One question a day: is Bitcoin about to have a big day. We put a number on it before the day opens, and the next morning the market answers. Right or wrong, the answer stays on the board.
Not "will Bitcoin go up". Nobody calls that reliably, us included, and we print how badly we do at it. The question here is whether today is going to be a big day at all. That one can be called.
Yesterday we said 42%. Bitcoin had to swing 2.35% to count as a big day. It swung 3.35%, so it was one, and the call went on the board as wrong.
14 indicators read the same day and each one votes. No single one of them matters. What matters is how many point the same way, and today 57% did. Watch that number, because it is usually lower than you would expect: most days they disagree.
The 14 are not a secret: every one of them has an article on this site and the simulator draws them for you. What is not published is the part that took the work: how a reading becomes a score, and what agreement is allowed to be worth. That part is ours, and it stays in the building.
Four readings go in and one number comes out. Every morning they get marked against what happened, and whichever one was furthest off loses some of its say. Keep being wrong and you stop being listened to. Nobody makes that call. The scoring does.
14 indicators on the daily candle, the only one of the four with measured skill of its own.
What the perpetual market is paying to stay long. Crowded longs lean the day slightly the other way.
Sentiment from the curated stream, bounded so it can nudge and never steer.
Trend state against the 200-day. Up or down, nothing in between.
Swipe for act two
Every forecast gets marked the next morning against what Bitcoin actually did. A day that did not resolve cleanly is thrown out rather than guessed. Here is the whole record, misses included.
The number is a pool of forecasters whose weights follow their own graded record, scored every day from 2026-09-26.
The Brier score is how far our number sat from what happened, so lower is better. 0.25 is what you score by saying fifty percent every single day and never thinking about it. Threshold is the middle of the last ninety days: half of them moved more than that, half less. It moved is what the day did, high to low as a share of the open. The record is young; judge it when the count is in the dozens.
The forecast above is about whether today is a big day. This is the other half of the same question: when in the day the movement actually lands. It is two years of hourly Bitcoin prices and one more day every morning, in New York time, because that is where it shows up.
Mean absolute hourly move, New York time · 17,520 hourly prices · one day at a time, no pooling · rank correlation across all 24 hours 0.74 between the two halves · loud against flat on weekday hours t = 20.4 · computed 2026-10-08 · source CoinGecko
It is not the American session. Bitcoin is loudest from 10:00 to 13:00 in New York, every weekday, and then it goes flat. Not when New York goes home, three hours later, but when London does. You are looking at the part of the day when Europe and America are both at their desks, and it ends when Europe leaves. The dead hours are early morning, 05:00 to 09:00 New York, after Asia has gone to bed and before America is up. Friday is the odd one out and empties in the evening instead. Sunday evening is busy, because that is Monday morning in Asia. Saturday has almost no rhythm at all. Seven dials and not one, because a Saturday looks nothing like a Tuesday.
We also went looking for the hours it peaks and bottoms out, across 1512 different windows. There is nothing there. The time of day tells you how much Bitcoin moves, not which way it goes.
A forecast nobody acts on is an opinion. So four paper accounts trade this one, five thousand dollars each, Bitcoin only, and every fill is published.
Reads the Barometer at full resolution: the pool behind the volatility number with each expert's weight, the four direction forecasters, the fourteen indicators behind the composite, and the intraday clock for the day of the week it actually is. Then it decides for itself: open, rest a bracket, close or hold. It is given no rule about what any of it means, and it is told plainly that the direction call is measured near a coin flip. Long and short.
The same reading of the Barometer as StormAI, from the same snapshot at the same instant, with one thing added: it is shown its own past decisions, the thesis it wrote for each one, the falsifier it named, and how that day was graded. Nothing else differs. So the gap between the two is exactly what it is worth to a model to be able to see its own record.
The control. It copies StormAI exactly: the same instant, the same size, the same stop, the same instrument. The only thing it decides for itself is which way, and it decides that at random. It exists to isolate one question. If reading the forecast is worth anything, StormAI has to beat a coin that copies every other choice it makes.
No model at all. It trades only when the Barometer calls a STORM or SEVERE day and stands flat otherwise: a fixed rule, kept as the benchmark it was registered as. Direction comes from the price, never from the forecast; the position closes at the end of the UTC day. It answers the question that was actually asked, whether the forecast is tradeable, without a language model anywhere in the loop.
Trading Trades from the first trigger 47 to go 6.77 graded a week done 2026-11-26
It is not on a schedule. Over the last seven days the watcher looked 672 times and acted 79 of them, standing still for the other 593. Every one of those looks is recorded, because otherwise patience and a broken job look identical.
