Home / Paper Indices / Methodology

Paper Index Methodology

Paper index: tickerseer calculates this from its own weekly ratings. It is not a fund or an ETF, it cannot be bought, and we show its returns before trading costs. Weeks before October 5, 2026 were calculated after the fact, from the ratings we published at the time.

The indices: Growth · Value · Short · all three.

Universe

Each week, the eligible names are the core companies in that week's data that also got a published rating. Up to the 2026-09-14 rebalance the core was the S&P 500. From the 2026-09-21 rebalance it is the MSCI World plus the S&P 500. Names we rate only because they report earnings that week, or because they trade next to a company that does, are never eligible.

From the October 5, 2026 rebalance, we leave out stocks traded over the counter, the five-letter tickers ending in Y or F (such as SFTBY or JPHLF), from all three indices and from the equal-weight benchmark. They trade thinly, and their prices are often stale conversions of the home listing. Our price source has returned negative closes for several of them, and they made up most of the names we could not price. They also cannot be borrowed, so a Short Index holding them would describe trades nobody could place.

From the November 16, 2026 rebalance, the fundamentals behind our ratings refresh once a month rather than every week. In the weeks between refreshes, the eligible names are rated on those figures repriced to the latest close, and every repriced row is recorded with the date of the figures it started from. The selection rules are unchanged.

From the November 16, 2026 rebalance until November 17, 2026, ratings followed the news, and the eligible names were chosen from those ratings. From November 16 to November 17, 2026, when a stock moved on a news story outside its earnings, its rating took a one-tier step in the direction of the move: down after a fall, up after a rise. A step was set to last up to 20 trading sessions, or until the stock's growth estimates were refreshed. Among the stocks we track that trade at least 50,000 shares or $50 million a day on average (or, with volume unknown, are worth $10 billion or more), a move counts only when it is at least 5% in one session, 2.5 times the stock's usual daily move (7.5% when it has too little price history) and differs from the S&P 500's move that day by at least half its own size, or at least 15% over ten trading sessions and twice its usual move over that span. We record at most 25 single-session moves a day, largest companies first. A move on an earnings report or in the five trading days after it did not count. Neither did a story about a deal for or by the company, or one from Seeking Alpha, The Motley Fool, InvestorPlace or Zacks. The step applied only to stocks priced in US dollars whose earnings our calendar had listed in the 120 days before the move or since. On November 17, 2026, a large insider or Congress trade also moved a rating one tier, and the news and the trade together moved it one tier at most. On November 18, 2026, every step ended, and news moves and large trades no longer change a rating. Both feed a separate short-term read for Premium members. A stock whose growth estimates were saved before a news move carries a risk flag with the date of the move, because its rating reads today's price against those older estimates. A stored week keeps the rating it was published with. The selection rules are unchanged.

RebalanceCore namesBasis
2026-08-03497calculated after the fact
2026-08-10500calculated after the fact
2026-08-17494calculated after the fact
2026-08-24501calculated after the fact
2026-08-31503calculated after the fact
2026-09-07505calculated after the fact
2026-09-14503calculated after the fact
2026-09-211224calculated after the fact
2026-09-281225calculated after the fact
2026-10-051209live
2026-10-121208live
2026-10-191206live
2026-10-261209live
2026-11-021209live
2026-11-091211live
2026-11-161211live

Rows calculated after the fact count the core names in that week's data, from our data-refresh log. Live rows count the names in our dataset after we remove the delisted and duplicate securities on our exclusion list, so the count drops by the size of that list at launch.

Selection

Weighting and rebalancing

Every name gets the same weight, reset each week. With fewer than 10 qualifying names, each holds 1/10 of the index and the rest sits in cash, which earns nothing.

A period runs from one rebalance to the next. That is usually a week, and two weeks when we missed a rebalance. The fact sheets count periods, not weeks.

Rules versions

A rule change applies from its date forward. Weeks already calculated keep the rules they were calculated under, and the level chains through the change.

Timing and prices

Ratings go out early on Monday, before the market opens, so a rebalance enters at that Monday's close, or the next trading day's close when Monday is a holiday. It exits at the next rebalance's entry close. Prices are closes adjusted for splits and dividends, so the two long indices include dividends. We calculate each week's return once and store it. We never overwrite it. A correction adds a new row beside it.

A name with no close on the entry day or the exit day earns 0% that week, and the fact sheet shows how many there were. That includes a company that stops trading partway through the week. We do not use an earlier price as its exit, because in our price data a halted stock looks the same as a day the data simply left out. The Short Index resets each week. Prices are adjusted for dividends, so a shorted company's dividend already counts against the index. We do not model the cost of borrowing the shares.

Benchmarks

There are two: SPY and the equal-weight core, both over the same sessions as the index. The equal-weight core is what the same money would have earned, by the same rules, spread evenly over every core name with a published rating that week, whatever the rating and under the same over-the-counter rule as the indices. It is the pool all three indices pick from, so it differs from an index only in the picking and, when fewer than 10 names qualify, the cash the index holds. The Short Index is read against its inverse. SPY weights companies by size and the equal-weight core does not, so the equal-weight core separates the effect of our picks from the effect of equal weighting. A period above the equal-weight core means the index beat taking the whole pool, which with no cash held is the picking alone. A period above SPY but below the equal-weight core means the pool beat SPY and the index trailed the pool. The Short Index profits when its holdings fall, so we compare it with the inverse of both. A period above the inverse equal-weight core means the index did better than shorting the whole pool. A period above inverse SPY but below the inverse equal-weight core means the pool trailed SPY and the index did worse than shorting the pool.

The equal-weight benchmark books a name with no usable price at 0% for that week, the same way the indices do. It never drops the name and reweights the rest.

No annualised figure is shown until an index has 52 periods of history.

Downloads

The CSV downloads (weekly returns, and every constituent of every rebalance) come with the Basic and Premium plans. See plans

Corrections

We never edit or remove a stored week. If we find we misread our own data, we add a corrected row beside the original and list it here with its date and effect.