Transcript for FiveThirtyEight: Introducing the 2016 general election forecasts
So it's mostly based on on polls and we're dealing with. Various problems I mean in theory to say well it's kind of taken hammer to the polls a figure out how much uncertainty there isn't and pre election that way. That's in the sky idea to simple idea. In practice things are complicated get to know how much to Wear a different polls also. Yes it is contested one C at a time. But the states are related Claire. If Ohio moves certain direct I hear -- countries regional probably will to bring out the core relations right so it's kind of like. Simple and principal in practice it gets rather technical and so we spend a lot of time produce the right way. And I would say it will where faith but model basically trying to there was if you are take a simple average right that would get to allow the way there but. These type of techniques can help. You sort of get from say 95% of the wait to 99%. Of the way there. And that kind of came and play big time in 2012 wearing Nate I was not what's at the time but Nate thought very cleverly used that lets add up the state also come up with a national estimate and that was slightly. I was staples have more information and yet there's no doubt we haven't had as many staples a little bit trickier but their point election almost start to see a lot more state always sort of I think now I'm surprised we haven't had as many yet but. But you're usually by the time we get to certainly after the convention sermon at September. You know we're getting for Ohio polls a week sometimes for Ohio polls a day would be more precise about that. And there's a public pension. That convention bounce out of ketchup bottle up the battle. Usually happens traditionally is that. A party has convention it. It's four days of free advertising basically in every major network. And you get abuse in particular. Number one you excite your partisans of voters who might have been reluctant but for work. They join the party figuratively and literally declared. And also my twelfth. You did so Obama got a bounce an estimate and Ronnie get much I didn't and and someone is now is like a better read of the polls than a month from now we're kind of in the shadow of the connections and attending it late August we'll get an even better read stale but. Yeah you know to me they can mention this kind of make. Kahuna that is where campaigns that seemingly we're dead kinda turned around if say we end up after the conventions in her mid August where that a month bout. And Donald Trump is still trailing. By seven percentage points. Back to me it's a very very bad bad sign up for his campaign. Accurately be you don't look trump is behind now if you're behind in June not as bad as being behind in September by seven points then we're kind of getting into it. Alar territory where there isn't really a lot of precedent you can argue that Reagan in eighty. Where that big a swing late in the election. I guess you know what what is unexpected that is really you know thump and that could. Chick terrorist attacks yep but I mean you know I'm a Bentley had to you are ready and I'm and I'm here's Robertson might run it right. Irked every attack. Is different but we've had in the past two weeks. A awful attack in Orlando. And with the kind of an exiled in this economic shot in the former wrecks that we're the markets were sent tumbling. Like neither man seems to have helped. Trump I think he's gonna have to become a much better campaigners the bottom line. I think there's no way that eat it mean to me it's kind of it's up from here from you there's nowhere to go but up. From where he stands out that's another and other economic shock happens or terrorist attack literally Tony about a new professional ice trap campaign. Called an afford it how. Maybe they have more of a nuanced response to it and they don't sort of get hit the polls like they do Perry's so you think. I don't. I could I don't dismiss that look at the tech but I bet this mister trump when I'm off the hepatitis ethnic Christmas when Clinton winning by double digits so here's a scary thing ever on here. Well that's what we saw the ball plus model says. Basically the outcomes are symmetrical with some caveats about but the distribution is right but that means that if Clinton's ahead by. By seven points. It would have to be debt by seven points to win roughly speaking. Any is just to do and at 20%. Our model says. Eras owners. A tossup but seems really weird or for use the same election over and over again. It's that would put winning by seven or eight points the trends are toward Democrats in Arizona we have to be open minded about different outcomes including crippling. We percent is a real chance by the same token. The map might expand and Clinton wins a can be pretty big win Electoral College points.
This transcript has been automatically generated and may not be 100% accurate.