Three bitcoin price targets crossed the wire this week. Galaxy Digital put $100,000 on the table. Fidelity put $300,000 on it. Volmex Labs put $500,000 on it. Read that list upward and it looks like a market growing more confident by the day. Read it against a calendar instead and the picture inverts, because $500,000 is not a new number. It is one of the oldest numbers in crypto, and every time its deadline has come due, the deadline has moved.
When a number survives four years, four or five authors and three quietly relocated deadlines, it has stopped being a forecast. I have watched too many teams ship a roadmap whose only living part was the date. A target like that is not a prediction about the future. It is a genre, with a fixed shape: a big round number, a named model, and a horizon set far enough out that nobody issuing it will be embarrassed this quarter. The useful question is not whether the number is right. It is what the number measures, and how it was produced.
“Pure Mathematics”
Before I argue about $300,000, I want to know exactly what was measured. Jurrien Timmer, director of global macro at Fidelity Investments, says in his latest technical report that bitcoin has finished its local bearish trend and is ready for long-term growth to $300,000 by 2029. The coverage says he backs the target with what it calls “pure mathematics.” Start with that phrase, because the phrase is doing more work than the model.
“Pure mathematics” is the coverage’s phrase for the Power Law model, which evaluates bitcoin using logarithmic linear progression and a 52-week Z-score of its ratio to gold, and which the report says shows current price fluctuations are “not random noise but regular cycles.”
U.Today, Gamza Khanzadaev, reporting Jurrien Timmer of Fidelity Investments
Every input in that sentence is bitcoin’s own past price, or a function of it. A power-law curve fitted to price history is a curve fit, not a law. You draw the line through the data you want to explain, point at the line, and call it a natural law of the asset. This tells you nothing about tomorrow. It tells you the fit of yesterday, wrapped in a confidence interval and sold as a discovery.
Calling a regression “pure mathematics” is a category error, and a revealing one. Pure mathematics is true by construction, independent of any dataset. A regression is the opposite: a summary of one dataset whose coefficients move the moment you move the sample. The price history here is measured. The curve is derived from that history. The $300,000 is asserted, and then presented as though the derivation had produced it. A model calibrated on the thing it is supposed to predict cannot confirm itself. Reporting that it fits its own calibration data is a tautology, not evidence.
“A Line in the Sand”
Timmer calls $60,000 a critical “line in the sand,” and says holding it “fully confirms the strength of the Power Law model” and opens the way to new highs, as reported in the same Fidelity report. A line in the sand is, by construction, something you can cross. So test it for falsifiability, which is the only test that matters. If $60,000 holds, the model is confirmed. If it breaks, what happens?
The coverage answers that for us. The local lows formed in the $57,742–$60,033 range. Read that range again: its bottom sits below the line. The market did not hold $60,000 in the sense a layperson would understand; it dipped under it, and the line was then redrawn as a range. A prediction that cannot be wrong is not a prediction. It is a description with a floor that moves whenever the price approaches it.
The same move sits in the momentum reading. Timmer notes that the weekly stochastic gauges, Fast %D and Slow %D, have climbed out of oversold territory, which he reads as large-scale buying resuming. But “oversold” is a state, not a direction. An oscillator that has left the bottom of its range has left the bottom of its range, and it resets. It is computed from price, so it cannot explain the price it is computed from.
“A Logarithmic Ruler”
The coverage describes Fidelity’s “logarithmic ruler that predicts growth by multiples,” and the model stacks three things together: a power-law regression, a 52-week Z-score of bitcoin against gold, and weekly stochastic oscillators. Be fair to each. All three are real tools used by real analysts. A Z-score against gold is a legitimate way to ask whether bitcoin is expensive relative to an older asset. A stochastic describes recent momentum. The problem is not that any one of them is fake. The problem is that they belong to three unrelated disciplines, and they have been bolted together to produce one number.
The claim underneath is that agreement between unrelated methods means more than one method speaking. That is only true if the methods were fixed in advance and the data was not. Here the order is reversed: the answer is a big round number, and the methods are assembled underneath it. Agreement between unrelated methods proves nothing if the methods were chosen after the answer was known. I can fit a ruler, a Z-score and an oscillator to any price series and produce a target. That tells you I can run three regressions. It does not tell you where bitcoin is going.
And a ruler, incidentally, does not predict. A ruler measures a distance that already exists. When a firm says its logarithmic ruler “predicts growth by multiples,” it borrows the vocabulary of measurement to make a claim of foreknowledge. One is a description of a chart someone has already seen. The other is a claim about a chart nobody has seen. The language trades on the first while promising the second.
“$500,000 by 2028”
Now the big one, because this is where the ledger actually matters. Cole Kennelly, founder and CEO of Volmex Labs, says bitcoin could reach as high as $500,000 within roughly two years. Bitcoin was changing hands at $84,269 at the time, according to CoinGecko, so the call needs a gain of roughly 540% — a bit more than sixfold. Note who is making it. Volmex measures volatility; its Bitcoin Volmex Implied Volatility Index is meant to capture the market’s forward-looking expectation for bitcoin, and the firm launched BVIV-US this year, an index that derives bitcoin volatility expectations from options tied to BlackRock’s iShares Bitcoin Trust (IBIT). A firm that sells expected volatility is entitled to a price view; its business, though, is expectations, which is not the same business as outcomes.
