With the conclusion of the Georgia runoff – and one Senator unexpectedly leaving the Democratic Party – control of the Senate in the 118th Congress is set. Democrats (and those who caucus with them) will hold 51 seats, one more than they held after the 2020 elections, surpassing many of the conventional expectations going into Election Day.

With the race for control of the Senate settled, it’s time to do some housekeeping and look back at our Senate ratings. This was our third cycle (and fourth election, as we also do presidential ratings) doing Senate ratings and we feel it’s always important to analyze how we did, how we compared to other outlets’ forecasts, and where we can improve. For anyone joining us for the first time this election cycle, we hope this turns you on to our way of analyzing the national race from the bottom up, with a candidate and story-driven focus in building out the national puzzle.

The Topline

This year, we mixed up our usual method of publishing the entire map with all ratings at the same time and instead opted to publish the ratings we had ready earlier, rolling out each Senate race’s rating state by state. We published our first initial rating on August 8 and our final initial rating for this cycle on September 22 – about seven weeks before Election Day. At the time we published our last initial rating, we forecast Republicans winning 21 of the 35 seats, resulting in a tied Senate of 50-50. Wary of national trends impacting any races, we held off a long time before changing any ratings. Ultimately, we ended up changing the rating for only two states: Pennsylvania and Wisconsin, the former from Lean Democrat to Toss-Up and the latter from Lean Republican to Likely Republican. Even moving these two states did not change the national forecast, however, and Republicans were still forecast to win only 21 seats on our map.

Going into Election Day, both FiveThirtyEight and The Economist’s separate models had Republicans picking up a seat, and both had Republicans as favorites to win the Senate. In the time between our initial rating of the entire race was out at the end of September and Election Day, our model did not budge from showing the Senate at 50-50 but both of their models responded to what appeared to be a shift in the polls, moving from Democrats picking up a seat at the beginning of October, to Democrats staying even, to Democrats losing a seat. By virtue of their models’ perceiving a national shift that didn’t materialize, our model beat them both and continued to forecast Democrats holding the Senate.

Why did our ratings hold steady? Well, for one, they didn’t – like I said, we did shift two states with competitive races towards Republicans in that same period and this did change the projection slightly (Republicans gained 0.35 seats from this, but not enough to round up to another seat). The way we calculate the seat total does not budge dramatically if one or two races slide one way or the other; it’s designed to be incredibly conservative in how it reacts to a shift, and this has historically served us well, as we have always either overperformed FiveThirtyEight and The Economist or (in our first ratings, in 2018’s Senate race) tied with them. Needless to say, we’re happy with how our own forecast held up this cycle – we may have missed the Democratic gain by a seat, but we did better than two incredibly rigorous data-driven models (and betting markets, oof).

The Swing States

We had 11 races rated less than “Safe” for either party. Of those we gave a Lean or Likely to a party (eight total), we correctly favored the winning party in all eight. Our closest of these eight was Wisconsin, which we moved to Likely Republican in the week before Election Day, but incumbent Republican Senator Ron Johnson only won by 1% statewide.

Where we were perhaps too cautious was in our three Toss-Up states: Georgia, Nevada, and Pennsylvania. All three of these ended up going to the Democratic candidate. Pennsylvania was not particularly close, especially for such a large state. Nevada was close, with the election coming down to under one percentage point, the closest Senate race this cycle (if only you had someone telling you this would be the most important race of the cycle for most of the past year… oh and as far back as July of last year too). And Georgia went to a runoff – which was somewhat predictable – and ultimately reelected the Democratic incumbent by about 3%.

With seven “Likely” states, the odds were in favor of one going the opposite way than what seemed likely. But not this cycle! And honestly that will weigh heavily on how we look at a few of these states going forward. Colorado’s monstrous swing towards Democrats was truly stunning, as incumbent Democratic Senator Michael Bennet won reelection by almost 15%, and it’s hard to make a compelling case that it’s not a Safe Democratic state going forward. But, as Colorado giveth, Florida taketh away. Incumbent Republican Marco Rubio soared to reelection by 16% in the Sunshine State – strengthening the case for a Safe Republican firewall in Florida. And the deeply flawed Republican Senate candidate in Ohio, J.D. Vance, still managed to win by over 6% of the vote in the Buckeye State. Both Florida and Ohio are large states where Democrats ran strong candidates, so it’s hard to see Democrats succeeding in major statewide races there in the near future.

North Carolina was as consistent as ever, handing Republicans a 3% win over a strong Democratic nominee. After the last three presidential elections where the Tar Heel state narrowly went for the Republican by even slimmer margins, and a couple of similarly narrow Senate races where the GOP perennially seemed to edge out the Democrat by a couple percentage points, it’s hard to see Democrats optimistic about their future in North Carolina. However, it’s a pretty tempting target for them, and not one they can afford to concede as parts of the Midwest move away from them and Georgia moves towards them. Bet on watching North Carolina for many cycles to come.

Looking Towards 2024

As we gear up for a big cycle in 2024 (another presidential year!), there are some things that we feel worked particularly well this cycle that we hope to continue or improve on for 2024. First of all, rolling out our ratings earlier and in pieces as opposed to all at once in the week before Election Day was a helpful way to adjust our expectations of the race and keep an eye on how the national story is building out. 

Going forward we’d also love to build out some more interesting analytics and a more sophisticated model that shows how we use both data and subjective metrics, starting earlier in the cycle, and being transparent on how our ratings have changed and performed over time. Not that putting the story of each race up on a map doesn’t get the job done, but wouldn’t you like to know more about why our ratings work and what goes into them?

This also lets us capitalize on the niche way in which we paint our picture of the national race: by telling the state-level and personality-driven stories that matter, as opposed to taking a national view. What a lot of purely-data driven models miss is in the nuances of characters and stories – but it’s impossible to understand the race for the Senate without them. It’s why we were incredibly bullish on Democrats in red Montana and deep red West Virginia in our 2018 ratings, why we highly favored Joe Biden in Michigan and Minnesota in the 2020 election, and why we gave Republicans an upper-hand in seemingly-tight North Carolina and Wisconsin this cycle – while keeping Nevada (which has gone for Democrats in the last four presidential elections) a Toss-Up. Candidate quality and the specific characters and demographics of each of these states varied wildly, and that matters a lot more than a model that looks down at each state from a national lens can show. That’s where we at The Postrider aim to set ourselves apart in our political coverage. So far, that’s proved a good bet.