Squeeze Histogram Study: We Replicated the 87,000-Squeeze Test on 20 Years of Data, 55,198 Fires Later
Replicating the 87,470-squeeze histogram study on Tapeboard's own daily data (55,198 squeeze fires across 609 names, 2006 to 2026), the squeeze histogram read at the last red dot called the fire day's direction 70.0% of the time and the 20-day moving average called it 76.9%, while all 16 signals readable at the fire day landed between 48.3% and 50.5% over the following 10 trading days.
TL;DR: We replicated the widely shared "87,470 squeezes" squeeze histogram study on Tapeboard's own daily data. Across 55,198 squeeze fires in 609 names over 20 years, the TTM Squeeze histogram read on the last red dot called the fire day's direction 70.0% of the time, matching the original's 69%. The 20-day moving average at the last red dot called it 76.9%, matching the original's 75%. And once the squeeze has fired, nothing we tested beat a coin flip: all 16 signals a trader can read at the fire day landed between 48.3% and 50.5% over the following 10 trading days. The edge in a squeeze is concentrated in the moment it fires, and it is gone within days.
The squeeze histogram claim we tested
In September 2026, TOS Indicators published a study asking whether the squeeze histogram can predict which way a TTM Squeeze fires. They measured every daily squeeze across 649 stocks and ETFs from 2004 to 2026, 87,470 squeezes in total, and reported four findings: the histogram read on the last red dot called the direction of the fire day 69% of the time, the 20-day moving average called it 75%, every signal they measured from the green dot landed between 49% and 51% over the following 10 days, and a squeeze that fired up was higher 10 days later only 49.5% of the time (their video and trade report carry the full claims).
Those are big numbers for a retail indicator, and the study deserves the compliment of being checked rather than repeated. So we rebuilt it end to end on our own data and our own indicator code, with every definition published and every number re-derivable from a committed script. This article is what our data says.
Our data: 55,198 squeeze fires across 609 names
- Universe: the 740 names Tapeboard hydrates for its squeeze coverage, the same names behind the squeeze leaderboard. 609 of them returned a full daily history through our data pipeline; 131 dead or renamed tickers returned nothing and are excluded.
- Window: October 9, 2006 to October 9, 2026, the full 20 years our historical candles pipeline serves, versus the original's 22 years.
- Data: 1,934,961 daily OHLCV bars, pulled through the same production endpoint the Tapeboard chart uses.
- Indicator: our own TTM Squeeze implementation, Bollinger Bands (20, 2.0) inside Keltner Channels (20, 1.5x average true range) with the standard linear-regression momentum histogram. The study runs the same indicator engine the live Tapeboard charts ship, so every figure below is measured on the exact signal a chart user sees, not on a re-implementation.
Definitions, so every percentage below is unambiguous. A red-dot day is a day the bands sit inside the channels. A squeeze episode is a run of one or more red-dot days. The fire day is the first day the squeeze turns off, the first green dot. The last red dot is the final red-dot day before the fire. Fire-day direction is the fire day's close against the last red dot's close. The post-fire outcome is the close 10 trading bars after the fire day, against the fire day's close.
We counted 55,198 squeeze fires. They split 27,455 up, 26,840 down, and 903 exactly flat, so neither signal below is riding a lopsided base rate. The median squeeze ran 5 red-dot days, the average 7.6, and the longest held for 188.
Reading the histogram at the last red dot: 70.0%
On the last red dot before the fire, a positive histogram says the squeeze fires up and a negative histogram says it fires down. Across our 54,295 directional fires, that read was right 37,987 times, 70.0%, against the original's 69%. The read works in both directions: among squeezes that fired up, 69.2% carried a positive histogram on the last red dot; among squeezes that fired down, 70.8% carried a negative one.
The signal is also era-stable on our data, which is the part that makes it worth taking seriously:
| Fire years | Squeeze fires | Histogram accuracy |
|---|---|---|
| 2006-2010 | 6,770 | 69.4% |
| 2011-2015 | 10,189 | 69.4% |
| 2016-2020 | 13,823 | 70.2% |
| 2021-2026 | 23,513 | 70.2% |
Our universe leans harder into small caps and squeeze candidates than the original's, our window is two years shorter, and our momentum window is 12 bars rather than 20. The number survives all of that.
The 20-day moving average at the last red dot: 76.9%
A simpler read beats the histogram. If the close sat above the 20-day moving average on the last red dot, the squeeze fired up; below it, down. That read called the fire day 76.9% of the time (41,737 of 54,292), against the original's 75%. This is a compression indicator doing trend-continuation work: a squeeze building above a rising 20-day average resolves upward far more often than chance, and the same holds below it.
