Friday, December 2, 2016

My apology

I've talked in person to multiple people who followed my blog. They tell me that my numbers helped them stay calm.

I wasn't trying to keep people calm. I was trying to get it right.

I failed massively, and for that I am deeply sorry. I have always distrusted the pundits, and after 2000 I wanted to do something on my own to get numbers I could trust. I was trained as a mathematician, but I'd never done statistics. Once I learned stats and learned how fast and loose the field is with methods, I decided to make my own system.

My system told me the electoral college was close in 2004 and it was. Bush needed to take both Ohio and Florida to win. Bush won the popular vote by 2.5%, Florida by 5% and Ohio by 2.1%, so it wasn't that close in terms of the real vote. Obviously, it was nowhere near as close as 2000, but it wasn't a landslide by any means.

The Wall Street Journal found my website in 2004, which was hard to do back in the day as it was a hidden page on my now-defunct website about Pascal's Triangle. They asked me for a prediction by the Sunday night before the election. The polls were close and my system said Kerry was the favorite because Bush's chances to sweep Ohio and Florida were under 50%. I made my prediction, but polls on Monday turned the tide slightly and Bush was a favorite. Still, my Sunday numbers were printed and I screwed up.

Obama's wins weren't close, and my system predicted the states and the district contests in Maine and Nebraska very well. It also did very well on the Senate races in 2012. These positive results made me confident, but I kept looking at where things might go wrong. I had no idea things could go as wrong as they did.

In previous years, I would miss one or two races. In the 56 contests for the electoral college in 2016, I got six wrong and all of the upsets favored Trump. In order from least surprising to most, I missed Maine's District 2, Florida, North Carolina, Pennsylvania, Michigan and Wisconsin. Of those six losses, only Michigan and Wisconsin are truly slender leads, and Pennsylvania is somewhat close.

I honestly don't expect recounts to change anything, much as I don't think Gov. McCrory's hope for a recount in North Carolina is going to go anywhere.

Let me ask myself some questions in public.

Did I have any doubts? Yes, but I didn't have the numbers that made me think the doubts were worth mentioning. Let's look at the three states now looking at recounts.

Wisconsin: 31 polls since May, Trump never lead in even the most optimistic Republican pollster's numbers. The median of the past few weeks said a 7 point lead for Clinton.

Pennsylvania: 54 polls since May, and only Google Consumer Surveys, a new polling company that often disagrees with the consensus, gave Trump a lead. The median said a 4 point lead for Clinton.

Michigan: 28 polls and only one poll even gave Trump a 50-50 chance, another new company named Strategic National. The median was a 2 or 3 point lead.

In simple terms, I had no "what if" I could ask that made these now crucial states look close.

How did the polls screw up so bad? That is a question I can't answer with certainty, but one of the known problems with polling is getting too many "refused to answer". Beyond the fact that Clinton and Trump combined might count for 90% or less in a poll when they averaged about 95% in the actual contest, there are the people who refuse to pick up the phone because they have caller ID, or the people who answer but hang up when they hear the noise of a call center in the background, or even someone like me who might stay on the line but be rejected because I truthfully answer the question about being a public employee. I have no way of knowing how the "refused to answer" crowd will vote or if they will vote at all. This might account for the difference, but I can't say that with any mathematical certainty, as I said before.

Any other factors look important? This was a very strange election and a lot of the strangeness is a matter of public record, not some conspiracy based nonsense. My problem is I have no way to quantify any of them. The oddest things to my mind were Comey's decision to try to sway the vote, the one-sided nature of the WikiLeaks information and the attempts by Russia to assert influence.  Another strong pro-GOP influence were the efforts to discourage voters in many states, whether it was ID laws that were difficult to comply with or allocating minimal polling places in areas with strong Democratic tendencies.

While I know for a fact all these things happened, I don't trust any of the numbers I can get to determine how important these factors were.

And so now we have President-elect Trump. I have no idea how bad things will get, but I have little hope these four years will work out for the best. When it comes to domestic policy, we will have to see how much in lock step the Republicans are on their worst ideas, which for me would have to include the privatization of Medicare. When it comes to foreign policy, my greatest fear is we have a gang who thinks a shooting war with Iran is a good idea. Those are just two of my policy concerns off the top of my head. When it comes to tone of our politics, it will now get worse than it was under Obama, when it was already not good.

