Listed's forecast leagues score one thing: how close your forecast of a reported figure came to the figure the company then reported. No prices are involved, and nobody decides the outcome by hand. The company's filing does. This guide follows one target from start to finish and works through the arithmetic. The league rules are the authority; this guide explains them.
All company figures below are illustrative round numbers, not any real company's results.
Why reported figures
A share price moves for many reasons, most of them unknowable. A company's revenue for a quarter that has already ended is a fact that exists before anyone knows it. Forecasting it rewards reading: the last few 10-Qs, what management said in MD&A, the company's own guidance, the season. The 10-Q settles the question, and every forecast is scored the same way.
The life of a target
1. The quarter ends. Say a company's fiscal quarter ends on June 30. The next day it becomes a target in the season's leagues, and the league page lists it with its period end and the date its 10-Q is due.
2. You forecast. You enter three figures for the quarter: revenue, diluted EPS and gross margin. The forecast page shows the company's last four reported values beside each field for reference. You can add a rationale of up to 1,000 characters. You can change your forecast as often as you like.
3. The first results filing locks it. Some weeks later, the company issues its earnings release and files it on an 8-K with item 2.02. The moment EDGAR accepts that filing, to the second, every forecast on the target locks. Your forecast now shows a rule through it, with the time it locked and the filing that locked it. Anything saved after that second does not count, even if the site had not caught up yet.
4. The 10-Q arrives. The release locked the forecasts, but it does not score them. Scoring waits for the 10-Q, whose XBRL data gives the figures as reported. Listed takes the figures as first reported in that 10-Q. If the company restates them next year, your score stays the same.
5. You are scored. Each of the three metrics gets a score from 0 to 100. Your forecast appears as a faint mark behind the reported figure, beside everyone else's.
The formula
Each metric is scored separately with the same formula:
score = 100 × max(0, 1 − |forecast − actual| / tolerance)
In words: work out how far off you were, divide by the tolerance, and take that share away from 100. Miss by nothing and you score 100. Miss by half the tolerance and you score 50. Miss by the whole tolerance or more and you score 0. The score never goes below zero.
The tolerance depends on the metric:
| Metric | Tolerance |
|---|---|
| Revenue | 10% of the absolute actual |
| Diluted EPS | 20% of the absolute actual, at least $0.05 |
| Gross or operating margin | 5 percentage points |
Revenue and EPS tolerances scale with the company, so a large company and a small one are equally hard to forecast. Margins use a fixed number of points, because a margin is already a ratio.
A worked example
An illustrative company reports these figures in its 10-Q:
| Metric | Actual | Your forecast |
|---|---|---|
| Revenue | $500 million | $480 million |
| Diluted EPS | $0.20 | $0.23 |
| Gross margin | 42.0% | 44.5% |
Revenue. The tolerance is 10% of $500 million, which is $50 million. You were $20 million off.
100 × (1 − 20 / 50) = 60
Diluted EPS. 20% of $0.20 is $0.04, which is below the $0.05 minimum, so the tolerance is $0.05. You were $0.03 off.
100 × (1 − 0.03 / 0.05) = 40
The minimum stops a company with tiny earnings per share from making every forecast a near miss. Without it, a company earning $0.01 a share would have a tolerance of a fifth of a cent.
Gross margin. You were 2.5 percentage points off, against a tolerance of 5.
100 × (1 − 2.5 / 5) = 50
Your total for this target is 60 + 40 + 50 = 150 out of 300. All three metrics scored above zero, so all three are hits.
Losses and negative numbers
The tolerance uses the absolute value of the actual, so losses work the same way. Suppose an illustrative company reports diluted EPS of −$0.50, a loss.
- The tolerance is 20% of $0.50, which is $0.10.
- A forecast of −$0.45 is $0.05 off and scores 50.
- A forecast of −$0.40 is $0.10 off and scores 0.
- A forecast of +$0.10, a profit, is $0.60 off and scores 0.
The fourth quarter
Companies file no 10-Q for the fourth quarter of their fiscal year. The 10-K reports the full year, and most companies no longer report the fourth quarter on its own. So the 10-K both locks fourth-quarter forecasts (if no earnings release came first) and supplies the year's figures, and Listed derives the quarter:
Q4 = fiscal year − Q1 − Q2 − Q3
Each figure in that sum is as first reported. With illustrative figures: a fiscal year's revenue of $2,000 million and quarters of $450 million, $480 million and $500 million give a fourth quarter of $570 million. Gross profit is derived the same way, and the fourth-quarter gross margin is the derived gross profit divided by the derived revenue.
Diluted EPS cannot be derived like this. Earnings per share do not add up across quarters, because the number of shares changes during the year. If the 10-K does not report a three-month diluted EPS figure for the fourth quarter, the EPS metric on that target is void for everyone. Nobody scores on it, and it does not count against anyone's hit rate.
Other void cases
A metric is void for everyone whenever the filings cannot support it: the 10-Q or 10-K does not report it, or actual revenue is zero, which leaves no tolerance for revenue and no margin to compute. If no 10-Q or 10-K for the quarter arrives within 120 days of the period end, the whole target is void.
Void means void. Listed never estimates an actual to rescue a metric.
Operating margin instead of gross margin
Banks, insurers and many service companies report no gross profit. For those targets the third metric is operating margin, scored with the same 5-point tolerance. The target page says which margin applies before you forecast.
From scores to a rank
Your season total is the sum of every metric you were scored on, across every target in the league. A blank metric scores 0, the same as a miss. Ties on total score go to the member whose last change to a counted forecast was saved earlier.
When the targets are scored, you can share a rank card showing your rank, your score and your best call. If you turn on a public profile, every scored call appears on your page with its error, your reasoning, and the discussion on it.
Leagues award badges only. There is no entry fee, no prize and nothing of cash value. A high rank shows that your forecasts of reported figures were close. It says nothing about investing, and nothing about next quarter.