This is for the President. parties can do better in Senate and Congressional races catering to their specific state issues on a local micro scale. This does not translate well to Presidential so hanging your hat on 2018 local races to imply 2020 success is no different than HRC resting her laurels on Trump faux pas and supposed leading polls to ignore campaigning in battleground states and hitting on issues voters cared about.
This is sorta true when this type of analysis is done by the unskilled, the press, political hacks etc to feed their hopes and bias. Skilled pollsters though, have usually identified specific districts that do in fact serve as a bellweather leading indicator for the the state or country. Not a lot of skill around these days in the polling community. Mostly hacks, people who hate or love a candidate, or a bunch of internet crap that asks who you love or hate and then reach all sorts of conclusions from that.
Polls are meaningless for predicting 2020. Even skilled pollsters cannot take a few Senate races for local candidates and extrapolate to Statewide voting patterns with DIFFERENT candidates and issues involved. If they do then they are simply proving how weak the science of statistics they are basing their conclusions on is.
One starts down the wrong road when even using the word "prediction." There is, however, often a lot of valid data that shows different aspects of a candidates performance, which serve as building blocks for their own thinking and strategizing. Lots of people seize upon that valid data and abuse it by running too far with it. Not to forget that I am the guy who sees countless polls showing Biden with an alleged overwhelming lead. I see those data points but do not consider them to be definitive of dispositive. I do not dismiss that data, but neither do I abuse it by driving off a cliff with it as some do.
The best polls is approval ratings,and they have been 100% in predicting presidential re elections since they were started.
Biden again leads the field in new polls from Texas +5, Michigan +6, and Nevada +8. Texas is interesting because O’Rourke was in that poll and polled at 16%. Michigan is interesting because Warren polls in 3rd place there.