I was recently reading yet another good historical novel by Gillian Bradshaw1 - in fact, a series of two she set in the English Civil War, London in Chains and A Corruptible Crown. She says in her afterward that despite hot disputes about the actual motivations of various personages there, "Luckily for me, I'm a novelist, not a historian, and the question I had to answer wasn't 'what really happened' but 'what would my protagonists think had happened?'... The history in this book is a Leveller view of the Second Civil War - as interpreted, of course, by a twenty-first century female novelist."2
This's quite correct, and she does it well. Regardless of whether King Charles or Cromwell were scheming tyrants, it makes sense for her characters to view them that way. Regardless of why Colonel Thomas Rainsborough was killed (modern historians still aren't sure), it makes sense for the characters to think he was assassinated by his political enemies in his own army.
But I think she's pointing to an even broader idea here about historical fiction.
There're two types of historical fiction, one which doesn't try to do as much as the other.
The first type - the looser sort - uses the historical period as a backdrop, has its own plot consisting completely of invented events, and doesn't mention actual events except perhaps in passing. As an example of this, we can take Bradshaw's Render Unto Caesar, a book about a fictitious merchant who comes to Rome during the early Empire to collect on a debt; or Rosemary Sutcliff's Frontier Wolf, a book about a Roman garrison commander in the late days of the Empire facing one of many attacks.
What these books need to do, like all historical fiction, is show us a valid view of the historical period: to show us enough of the historical period to make us feel that we're there and we understand the setting enough to understand the characters and story. Just like a fantasy writer has to worldbuild their invented world and show it to the reader, a historical fiction writer needs to show the reader the worldbuilding of the actual past. We need to understand what it was like to live in early imperial Rome, or 1930's England, well enough to envision these characters as they live their lives in the book. We need to know what they'd be thinking; what would be ordinary to them, what sort of choices they imagine themselves to have, and what they'd be imagining as they make decisions in the plot.
To write this well, the writer also needs to make their characters realistic to the period. One of the things the novel is showing us most directly is those characters, so it can't show us a realistic view of the past unless the characters are also realistic people from the past. Many books fail at this. One place where the failure's more obvious is H. Rider Haggard's The Brethren. He was a Victorian author; he set his book during the Crusades, but his protagonists and most of the supporting characters think and act like Victorians. Their morals and sense of fair play are much more Victorian than medieval.
Also, the writer needs to make the real past believable to us. Writer Jo Walton called this challenge "The Tiffany Problem": "Tiffany" was an actual medieval name, short for "Theophania", but it sounds so modern that no modern reader will believe it was actually medieval. Similarly, ancient and medieval religion was very different from most modern people's conception of it. An author can explain all this for the reader - but it will take pagecount and effort.
If you take this seriously, it's a daunting task. This's one reason I've hardly ever written any historical fiction.3 I know I'd hesitate at every detail and second-guess it and demand more research. In part, this's my own insecurities, and I wouldn't hold them over other authors. But in part, it does demand substantial research - as Bradshaw (for example) openly admits. The only people who don't need to research are people who've lived through the period they're writing about or already talked with people who have - but that would be a very limiting principle.
The second type of historical novel - the more difficult type - is one that doesn't only use the historical period as a backdrop, but talks about real historical events. Cromwell isn't just out there somewhere fighting the Levellers outside the book's view; he's doing it in the pages of the book directly.
Sometimes the historical events are referred to in the course of other events, like in many kids' historical novels (one example I remember happily is Across Five Aprils about the American Civil War where our protagonists on the home front react to news from the war); or occur in just one moment, like Rosemary Sutcliff's Dawn Wind where we briefly see Augustine of Canterbury's arrival in England. I'd say these are a blend of the two sorts.
But sometimes the book is about the historical events themselves. In the extreme, we get stories like the film The Death of Stalin where the plot is entirely around Stalin's death and funeral; or the film Lincoln where it's entirely about the Thirteenth Amendment abolishing slavery. Less extreme are Bradshaw's books London in Chains and A Corruptible Crown: our protagonists and their personal relationships and struggles are fictional, but the plot is dominated by the real historical events taking place around them and influencing them at every turn.
