If you're thinking about exercising your ISOs, you've probably been told to do it before year-end.
That advice isn't wrong. It's just incomplete. Exercising in December locks you into a tax outcome the moment the calendar flips. If your stock tanks in January, you'll be paying AMT on a phantom gain that no longer exists.
Exercising on January 31 gets you almost the same tax-year result, plus eleven extra months to watch the stock and decide whether to hold for a qualifying disposition or bail out before AMT bites.
Here's how the strategy works, with the math.
If the qualifying-disposition rules are new to you, Incentive Stock Option Taxation, From Exercise to Sale walks the four sale paths and Equity Compensation and the Alternative Minimum Tax covers what triggers AMT in the first place. Start with the equity compensation guide if ISOs themselves are new.
Why January Beats December
The early-year exercise sets up two doors that close at different times.
Door A: Hold for a qualifying disposition. If you exercise on 1/31 of Year 1, you can sell after 1/31 of Year 2 and have the spread treated as long-term capital gain. The exercise adds the bargain element to AMT income in Year 1, which may or may not produce an actual AMT bill depending on the rest of your picture. If it does, a credit starts coming back in Year 2.
Door B: Bail out before year-end. If the stock craters between the exercise date and the end of Year 1, you can sell before December 31 and trigger a disqualifying disposition. The bargain element stops being an AMT preference item and becomes ordinary compensation income: taxed at the actual sale price, not the higher exercise FMV. AMT goes away. You take the tax hit at a much lower number.
Exercising in December gives you Door A and almost nothing else. By the time you'd want to bail out, the year's already over.
Exercising in January gives you both doors for eleven months. That's the whole strategy.
The cost of getting it wrong on either side, paying AMT on a position that's down 60% or missing the qualifying disposition by a few weeks, runs into the tens of thousands. Sometimes more.
The Setup, in Numbers
Pick a hypothetical. A senior engineer and their spouse expect $425,000 of ordinary taxable income for the year, MFJ, putting their baseline federal tax around $89,000.
On 1/31 of Year 1, they exercise 5,000 ISOs granted four years earlier with a $10 strike. The current FMV is $60. They write a check for $50,000 (5,000 × $10) to take ownership of the shares. The bargain element, what the tax code says they "got" by exercising, is $250,000 (5,000 × the $50 spread).
ISOs don't trigger ordinary tax on the bargain element at exercise. But the bargain element is an AMT preference item, added to AMT income. If the shares are still held at year-end, that $250,000 flows into the AMT calculation.
For this household, that $250,000 add produces roughly $65,000 of AMT, bumping their total Year 1 federal tax from ~$89,000 to ~$154,000.
Hypothetical illustration using 2026 figures. AMT income is taxable income with the standard deduction added back, plus the $250,000 preference. Exact AMT depends on filing status and the year's exemption schedule. These figures also exclude the 0.9% Additional Medicare Tax on earned income above $250,000 for joint filers, reported on Form 8959; it sits outside the AMT comparison, so it doesn't move the AMT number, but it does raise the total bill.
That $65,000 isn't purely a cost. Paying AMT generates an AMT credit of the same amount, recoverable in any future year where regular tax exceeds tentative minimum tax. Scenario A is what it looks like when it starts coming back.
This is the moment most ISO advice ends. The strategy is just getting started.
Scenario A: When the Stock Cooperates
The shares appreciate. By February of Year 2, they're trading at $72.
Five days after the qualifying disposition window opens (more than one year from exercise, more than two years from grant), our hypothetical exerciser sells all 5,000 shares for $360,000.
Two things happen in the tax accounting that often surprise people.
Regular tax sees a $310,000 long-term capital gain. Basis for regular tax purposes is what they paid out of pocket: $50,000. Yes, $310,000 is a big number. It's at LTCG rates, not ordinary, which is the whole point of waiting for the qualifying disposition.
AMT sees a $60,000 gain. Because they already paid AMT on the $250,000 bargain element in Year 1, AMT basis is $50,000 + $250,000 = $300,000. The AMT-side gain is just $60,000. This is dual basis: the tax code's mechanism for keeping you from getting taxed twice on the same dollars.
In Year 2 the AMT calculation carries a negative $250,000 adjustment on Form 6251 to undo the prior year's preference. That pulls AMT income well below regular taxable income, and the gap between the two systems is what the Year 1 credit can be claimed against:
The credit can only be claimed to the extent regular tax exceeds tentative minimum tax. In Year 2 that gap is $48,000, so $48,000 of the $65,000 balance comes back and about $17,000 carries forward. Two things hold the rest back: the 3.8% net investment income tax isn't regular tax, so no credit can touch it, and AMT still taxes the $60,000 AMT-side gain at long-term rates.
Total Year 2 federal tax after the credit, including NIIT: roughly $105,000.
The useful number isn't the two-year total, because most of that is tax on a salary they'd have owed anyway. It's the total tax the exercise and sale actually produced:
That $64,000 is exactly the long-term capital gains tax plus NIIT on a $310,000 gain. The AMT was never a cost. It was a prepayment, returned through the credit, with a tail of about $17,000 still on the books at the end of Year 2.