The wifi went down.
Stepped out for five minutes.
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
The STORM call is already settled for today. The range so far is 3.22% against a 2.35% threshold, or 137% of it, so the forecast says nothing about the hours left. Spot is sitting on the day low at 80783. A short there would be a direction bet, and the direction signals are noise (AUC 0.52). A breakdown bracket just below would be chasing a move that has already happened. The loud 10 to 12 New York block is over and the clock says the coming hours are quieter, which leaves little room for a fresh break to run. The band record also shows no skill so far: STORM was right on 2 of 5 graded days, and the overall Brier score is 0.2528, slightly worse than always guessing 50/50 (0.25). I am holding. If BTC prints below 79570, about 1.5% under the current low, before the UTC day ends, a downside bracket would have been the right trade.
Wrong if: a print at or below 79570 before 2026-10-09T00:00:00.000Z
I'm staying flat at $4,957.48 with nothing pending. Spot of 80,783 is the new Thursday low, so the range from 83,383 is 3.22%, which is 137% of the 2.35% threshold. The STORM call at 0.63 is already answered, and a short or bracket from here would only bet that the fall continues. Direction has no measured edge (AUC 0.52), and the loud 10:00 to 12:00 New York block has ended, with the hours ahead averaging 0.29% to 0.37%. Today's 1% stop-out on a break that reversed is the same mistake I'd risk repeating by chasing a low that has already run this far. If price prints below 79,950, about 1% under spot, before midnight UTC, then a short here would have been the right call.
Wrong if: a print at or below 79950 before 2026-10-09T00:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
I am closing the short from 82,124.59 with about $19 of paper profit. Today's range has reached 2.53%, past the 2.35% threshold, so the STORM call has already come true. That was the only part of the forecast with any claim to skill, and its live record is thin: 2 right out of 5 STORM days, and overall only about level with a coin flip. Keeping the short past this point would be a bet on direction alone, which the snapshot says is noise (AUC 0.52). It would also sit at the day low during the loudest hour of the day, where a bounce against the position is as likely as more drop. The snapshot does not show the stop or target the position was opened with, so I am deciding without them. Closing turns out to be the wrong call if BTC prints 80,500 or lower before midnight UTC, about 1% further down, which would mean the break kept running after the range was spent.
Wrong if: a print at or below 80500 before 2026-10-09T00:00:00.000Z
I'm staying flat at $4,957.48 with nothing pending. Spot of 81,327 is a new Thursday low, so the range from 83,383 is now 2.53%, or 107% of the 2.35% threshold. Today is a big day and the STORM call at 0.63 has been answered. Any trade from here is only a bet that the fall keeps going, and the snapshot gives no measured edge for that: direction has an AUC of 0.52, and the 11:00 New York hour's 0.58% average move can go either way. My record holds a warning on both sides: my 1% stop on the short was hit by ordinary loud-hour noise before price fell to where I had aimed, and the 81,300 level I named on my last three holds is now only $27 away, so chasing this low would repeat that pattern with a worse entry.
Wrong if: a print at or below 80450 before 2026-10-09T00:00:00.000Z
Mirror of the treatment's close.
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
I am holding the open 1x short of about $2,000, entered at 82,125. It was woken because its stop is within reach, but the snapshot does not give the stop level, so I cannot say exactly where it is. The direction calls are noise, so closing now would be a guess about direction and not a better decision. The difference between closing now and letting the stop fire is small, and the stop caps the loss at well under a third of equity. Today is a STORM band with 1.95% of the 2.35% threshold used, and STORM has graded only 2 of 5 right, so I am not adding size. The current hour (11:00 New York) is the loudest of the day, which means the stop or the breakdown will be tested soon in either direction. If price prints above 83,000 before 20:00 UTC, the break below failed and holding was the wrong call.
Wrong if: a print at or above 83000 before 2026-10-08T20:00:00.000Z
I'm staying flat at $4,957.48 with nothing pending, the same call I made at 14:35Z and 14:50Z. The stop-within-reach alert refers to nothing I hold. Spot is 82,634, and Thursday's range is still 81,792 to 83,383, which is 1.95% or 83% of the 2.35% threshold. A big day needs a low near 81,420 or a high near 83,720, each about 1.3% from spot, so a bracket there would only pay if a break keeps running after most of the day's range is already in. My record shows what that costs: this morning's bracket filled 0.55% past its level because the watcher only looks every fifteen minutes, the fall reversed, and the stop cost $24.76. STORM calls are right 2 of 5, and the pool's Brier is about a coin's, so I have no measured edge to try it again during the loud 11:00 New York hour.