From the green dot: 16 signals, no directional edge
The original study's sharpest finding was what happens after the fire. We tested 16 signals, every one readable at the moment a trader would enter, the fire day itself, against the direction over the following 10 trading days:
| Signal readable at the fire day | Squeeze fires | Higher 10 bars later |
|---|---|---|
| Histogram positive on the fire day | 27,591 | 50.2% |
| Histogram rising on the fire day | 28,169 | 50.0% |
| Close above the 20-day MA on the fire day | 27,501 | 50.4% |
| Squeeze fired up (close over prior close) | 27,388 | 50.3% |
| Fire-day gain over 1% | 18,875 | 49.3% |
| Fire-day loss over 1% | 19,019 | 49.8% |
| Fire-day volume over 1.5x the 20-day average | 12,452 | 48.6% |
| Fire-day close over open (green candle) | 26,715 | 50.1% |
| Histogram positive at the last red dot | 27,153 | 50.3% |
| Close above the 20-day MA at the last red dot | 27,325 | 50.4% |
| Histogram positive and rising at the last red dot | 19,964 | 50.5% |
| Histogram negative and falling at the last red dot | 19,914 | 50.2% |
| Long squeeze: 7 or more red-dot days | 21,859 | 49.8% |
| Short squeeze: 1-2 red-dot days | 16,868 | 50.4% |
| Fire-day gain over 2% | 13,082 | 48.3% |
| Fire-day loss over 2% | 13,643 | 49.4% |
Every row sits between 48.3% and 50.5%. The same histogram that calls the fire day at 70.0% calls the 10-day move at 50.3%. The 20-day MA that calls the fire day at 76.9% calls the 10-day move at 50.4%. The signal is not weak after the fire; it is absent.
The single most practical row is the fourth: a squeeze that fired up was higher 10 trading days later 50.3% of the time, against the original's 49.5%. Chasing the green candle after a squeeze fires is paying the pattern's least informative price.
Where our study differs from the original
Our replication agrees with all four of the original's headline numbers, with differences worth naming. We measured 55,198 fires against their 87,470, because our pipeline serves 20 years of history against their 22, and 609 names against their 649. Our density is lower too: 4.5 fires per symbol-year against their 6.1, partly because our universe includes names with shorter listing histories, partly because the original's 16 post-fire signals are not published in detail and ours are our own definitions, listed in full above. The one result we could not compare directly is the era stability of their 20-MA signal, which they did not publish; ours held between 76.7% and 77.2% across all four eras.
Limitations
- Survivorship. Our pipeline cannot serve history for 131 of the 740 names in our universe, delisted bankruptcies and acquired tickers among them. Squeeze fires in names that fully disappeared are undercounted, the same class of bias any long-horizon retail backtest carries.
- Universe composition. Our 609 names skew toward small caps and squeeze candidates, so they fire more often than a broad-market sample would. The 70.0% fire-day result is stable across eras on this universe; we did not test a large-cap-only universe.
- One definition per signal. Fire-day direction is close over close. Measured close over open, the histogram read runs 67.7% and the 20-day MA 73.9%, still signal, slightly weaker. The histogram result is also sensitive to the momentum window: at the 12-bar window our chart indicator uses it reads 70.0%, and at the 20-bar window common in other TTM implementations it reads 65.0%. Every variant stays well above a coin flip, but the exact percentage depends on the exact histogram.
- Daily bars only. Intraday fires are invisible at this resolution; the original study shares the limitation.
- This is a measurement of an indicator, not a trade. No stop, no slippage, no fees are in these percentages, and none of this is trading advice.
How the study was run
Every number in this article comes from one reproducible pass over Tapeboard's own historical candle data. The same TTM Squeeze indicator that powers the live charts is applied to all 609 names, day by day, across the full 20-year window. Each squeeze episode is detected and labeled with its fire day, the statistics are computed from those labeled episodes, and the resulting percentages are read directly into the tables above. The row counts, the window, the universe, and the indicator outputs are audited against the shipped chart implementation before publication, so the study measures the product a subscriber actually uses.
FAQ
Does the TTM Squeeze histogram actually predict direction?
Around the fire day, yes, and only there. On 54,295 squeeze fires across 20 years of daily data, the histogram read on the last red dot called the direction of the fire day 70.0% of the time, stable between 69.4% and 70.2% across every four-year era we measured. For the 10 trading days after the fire, the same read was right 50.3% of the time, a coin flip.
What is a squeeze fire day?
The fire day is the first day a TTM Squeeze turns off: the first day the Bollinger Bands sit back outside the Keltner Channels after one or more red-dot days. It is the day the compression resolves, and the study's whole point is that the measurable edge lives on that day, not after it.
What is the last red dot?
The final red-dot day before the fire, the last day the squeeze is still on. Reading the histogram or the 20-day moving average on that day is reading the squeeze from inside it, which is where the 70.0% and 76.9% numbers come from. Reading anything on the green dot or later is reading it from outside, where we found no edge.
Is the 20-day moving average better than the histogram for squeezes?
On our data, yes. The 20-day MA read at the last red dot called the fire day's direction 76.9% of the time against the histogram's 70.0%, and both signals measured from the fire day forward were statistically indistinguishable from a coin flip over 10 trading days. The MA is the cheaper read and the stronger one.
What is a slingshot squeeze?
A slingshot squeeze is a short squeeze setup built on the TTM Squeeze pattern: compression in the bands, fuel in the borrow, and a fire-day break in the trend direction. The name is used by the site behind the original study (SqueezeSetups), and our fire-day numbers describe the directional part of that setup; the follow-through part, per 55,198 fires, is a coin flip after entry.
If a squeeze fires up, does the stock keep going up?
No. Squeezes that fired up were higher 10 trading days later 50.3% of the time across our 27,388 up-fires. The direction of the fire day itself carries essentially no information about the next two weeks.
Source: Tapeboard squeeze-universe daily history, 609 symbols, October 9, 2006 to October 9, 2026, 55,198 squeeze fires, computed by the same TTM Squeeze indicator Tapeboard's live charts ship, applied to that history. Original study: TOS Indicators, "I Tested the Squeeze Histogram on 87,000+ Signals." Prior data study in this series: We Analyzed 30 Days of Borrow Fees.