To isolate what the qualifying disposition is worth, hold the price flat and run the same $250,000 spread both ways.
Same shares, same price, same spread. The $35,000 difference is entirely rate: ordinary income on one side, long-term capital gains plus NIIT on the other.
The timing differs though. The same-day path settles its $85,000 in Year 1 and is done. The qualifying path pays about $69,000 across the two years and gets the rest back as the remaining $19,000 of credit unwinds in later years.
There's a second, less obvious benefit: the Year 1 AMT bill comes due 4/15 of Year 2, after the early-February sale. The proceeds fund the AMT. No need to scrape together $65,000 from outside cash.
Scenario B: When the Stock Tanks
Now run the same setup, but the stock falls 50% to $30 by November of Year 1. The full position is worth $150,000, against an AMT bill of about $65,000 if it's held through year-end.
Hold the position through 12/31, and the AMT calculation doesn't care that the stock is down. The $250,000 bargain element is locked in at the exercise-date FMV. The AMT bill is still about $65,000, on a position now worth less than half of what it was when it triggered.
If the household can't pay $65,000 from outside savings, they're selling into a depressed price to fund a tax bill on a gain that no longer exists.
This is where the bail-out earns its name.
By selling before December 31, our hypothetical exerciser triggers a disqualifying disposition. The tax treatment shifts:
The bargain element gets re-characterized as ordinary compensation income, calculated at sale price minus strike ($30 − $10 = $20/share × 5,000 = $100,000), not at the original exercise-date $250,000.
No AMT preference. The position never crosses into AMT territory because the disposition disqualified the ISO treatment that triggered AMT in the first place.
No capital gain on the sale (basis = strike + ordinary comp income recognized = $30/share, which equals the sale price).
Hypothetical illustration. Specific tax outcomes depend on household income and filing status.
The bail-out costs about $33,000 less in federal tax than holding through year-end. It also converts a paper position into $150,000 of cash that can be redeployed.
Exercise the same ISOs in December instead of January, and none of this is available. The stock can fall 80% in the first week of the new year and the AMT is already baked in.
The AMT Credit Comeback
The AMT credit isn't a refund. It's a deferred recovery. You get to apply it in any future year where your regular tax exceeds your tentative minimum tax, and the qualifying disposition naturally creates that scenario, because the negative AMT adjustment in the sale year drops AMT below regular tax.
Two things people miss about the credit:
It doesn't apply if you bail out. The AMT credit only exists if you paid AMT in the first place: you held through year-end. Bail out, and there's nothing to recover. There's also nothing to recover from, because you avoided the AMT entirely.
It can take years to recover if the stock falls and you held. Hold through year-end, get hit with AMT, then sell at a capital loss in the following year, and the AMT capital loss limitation can stretch the credit recovery across multiple tax years. You'll get the money back. "Eventually" can mean five years.
This is exactly the case where the bail-out outperforms holding. Take the smaller, certain hit in Year 1 instead of the big AMT hit followed by a slow-motion credit recovery.
What the Math Doesn't Show You
Everything above is the federal tax math on a single equity event. In a real planning situation, the early-year exercise sits inside a half-dozen other moving parts, and the right answer changes when you zoom out.
Cash flow. The $50,000 to fund the exercise has to come from somewhere. So does the $65,000 AMT bill in April. If you're funding both from a brokerage account, you're triggering more capital gains. If you're borrowing to cover the AMT, you're carrying debt alongside a concentrated equity position. None of this shows up in a single-year tax calc.
State tax. California runs its own 7% AMT on the same bargain element, so a large exercise triggers a state bill on the same gain you haven't collected. Other states treat ISOs their own way, and where you live when you sell can change the answer. The federal math above says nothing about any of it.
Concentration. Holding 5,000 shares from January to February of the next year means twelve months of single-stock exposure instead of none. That's a real risk position separate from the tax math, and it cuts in both directions.
Other equity events. A secondary sale, an IPO, a tender offer, an RSU vesting cliff: any of these compound or interfere with the ISO decision. The right exercise plan depends on what else is happening across your full equity comp stack, not just the ISOs in isolation.
Future-year tax planning. The AMT credit only recovers in years where regular tax exceeds tentative minimum tax, so how fast you get it back depends on what your income looks like in those years. A large vest, a tax-gain harvesting move, or a lighter earning year each change the pace, and that changes the optimal exercise size today.
These aren't separate problems. They're the same decision viewed from different angles.
The tax math on a single ISO exercise is the visible part. Coordinating it with the rest of your financial life is most of what actually matters, and most of what generic ISO advice doesn't touch.
More Doors, Not Fewer
The early-year exercise isn't a "strategy" in the punchline sense. It's a setup. The actual strategy is the flexibility you give yourself by exercising in January instead of December: the right to look at the stock in November and choose whether the qualifying disposition or the bail-out fits your situation better.
That logic runs through almost every good equity comp decision. You're not optimizing for a known outcome. You're making sure you still have a move to make when the stock does something you didn't predict.
Which is, conveniently, exactly how the rest of your financial picture should work. Every decision shaped to give you more doors, not fewer, when life and the market do what they do.