Wrong if: a print at or below 81300 before 2026-10-09T00:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
I am keeping the $2,000 1x short I already hold (entry 82,125, spot 82,519, down $9.60). I will not close it by hand, because the watcher's stop already caps the loss, and a close made only because price is near that stop would be a direction call, which I have no edge on. The day is a STORM call and has used 83% of its 2.35% threshold during the loudest hours of the day (10 to 12 New York time). For the range to clear the threshold, price must print below about 81,420 or above about 83,750, so a break from here is a real possibility. That said, the record in this band is only 2 right out of 5, and overall it is level with a coin flip, so this is not an edge. The snapshot does not show the exact stop level, so I am letting the engine enforce it rather than guessing it. A print above 83,400 (a new day high) before the US close would mean this was a squeeze up rather than a breakdown, and keeping the short was wrong.
Wrong if: a print at or above 83400 before 2026-10-08T21:00:00.000Z
I'm staying flat at $4,957.48 with nothing pending. I hold no position or bracket, so the stop-within-reach alert that woke me has nothing to refer to. Little has changed in the 15 minutes since my last hold. Spot is 82,519 and Thursday's range is still 81,792 to 83,383, which is 1.95% or 83% of the 2.35% threshold. A big day still needs a low near 81,420 or a high near 83,720, each about 1.3% to 1.4% from spot. A bracket that far out would be a bet that a break keeps running, and the record gives that no edge: STORM calls are right 2 of 5, the pool's Brier of 0.249 is about a coin's 0.25, and today's one fill broke, reversed and stopped out for -$24.76. My falsifier is a print below 81,300 before midnight UTC, which would mean a short on the break would have paid and staying flat was the wrong call.
Wrong if: a print at or below 81300 before 2026-10-09T00:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
I hold a 1x short of about $2,000, entered at 82,125. Spot is 82,601, so the trade is down $11.59, which is 0.6% against the position and about 0.2% of equity. The snapshot does not show the stop and target on this position, so I am leaving both to the watcher rather than closing on the wake-up alert, because direction calls have no edge and an early close just locks in a loss the stop already caps. The day's range is 1.95% against a 2.35% threshold, and the STORM band (2 right of 5 graded) still expects more range. That means a print below about 81,420 or above about 83,717, and it is now 10:35 New York time, one of the three loudest hours of the day. If price prints above the day high of 83,383, the short was wrong and the extra range is coming on the upside.
Wrong if: a print at or above 83383 before 2026-10-09T00:00:00.000Z
My $2,000 short was stopped out near 82,570 for -$24.76. I'm flat at $4,957.48 with no brackets pending. Thursday has run from 81,792 to 83,383, a 1.95% range that is 83% of the 2.35% threshold, so a big day now needs a low near 81,465 or a high near 83,720. Each is about 1.4% from spot of 82,601. A new bracket at or past the day's extremes would mostly be a bet that a break keeps running. The record gives that no edge: STORM calls are right 2 of 5, the pool's Brier is about a coin's, and my one fill today overshot its level by 0.55% and then reversed. So I'm not re-arming straight after a loss in the loud block, where a move against a position is as large as one for it. If price prints below 81,300 before midnight UTC, a short-side bracket would have paid and this hold was wrong.
Wrong if: a print at or below 81300 before 2026-10-09T00:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
I'm short from 82,125, the stop is close and spot is 82,511. The snapshot doesn't show my stop level, so I can't say exactly how close. The direction call is noise, and nothing new tells me which way this goes, so closing by hand now would be a guess. I'll leave the exit to the stop I already set. Today has used 83% of its range threshold. The 10:00 to 12:00 ET block is the loudest of the day, which gives a breakdown room to resume, though it can just as easily take out the stop. STORM-band calls have been right only 2 times in 5. If price prints above 83,000, back near the day open, the short was wrong; the stop should already have closed it by then.
Wrong if: a print at or above 83000 before 2026-10-09T00:00:00.000Z
I'm keeping the $2,000 short at 1x from 81,753 with its stop near 82,570 and target near 80,445. Spot is 82,511, so the stop is only 0.07% away and the trade is down $18.54. The day ran 81,792 to 83,383 (1.95%, 83% of the 2.35% threshold), and the bounce of almost 0.9% off the low means the break I was betting on has mostly failed. Closing now would save about $1.40 against the stop, and I have no measured edge on direction (AUC 0.52) to justify stepping in early. So I'm leaving the stop as the exit I set in advance, accepting that in the 10:00 New York hour (about 0.57% average move) the watcher could fill it past 82,570. The read is wrong, and I'm stopped out, if 82,570 prints before 17:00Z, which I think is likely.
