The method almost everyone starts with
Spread grain in a single layer on a sheet pan. Heat the oven to 350°F. Slide the pan in, set a timer for ten minutes, stir, set it again, stir again, and pull the grain when it looks and smells right — usually somewhere around the half-hour mark for amber. Let it cool, leave it a week or two, then use it.
That is the standard home method, it appears in every homebrewing forum thread on the subject, and it deserves better than the dismissal it usually gets from people selling roasters. It works. For half a pound of grain it works well enough that buying equipment to do it would be silly.
What follows is the fair version: what the oven genuinely does, and the four specific places it runs out of road.
What the oven does well
An oven holds a broadly stable air temperature in a well-insulated box, which is most of what an amber roast asks for. The 350°F for about thirty minutes region is slow and forgiving. Colour arrives over minutes. You have time to open the door, look, take a few kernels out, cut one, taste it, and decide. Very little can go badly wrong in the ten seconds between one look and the next.
It is also free. You own it. There is no setup, no gas bottle, no cleanup beyond a pan, and nothing to store afterwards. If you want to find out whether a half pound of home-toasted malt does anything interesting in your beer, the oven answers that question this afternoon for nothing. Anybody who tells you to buy hardware before running that experiment is selling rather than advising.
Keep using it while it is doing the job. The reasons to stop are specific, and if none of them apply to you then none of them apply to you.
Where it stops
Nothing moves the grain between stirs
This is the big one, and it is structural rather than a matter of technique. Between stirs, every kernel sits exactly where it is. The grain touching the hot pan takes heat by conduction and runs ahead. The grain on top takes it from the air and lags. Grain near the edges of the pan sees the oven walls and the hotter perimeter; grain in the middle of the mass does not.
Stirring every ten minutes does not fix that. It shuffles the deck three times over a thirty minute roast, which averages the extremes a bit but leaves you with a batch that is a spread of colours rather than one colour. Cut open twenty kernels from an oven batch and you will find pale ones and dark ones in the same handful. Every one of those kernels contributes its own flavour, and the finished character is the average of a wide distribution instead of a narrow one.
A drum turning continuously does not average the extremes. It prevents them, because no kernel stays against the metal long enough to run ahead of its neighbours.
The dial reads air, not grain
The number on the front of the oven describes the air in the cavity. Your grain is a separate object with its own moisture load, absorbing heat at its own rate, and the gap between those two numbers changes continuously as the roast dries out. There is no offset you can apply to convert one into the other.
In practice that means an oven roast is controlled by the clock and the colour of the grain, which is why the same recipe gives different results on different days. Change the sack, change the humidity, change how long the oven preheated, and thirty minutes lands somewhere else. Nothing about this is unique to ovens — it is true of any roast where you cannot see product temperature — but the oven gives you no route to fixing it, because there is nowhere to put a probe that touches the grain and stays there.
Capacity is pan area, not oven size
The method depends on a single layer. That is what keeps the bed shallow enough for the air to reach most of it between stirs. One sheet pan in a single layer is somewhere around half a pound of grain, and that number does not scale the way you want it to.
Doubling the depth on the same pan does not double your yield, it doubles your unevenness, because the middle of a deeper bed is now insulated by grain on all sides. Adding a second pan on another rack gives you two different roasts running at once, because oven racks are not the same temperature. So the realistic ceiling is about a pound a session, in a session that took the best part of an hour with the preheat. If you need five pounds of amber malt, you are looking at a full day of pans, and every batch will be slightly different from the last.
Dark roasts fill the house with smoke
Amber is fine indoors. The dark end is not.
Taking grain to roasted barley or chocolate malt colour produces real smoke, in real volume, and it smells like a roastery rather than like baking. A domestic extractor hood is not built for it. What actually happens is that the smoke goes into the kitchen, then the rest of the house, then the soft furnishings, and it is still faintly there two days later. Fine chaff burning on the oven floor adds its own note. This is the single most common reason people who have been happily oven-roasting amber for a year suddenly start looking at outdoor equipment.
The oven also cycles
Worth knowing, though it matters less than the four above. Most domestic ovens hold a set point by switching an element on and off, so the air temperature swings around the number on the dial rather than sitting on it. On a slow amber roast the grain mass smooths most of that out. Closer to the dark end, where a minute changes the outcome, the swings start to show up in the batch.