Often these stories aren't exactly accurate in every detail. Sometimes they try to be, like Bradshaw, or like Nordhoff's Mutiny on the Bounty where he explains that the one intentional change he made was to insert his fictional protagonist onto the ship's roster. Or, as I wrote earlier about The Death of Stalin and Lincoln, sometimes they don't try to be. It's of course more difficult to get every detail correct - but that can be more rewarding as well if a writer can weave a story through all the known events of history.
But, there're other rewards there too. The two films I mentioned meet the other goal and challenge of historical fiction very well: if a story's talking about historical events, it can't avoid giving some message about them. Putting historical events into a story can hardly avoid imbuing them with some meaning. You're making them into a story; just like any other events in a story, they need to relate to the story with some clear theme or connection.
You can give them the same meaning it had to some people at the time, as Bradshaw does in London in Chains, which is indeed "a Leveller view of the Second Civil War" - essentially how the Levelers at the time did interpret those events. Or you can interpret it how the people there at the time studiously avoided interpreting it, like The Death of Stalin which tells the whole thing as a dark comedy. Or, you can give it some totally different meaning, like Sutcliff does in Dawn Wind where she interprets Augustine's arrival as a symbol of the return of civilization to Britain.
Sometimes, of course, those messages can be really weak. Haggard's The Brethren seems to almost trivialize the Crusades as mere adventures. If he'd used that to criticize them, that could be done powerfully; but he seems to consider them a good thing even as he ignores any higher purpose than adventure. Perhaps this's just weak writing, but I suspect Haggard's message falls flat because of a moral disconnect between me and Haggard.
When done well, this's the power of historical fiction: that - in addition to all the power of any good story - it can give new insights and perspectives on history much more vividly than nonfiction history books.
Reading actual history books is very useful, and fun too. But to actually appreciate what it was like in history - whether amid actual historical events, or just what it was like to live in a historical era - we need historical fiction. Factual history books inevitably leave us with all sorts of gaps and open questions. Even if we had a time machine with a spy camera, we'd still be left to wonder what people were thinking and how they reacted to events; without that spy camera, we're left wondering at so many other details of connections and daily life. History, as we read it, is not just the past; it's the stories we tell each other to bring out meaning from the past and connect it together.
Take Bradshaw's protagonist: a woman printer in Civil War London. There were real people fitting that description. If we properly align facts from history books, we can plausibly approximate her perspective on events and envision her life. But even if we go to all the work to do that, we're usually left without records of where she came from or where she went. And even if we (improbably) had those records, we'd be left to wonder at her motivations.
That's where the best historical fiction can come in: to fill in plausible answers for those, take out the work of research, and help us in our imagination. And by doing that, it excites our sympathy and brings history alive.
Yes, I've been reading a lot of Gillian Bradshaw; after some strong recommendations, I resolved at the start of last year to read every novel she'd written. So far, it's been a mostly-enjoyable resolution!
The Levelers were a faction in the 1642-1649 English Civil War (sometimes divided into First and Second Civil Wars) committed to government by the people and equality before the law. They supported Parliament against King Charles I, but afterwards opposed Parliament as favoring upper-class interests. In the end, they were suppressed by Cromwell (a general in Parliament's army) as he established his dictatorship.
Except for some play scripts when I was a teenager - the conventions of amateur theater let me get over my second-guessing about inaccuracies.
The two writers I've read who do what you call the second type of historical fiction are Robert Graves and Mary Renault. I'm not knowledgeable enough to judge how accurate either one is in terms of their use of the surviving sources, but it's clear that they're consulting those sources. (On the other hand, Graves' King Jesus appears to be wildly speculative.)
Tim Powers seems to do the same sort of thing, but with the invented material containing overt fantasy, as in The Stress of Her Regard, where the famous weekend at the Villa Diodati is surrounded by struggles with ancient vampiric beings.
I think the technical term for this sort of thing might be "saving the appearances": You're free to hypothesize, but your hypothesis must not imply anything contrary to the data.
Your Gillian Bradshaw quote reminds me of some comments by Hilary Mantel on her Wolf Hall series (which I suspect you've read). On the one hand, she made a commendable effort to avoid contradicting the historical record where it exists - past the point of duty, in my opinion - but on the other, she tells the story of Henry VIII in a way that rather unusually casts Thomas Cromwell as the hero (or, at least, less villainous than his opponents) because the book is written from Cromwell's point of view and he'd probably see things that way.