Wrong if: a print at or above 82570 before 2026-10-08T17:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
I'm keeping the existing 1x short of about $2,000. It is near breakeven at 82,150 against an 82,125 entry, and the snapshot does not show its stop or target levels. Its direction is no better than a coin, so the only reason to keep it is size. The pool calls STORM, but the band has been right on only 2 of 5 graded days. Today has already used 83% of the 2.35% threshold. The loudest three hours, 10:00 to 13:00 New York time, have just started, and that gives the break room to carry on to a new low under about 81,420. Closing now would only pay the spread to swap one coin flip for another. If price climbs back above 83,000, near the day's open and high, before the loud block ends, keeping the short was the wrong call.
Wrong if: a print at or above 83000 before 2026-10-08T17:00:00.000Z
I'm keeping the $2,000 short at 1x from 81,753, with the stop near 82,570 and the target near 80,445. Spot is 82,150, down $9.71, and the stop is 0.51% away, which is less than the 10:00 New York hour's average move of 0.57%. That move can go either way, and the loud block that just started gives no edge on direction. Thursday's range is 81,792 to 83,383, which is 1.95% or 83% of the 2.35% threshold. A low near 81,465 or a high near 83,720 would make it a big day, and this trade only pays if the break below the low resumes. Closing now would be a direction call on a 0.44% bounce, and nothing in the snapshot backs one, so I'm letting the stop and target I set decide. Watcher checks are 15 minutes apart, so a stop-out could fill past 82,570, but at about $20-25 the loss is small next to equity of $4,972.
Wrong if: a print at or above 82570 before 2026-10-08T17:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
I am keeping the existing 1x short of about $2,000, entered at 82,125. Its stop and target are not in the snapshot, so I am not changing them. Price is printing a new day low at 81,792. The day's range has used 83% of the 2.35% threshold, and the loudest hours of the clock, 10 to 12 New York time, are about to start. That gives a breakdown room to run, though the clock says nothing about which way. The STORM band has been right only 2 times in 5, and the vol pool's overall Brier score is no better than a coin, so I am not adding size or leverage on the band. The direction inputs are noise and I am not using them. If price climbs back above 82,700 before the loud block ends, the break has failed and holding the short was the wrong call.
Wrong if: a print at or above 82700 before 2026-10-08T17:00:00.000Z
The bracket's short side has filled. I'm holding $2,000 short at 1x from 81,753, which is 0.55% below the 82,200 level because the watcher only checks every fifteen minutes. The stop is near 82,570 and the target near 80,445, both measured from that fill. Spot of 81,792 is a new Thursday low, so the range is now 1.95%, or 83% of the 2.35% threshold, and a low near 81,465 would make it a big day. The loudest block (10:00 to 12:00 New York, about 0.56% an hour) starts in ten minutes. The trade was a bet on the size of the day, not on direction, which has no edge (AUC 0.52). The stop also gives back 1% just as easily in that loud block, so I'm leaving the stop and target where they are and not adding size. STORM days have been right only 2 of 5 and the pool's Brier is 0.249, about a coin's 0.25, so I don't hold this with much conviction.
Wrong if: a print at or above 82570 before 2026-10-08T17:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
Spot is at 82,358, which is today's low and $58 above the bracket armed earlier at 82,300 (below) and 83,700 (above). That bracket expires at 20:00 UTC, carries $2,000 with a 1.1% stop and a 2.2% target, and already trades this break with the move, so opening a short now would only chase it a few dollars early. The day has covered 1.24% of a 2.35% threshold. The band is STORM, but STORM calls have been right only 2 of 5 times and the overall record is level with a coin flip (Brier 0.2528 against 0.25), so I am not adding size. The New York hours from 10 to 12 local are usually the loudest and start in under two hours, which gives the existing bracket room to run if the break holds. My read is wrong, and a bigger position now would have been the better call, if price trades below 81,400 before 20:00 UTC: that would put the day's range past the threshold on a strong, clean downside expansion.
Wrong if: a print at or below 81400 before 2026-10-08T20:00:00.000Z
I'm keeping the 05:20Z bracket unchanged: $2,000 at 1x, long at 83,500 or short at 82,200, with a 1% stop and a 1.6% target, expiring 17:00Z. Another new Thursday low woke me. Spot is 82,358 and the day has run 83,383 to 82,358, which is 1.24%, or 53% of the 2.35% threshold. To count as a big day the low still has to reach about 81,420. The short level is 0.19% below spot. If it fills, the stop is near 83,020 and the target near 80,880, past that threshold level, and the bracket covers the loud 10:00 to 12:00 New York block, which starts in about 1h40m and averages about 0.56% an hour. The forecast's record is still about a coin (Brier 0.249, STORM right 2 of 5), and two brackets on Wednesday were cancelled even though their levels printed, so I'm keeping the size small rather than adding to it. The falsifier is a move back above 83,050 before 17:00Z, which would mean the break did not run.
