How Often Are We Right? Auditing 1 Year of Hot 100 Predictions

Each week, we publish a ‘Final Billboard Hot 100 Predictions’ post on X ahead of the real chart: a full 100-position chart, estimated from tracked US sales, streaming, and airplay data.

This article reviews how those predictions performed across the past year (July 12, 2025 to July 4, 2026), measured against the chart Billboard ultimately published.

---

The Results

We ourselves have never been sure how accurate our predictions were in an objectively quantified way, so we went back and calculated the accuracy of a recent 52-week sample. Receipts at the bottom of the article.

---

Accuracy of the #1

Getting a #1 incorrect is the most visible miss our predictions can make. When it’s close and within margin of error, it’s the call we sweat over the most before we publish the final predictions on X.

Several times over the analyzed 52 weeks here, the top spot came down to a projected margin of only a few points; several weeks were near coin-flips, and one, in August 2025, was a practical tie, with a margin of less than 0.5%.

Olivia Dean’s “Man I Need” notably missed the top spot by less than 1% twice, according to our final predictions.

When We’ve Gotten a #1 Wrong

While our track record for #1 predictions was perfect in the last year, we had two notable misses in 2024 that some may remember.

In 2024, the two weeks with an incorrect #1 both predicted Shaboozey’s megahit “A Bar Song (Tipsy)” would be dethroned during its long reign, only for it to hold on.

On the chart dated August 31, 2024, the week Post Malone’s album ‘F-1 Trillion’ was released, our prediction called Post Malone and Morgan Wallen’s “I Had Some Help” to return to #1 by a margin of 1 single point, but “A Bar Song” stayed put. Given “I Had Some Help” already spent many weeks at #1, this was not quite as visible as the next one, and it was largely ignored by our audience.

The next one, however, was significantly more visible among our audience. We predicted Lady Gaga & Bruno Mars’ “Die With A Smile” would reach #1 by a small margin of 2 points on November 30, 2024, its first week at the summit. However, “A Bar Song” held once more, spending a then record-tying 19th week at #1.

Over the past 131 weeks (since the beginning of 2024), that works out to 98.5% accuracy for #1 predictions, with just two misses.

---

Accuracy of the top 10

Of the 520 top 10 positions predicted across the 52 weeks, 476 (91.5%) landed in the exact rank predicted. Counting membership rather than order, 514 of 520 songs placed in a predicted top 10 actually finished there. Across the entire year, six predicted top 10 songs slipped out, and each landed at either #11 or #12.

33 of the 52 weeks were fully correct, in exact order. Of the 19 weeks with an error, most were a single adjacent swap or one song sitting a place or two out of position.

Around 2 in every 3 weeks gets a perfect top 10. Each bar is a week’s tally of songs placed in the exact right rank, out of 10, shaded by that week’s grade: the perfect A+ weeks are lightest, and the shade darkens as the grade slips through A, A-, and B+.

---

Errors

Grading follows a simple positional method: a song one rank away from its actual position contributes one point of error, two ranks away contributes two, summed across the ten slots, plus a 2-point surcharge if the #1 is called wrong (the top spot matters most).

A flawless week is graded A+; a single adjacent swap (up to two points) is an A, each further two points steps the grade down through A-, B, C and D, and a week 24 or more points off would be an F.

According to this method, nothing in the last year was graded below B+.

Every week that was not fully correct is listed below.

---

The full top 100

This review covers only the top 10. Accuracy falls off sharply beyond it: songs cluster within a few points of one another, so a full top 100 figure would look very different from the numbers here.

We’re quite certain we’ve never gotten an entire top 100 correct in a single week, and leave it to separate analysis.

---

Engagement

Across the 52 weeks, the prediction posts drew a combined total of roughly 30.5 million views, 128,000 likes and 32,000 reposts. That’s an average of about 590,000 views, 2,500 likes and 600 reposts per post.

Engagement seems to track the size of the week’s chart story or fanbase.

The most-viewed post of the year — 3.6 million views for the chart dated June 13, 2026 — was the result of a tense battle for #1: Ariana Grande’s “hate that i made you love me”, third in the early projections and second in midweeks, managed to grab the #1 spot by final predictions.

The most-liked post in the last year (21,000 likes, April 4, 2026) was the week BTS’s “SWIM” reached #1.

---

Early Predictions

So far we’ve only discussed the account’s final predictions, posted right before the chart is revealed, when the data is at its most complete — so they are the easiest of the account’s forecasts to get right.

We also publish an early prediction, while much of the tracking week has yet to elapse. Early predictions are the harder, and arguably more interesting, test: we have to project missing days of data, and the early call is often made with only one or two days of sales, streaming, and airplay data.

We graded the early top 10 for 21 of the 27 2026 charts by the same method used above. (Six weeks were absent.)

In terms of accuracy, it trails the final:

Getting a #1 wrong, all but unheard of in the finals, is not rare in the earlies: the early call names the eventual chart-topper 16 times in 21 (76%), missing roughly one week in four. The final days of tracking data lift that to a perfect 21 of 21.

The earliest miss of the year came down to the points being too close to call, while later ones came down to a late move the early couldn’t yet see.

Position for position, the early places 58% of its top 10 in the exact rank compared to the final’s 95%. It is the exact ordering that the final improves on, because even the early puts 93% of its picks somewhere in the real top 10, against the final’s 99.5%.

Most early weeks still land at a B rather than an A+. The best and worst weeks below:

Despite the final being more accurate, the early is the bigger draw among our audience. Across the same 21 weeks, the early posts average 1.17 million views to the finals’ 663,000 (1.77 times as many), and out-viewed the final in 16 of the 21.

The first look at a week’s projected chart seems to spark the most debate; the final reads more as confirmation. Likes and replies lean the same way. Only reposts tip toward the finals, carried by the odd blockbuster result-reveal like BTS’s “SWIM” hitting #1.

The midweek predictions, which sit between the two in data completeness, aren’t graded here. It’s likely they would land between the early and final in accuracy.

---

Receipts

To reproduce the data here, you can check the 52 weekly posts on our X account. Search, for example: from:@talkofthecharts “final billboard hot 100 predictions” and choose ‘Latest’.