Here is the ledger of this same call, all from one piece. Standard Chartered forecast exactly $500,000 by the end of 2028; in December 2025 it pushed that target back to 2030, citing weaker corporate buying. In September 2021, the ARK Invest CEO said bitcoin could rise roughly tenfold to around $500,000 over five years, and per the coverage, it never happened. Galaxy Digital’s Mike Novogratz repeated a $500,000 call in 2022 within five years, a deadline near 2027 that the coverage itself calls highly unlikely. Robert Kiyosaki, of Rich Dad Poor Dad, went further on timing: in February 2023 he predicted $500,000 by 2025. That one failed too.
Same number, five years, four or five authors, not one expiry met. So name what this is. A pushed-out deadline is not a corrected forecast — it is a subscription. A forecast dies on its date when it is wrong. This thing does not die; it takes a new date and keeps going. The coverage states plainly that $500,000 by 2028 is not an unprecedented Wall Street target. That is the more damning fact, not the reassuring one. The target’s pedigree is the reason to distrust it, not the reason to trust it.
“Should Be Taken With a Huge Grain of Salt”
The coverage says this itself: such targets should be taken with a huge grain of salt. I agree, and the reason is incentive, not stupidity. Nobody is scored on a target being missed. A wrong target is still a headline, still a segment, still a quote in a piece like this one. The cost of the miss is zero, and the benefit of the miss equals the benefit of a hit. When the payoff is identical on both branches, do not be surprised that the branches are not treated with equal care.
I have seen too many teams ship a forecast with no cost of being wrong. In a production system, a bad number costs you something: a pager at three in the morning, a rollback, a postmortem that names the assumption which failed. That cost is the discipline; it is why the estimate gets checked before it ships. Here the cost is absent. The person who says $500,000 and the person who says nothing are treated identically, except the first gets quoted. Reverse the incentive and the supply of targets collapses overnight. Nothing about the technology changed. Only the accountability changed.
That is also why the numbers get larger, not smaller. A modest target is a risk with no reward, because if it lands it was obvious and if it fails it was still wrong. A large number is a free option: it fails harmlessly, and if it lands once you are a legend. This market rewards variance, not accuracy, and it rewards variance most where the horizon is long enough that nobody is keeping score. That is the salt. The grain is the number.
“I Still Like the Chart”
Now the part I actually enjoyed. Mike Novogratz, CEO of Galaxy Digital, remains bullish on bitcoin and says a move to $100,000 before the end of the year would not surprise him. His words, from a recent All Things Markets discussion, per U.Today.
“I still like the chart,” Mike Novogratz said. “It will not surprise me if we’re at $100K by year end.”
Mike Novogratz, chief executive of Galaxy Digital, All Things Markets discussion
Sit with that for a second. This is one of the men whose 2022 $500,000 call sits in the failure column two sections up. Same person, same method, a much smaller number and a much nearer date. I note it without malice, because it is not hypocrisy. It is closer to convergence. A nearer deadline and a smaller number is, in fact, what the evidence supports, since the multiple and the horizon trade against each other. You cannot have both a huge multiple and a near date without claiming to know something the market does not. His revised call is smaller and closer. That is the direction a disciplined estimate moves when it is marked to reality instead of marked to narrative.
The puzzle is why it is still packaged as a chart call at all. “I still like the chart” is a statement about a man’s feelings, not a statement about the chart. A chart is a record of prices that have already printed; it has no opinion, and it cannot be liked or disliked any more than a thermometer can be liked. What he means is that he expects the price to rise, which is a view about the future, not a reading of the past. I am not mocking the man. I am pointing at the grammar, because the grammar is how a feeling gets laundered into a signal. When someone says he likes the chart, translate it: he has a prior, and the prior is not derived from the chart.
What Is Actually Measurable
Set the models aside. These numbers need no curve fit and no ruler, only arithmetic anyone can reproduce. Using CoinGlass data, the coverage reports bitcoin is on track for its second-best third quarter on record, a gain of roughly 43.5% that trails only 2017’s 80.41% and beats 2013’s 40.6%. The other positive quarters: 25.01% in 2021, 17.97% in 2020, 6.31% in 2025, and under 1% in 2024. The substantial declines were 39.74% in 2014 and 22.86% in 2019. CoinGlass puts the average Q3 return at 8.73% and the median at only 2.29%; eight of the 14 quarters in the table were positive. Bitcoin had lost 22.2% in the first quarter of 2026 and 14.09% in the second, so a near-43.5% third quarter would snap a two-quarter losing streak. It traded near $84,800 when that piece was written, according to CoinGecko.
Now read the two summary statistics against each other, because the gap is the story. The mean is 8.73% and the median is 2.29%. When the mean sits far above the median, a few enormous quarters are dragging it, and 2017 and 2013 are doing most of that work. The typical quarter you might actually get is near the median: a hair over two percent. Anyone quoting 8.73% as “what Q3 usually does” is quoting a number that almost no individual quarter delivers.
That is a distribution, not a trajectory. It gives you the range of outcomes and roughly how often you land where, and it refuses to say what happens next. It is boring, it is incomplete, and it is the honest thing to look at. I would rather reason from a distribution I can check than from a ruler that claims to predict.
So let me mark the edge of what I know, which is the only part I can defend. I am not forecasting bitcoin. I have no $100,000 call and no $500,000 call, and giving you one would mean inventing it. A large rally is entirely possible; the Q3 distribution holds both 80.41% and -39.74%, and I do not know which regime we are in. What I can say is narrower, and I think more useful: the target genre cannot tell you the price in 2028 or 2029, because the method that produces the number is calibrated on the number it claims to predict. The one thing you can check, cheaply, is the single behavior the genre performs reliably. Watch the next deadline. When it arrives and the target has not printed, look for the quiet line that moves it — from 2028 to 2030, from five years to a few more. That extension is not a correction. It is the target admitting what it always was. And that, unlike the chart, is measurable.