Wrong if: a print at or above 83050 before 2026-10-08T17:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
I'm holding because the bracket armed at 03:20 UTC already covers this view. It goes short at 82300 or long at 83700, with $2,000 notional, a 1.1% stop and a 2.2% target, and runs until 20:00 UTC, which includes the loudest New York block from 10:00 to 12:00. Spot at 82389 is just $89 above the lower trigger. The day has used 51% of the 2.35% threshold, and the STORM band calls this a big day, but STORM's record is 2 of 5 and the forecast overall is about level with a coin (vsCoinflip -0.003). That record doesn't support adding size or opening a position now in the second-quietest hour of the day. The direction inputs (score 5.6, pCloseAboveOpen 0.507) are noise, so they give no reason to pick a side, and funding is a missing input, not a neutral one. If 81400 prints before 20:00 UTC, the day has cleared the threshold to the downside and a bigger or earlier short would have been the better call than waiting on the existing bracket.
Wrong if: a print at or below 81400 before 2026-10-08T20:00:00.000Z
I'm keeping the 05:20Z bracket as it is: $2,000 at 1x, long at 83,500 or short at 82,200, 1% stop, 1.6% target, expiring 17:00Z. I was woken by a new Thursday low of 82,389. That puts the range at 1.21%, 51% of the 2.35% threshold, so the day still has to reach about 81,420 to count as big. The short level is now only 0.23% below spot, still a real break of the day's low. If it fills, the stop sits near 83,020 and the target near 80,880, past the threshold level, with the loud 10:00 to 12:00 New York block still ahead. Opening a short now would just be paying up for a direction the snapshot says is noise (AUC 0.52). The STORM record (2 of 5) doesn't justify more size, and the snapshot doesn't explain why my earlier brackets were cancelled after their levels printed. I'm naming above 83,050 by 17:00Z as the falsifier: if price prints there, a short fill near 82,200 would have reversed into its stop, and the bracket's short side was the wrong call.
Wrong if: a print at or above 83050 before 2026-10-08T17:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
The price-break wake came from spot (82752) moving toward the day low (82430). It is still above the pending bracket's lower level of 82300, and that bracket ($2000, stop 1.1%, target 2.2%, open until 20:00 UTC) already expresses today's STORM call without picking a direction. The range so far is 1.16%, about half of the 2.35% threshold. The loudest New York block (10:00-12:00, 14:00-16:00 UTC) is still ahead, so there is time for the day to reach the threshold through either bracket level. The volatility record does not justify adding size: overall Brier is 0.253 against 0.25 for a coin, and STORM calls are 2 right out of 5, so a second bet on the same view would double up on a weakly measured edge. The direction score of 5.6 is noise and is not a reason to open now. I'm keeping the existing bracket unchanged; if 81400 prints before 20:00 UTC, the day went past the threshold and a larger short commitment would have been the better call.
Wrong if: a print at or below 81400 before 2026-10-08T20:00:00.000Z
I'm keeping the 05:20Z bracket as it is: $2,000 at 1x, long at 83,500 or short at 82,200, with a 1% stop and a 1.6% target, expiring 17:00Z. The day's high and low are still 83,383 and 82,430, so this price-break adds nothing about size: the range is 1.16%, 49% of the 2.35% threshold. At spot of 82,752 the short level is 0.67% below and the long level 0.90% above. The bracket's expiry covers the loudest block (10:00 to 12:00 New York, about 0.56% an hour), which starts in a little over three hours. STORM at 0.63 has been right only 2 of 5 times and the pool's Brier of 0.249 is about a coin's, so I'm not adding size or picking a side on direction, which is a 0.52 AUC coin. My hold is wrong if the price falls through 81,450 (the downside threshold) before 17:00Z, because that would mean a short opened now at the market was the better call than waiting for 82,200.
Wrong if: a print at or below 81450 before 2026-10-08T17:00:00.000Z
the treatment did not act, so there is nothing to mirror
Wrong if: None of its own. This arm has no thesis, which is the point of it.
Already holding. This arm carries one position at a time.
The forecast goes out every day whether anyone reads it or not. If you want to find out whether you can do better, the simulator runs on the same live prices with a paper account.