Your question is an i portant one, and one that hadn't been asked frequently enough by historians until about 30 years ago. Today, we're very aware that most of our narrative sources - even those which claim to be factual accounts of events - are forst and foremost works of literature, not strict records of events. But historians have been less skilled at making these distinctions in the past (and occasionally continue to be, in a few cases).

Historians have always been concerned with questions of accuracy. If you read Ploybius' history of Rome (written in the second century BCE), he has a rant about how this other historian wasn't an eye witness, and relied on others' written accounts, and as a result was complete rubbish. Thicydides criticized Herodotus for the same lack of eye witness accuracy several centuries earlier. And in the following two millennia, historians continued to question whether the accounts they were reading were teustworthy, using a variety of tests (ranging from whether the text was an eye witness to whether the sogns and portents recorded by the author were sufficiently impressive to indicate accuracy).

The big shift in modern history happened in the 19th century, when a German historian named Ranke argued that we shouldn't rely on past historical accounts to know what happened, but should instead read the surviving legal and administrative documents preserved in government archives. Such as approach, he argued, would let us see events as they really happened, rather than trusting the dubious accuracy of people who purported to have been there and observed it all. Ranke's practixe continues to form the basic method for many modern historians (going to archives, reading administrative documents, and piecing together the lives they describe into narratives based on this data).

But the earlier you go,mthe harder it is to find surviving administrative documents. We've been lucky in a few cases - there are some great late medieval archives surviving, and we have some legal cases from the early middle ages. We even have administrative documents (tax assessments, court cases, etc) from parts of the ancient world. But these sources are few and far between, and it's often the case that the only texts we have surviving are histories written by near contemporaries in the ancient world.

How do we decide if these texts can be trusted, or if they're fictional?

It used to be that historians would trybto tease out the reliable facts from documents, separating truth from exageration, inaccuracy, and outright authorial invention. If, for example, three different authors agreed about an event, historians would assume that was how it happened. If they disagreed, historians might side with the author who usually got his facts straight, rejecting the others as being less trustworthy. Some sources that claimed to be reporting events accurately might be thrown out altogether as being works of historical fiction, if historians agreed they had too many errors. To an extent, this approach is still how many historians read their sources today.

But starting in the 1980s, many historians began to question some of the assumptions implicit in this approach to reading ancient authors. The biggest assumption was that ancient historians shared our distinction between truth and fiction. If an ancient author claimed to be telling events how they happened, the assumption was that they would then try to give the facts as objectively as possible. Because ancient authors were trying to be objective, modern historians' goal was simply to test ancient authors' accuracy, work around their biases, and figure out whether they were credible witnesses to the events they described.

But starting in the 1980s, historians like Walter Goffart began to show that ancient and medieval authors did not care about the difference between truth and fiction in quite the same way we do today. Goffart (1988) looked at stories that claimed to tell the origins of early medieval ethnic groups like the Anglo-Saxons, the Lombards, and the Goths. Previous historians (like Wolfram) read these origins stories and assumed they were, if not completely accurate, still written with the goal of describing objective information about these people groups' origins. Goffart, however, noted that these histories were first and foremost works of literature, written for specific purposes for specific audiences. Before we could assume they were works of objective non-fiction, we needed to figure out what their authors hoped to accomplish when writing them. To do that, we had to begin by asking not what facts might be trustworthy in the story, but instead what points or lessons the stories were trying to tell to their readers, and then analysing how the stories' events fit into those larger objectives. Reading the stories as works of literature with themes, carefully crafted plots, and a message that they were trying to prove made it clear that authors like the medieval monk Bede were more concerned eith convincing their audiences to believe a particular version of the past than they were with getting all their details right. Histories which we had assumed were trying to report the facts (albeit sometimes with heavy bias) were, instead, frequently more concerned eith telling a compelling story.

The impact of Goffart (and others, using similar methods) has been profound. In the field of Late Antiquity (the end of the ancient world and beginning of the middle ages), historians have stopped looking for 'facts' in ancient historical accounts. We've stopped believing they're trustworthy works of non-fiction. Instead, we read each text - including the ones that claim to be written by eye witnesses - as literary compositions - stories - and we ask questions like: what did the text talk about? What did it leave out? Why did it tell this story in a particular way when other texts describe the events differently, and can we learn anything about the time when these stories were written by studying the reasons that led them to tell two very different cersions of the same event?

