X @WojciechBernPL

    New studies go out on X first.

    Follow @WojciechBernPL