The result if this has been to, on the one hand, cause us to be more critical of some texts. You can't just say, 'that guy is pretty reliable!' and assume something someone wrote is True anymore. You have to ask the question you did here: might it just be a fictional story?

But on the other, realizing that many of the texts we used to think were mostly factual are, instead, literary compositions has led us to start looking at many of the 'less accurate' authors with fresh eyes. Historians whose writings had been dismissed as 'too biased' to be trusted, when read as works of fiction, can tell us a lot about the societies in which they were writing. Sure, their facts might be wrong - but why did they believe them? A colleague recently completed a dissertation lookig at a church historian who has been mostly ignored because he was a heretic and, therefore, assumed to be less accurate than the orthodox authors; in fact, what she doscpvered was that he had a perspective on late roman history which was engaged in a very lively debate with many contemporary political issues, and reading him as a participant in these political debates reveals both why his history looks different from his peers, and helps us understand the debates of which he was a part much more clearly. Fiction, literature, propaganda, and unreliable histories can tell us a TON about the peop,e who wrote them, and the world in which those people lived.

So, to circle back around to your question: ancient and medieval historians generally don't trust ancient historians to be accurate anymore. Instead, we read the, as products of their time, windows into ideas and attitudes of the cultures in which their authors lived. We do still try to figure out whrther the events they described were real, but we also recognize that the lines between fiction and fact were fuzzier in the ancient world than they are today, and the best we can do often is to read with skepticism and broaden our research to include other kinds of evidence (like archaeology) which give us alternative perspectives with which to test and evaluate our unreliable ancient narrators.

If you want to read more about how early medieval historians tackle the problem of unreliable narrators, and how archaeology can (and can't) help, I'd recommend Guy Halsall's attempt to answer the question: was there a historical King Arthur? in his 2013 book, Worlds of Arthur: Facts and Fictions of the Dark Ages. He spells out very clearly all the difficulties of separating fact from fiction in sources that claim to be accurate records of events - but aren't.

Today, we have books like the Hunger Games, Game of Thrones, The Shining, etc. but if a thousand years from now someone dug one of these books up and read it, would they think that the story of a girl who murdered her peers while also finding government corruption to be real?

Probably not, there wouldn't be any corroborating accounts and the book specifically says it's fiction.

Why are most texts and writings we find today that were written about 600 or more years ago considered to be actual accounts?

They're not just assumed to be first hand accounts of actual events; why do you think this is the case?

Why can't some just be fictional stories or author thought theories?

Many are. The Historia Regum Britanniae, for example.

The only one I can think of that is a fictional story is the Epic of Gilgamesh.

The Canterbury Tales, all the Homeric epics, The Tale of Two Lovers, Blanquerna, Romance of the Three Kingdoms, &c.

One thing I can point to is that in the Islamic tradition in the Middle Ages there's actually a cultural aversion to prose fiction because fiction could be construed as a lie.

There were basically two ways to get around that. One was to construct stories that were so obviously fictional that they couldn't be construed as lies. Qalila wa Dimna, a collection of animal fables in which the animals talk, would be one example of that kind of obvious fiction.

The other way to get around it would be have the author say that the events being recounted were told to him via a chain of transmission so that the lie gets shifted onto someone else. This can include the chain of transmission having one really obviously weak link that marks it as fiction so "James J. Peabody told me that Sarah S. Wentworth told him that she met a guy at the bar named Joe who told her the following story."

That practice is a parody of the historical and religious tradition in Islamicate history of maintaining a chain of transmission from the person who saw the event to the person who recorded it. In analyzing history, particularly in the early Islamic period where we don't have much in the way of documentary sources, this becomes problematic because even though the historical writers were supposedly careful to record these chains of witnesses, we have no way of verifying the chain. "Hadith science" purports to be able to do this for the chains of transmission of the sayings of the Prophet Muhammad but most historians of Islam are, at the very least, skeptical of those chains.

Then of course you have issues of hagiography, or polemic that sneaks into historical works.

It's a tough call, it's a major issue in historiography to separate that myth from legend in the middle ages. It helps sometimes that, as in our own period, some historians just seem to be better or more serious than others. Tabari is a good example of that, where even critics of the early Islamic historians often recognize that Tabari's project and methods were really ahead of their time.

In God's Path by Robert Hoyland has some good writings on this in the appendices on sources.