If my numbers said this was a serious possibility, I would have made that as clear as I could. But they didn't say that and so I passed out incorrect information, believing it to be reliable.

For that I am truly sorry.

Saturday, November 19, 2016

A critique of Allan Lichtman's system.

Allan Lichtman got a lot of press for his system which predicted Trump's win.

Here's a news flash; it didn't predict it at the end.

Here is a link to his 13 true-false questions. 

His threshold is simple enough. Five false answers or less means the incumbent wins. Six false answers or more means the White House will switch parties. As you can see on the link, there were six false answers. So game to Trump and Prof. Lichtman wins again.

Hold on a second.

My first critique: Too many questions are subjective. Fort example, he says neither Trump or Clinton is charismatic (questions 12 and 13). His questions are set up so that Clinton's lack gets a false answer and Trump's lack gets a true answer. (This makes sense in his system, since false answers are bad for the incumbent side true are true answers are bad for the challenger.) While I am immune to both their charms, Trump clearly had very loyal followers who did find him charismatic, in large part because he a TV star. I also personally know people very committed to Clinton. Charisma, like beauty, is in the eye of the beholder.

Question 9 gives the Obama administration a point for being unmarred by a major scandal. Again, while I agree that should count, partisans who have been bombarded by Benghazi stories for years on end think of the administration as completely tainted. Like charisma, this is probably better ranked on a sliding scale.

Questions 10 and 11 give Obama a split, no major foreign success or failure. Here I would disagree, as I think the Iran deal should be counted as a success, and of course, conservatives count Benghazi as a major failure.

Lichtman gives the Obama administration a 0 for effecting major policy. Unless he means in the past four years, Obamacare counts as major, a BFD in the language of our beloved vice president.

Maybe Lichtman doesn't count first term stuff. If so, that would be a yet another serious flaw in his system. Trump and the GOP are still running against it and the Democrats were running to protect it.

My second and more specific critique: Question 4 is stated "There is no significant third party or independent campaign". Lichtman's answer is false, which counts as a black mark against Clinton, in his system the fatal black mark.

This one is clearly quantified. At or above 5% for a third party of the vote counts as significant, under 5% is insignificant.

Here is the current count.

The Libertarians have 3.28%. This should count as true, no significant third party. By Lichtman's count, this is a true answer and there are only five false ones.

Lichtman's system said Clinton would win and no one is calling him on it. As far as I can tell, you heard it here first.

With every state now decided, the Libertarians only got 

Monday, November 7, 2016

Final predictions,
7 November 2016

Click the map to create your own at

Here is the map I am expecting for Clinton vs. Trump. Iowa and Nevada were both 50-50 toss-ups this morning, but new polls have split the difference, with Iowa leaning Trump and Nevada leading Clinton. Here's the numbers I am hanging my hat on.

Electoral college count
Clinton 323
Trump 215

This does not take into account two electors from the Clinton delegation in Washington saying they won't vote for her. Who knows, they might get kicked out of the club, we'll wait and see.  Here is the graph of my weekly snapshots. The odds right now are about 99% to 1% in Clinton's favor.

   I've read accounts that if Trump can't win, he'd like to out-perform Romney's 206. He's got a good chance to do that, but he isn't facing Obama, who is a much more popular figure in his eighth year than either Clinton or Trump.

The Senate prediction

This prediction is split into good news and bad news. The order depends on your political persuasion.

The GOP currently holds 54 seats to the Democratic/Independent coalitions' 46.  My algorithm considers the Democrats will hold all ten seats they are defending, including the close race in Nevada for the seat Harry Reid is vacating. The algorithm predicts three losses for the GOP in Illinois, Wisconsin and Pennsylvania. There are four very close races still in the mix, but the forecast from my algorithm says the GOP will hold New Hampshire, North Carolina, Missouri and Indiana. This will mean the GOP will have 51 seats and retain control of the Senate. I do not have the resources to predict the House, but this doesn't look like the kind of landslide that can make that chamber switch sides.

My competitive side would love to see my algorithm go 90-0 in predictions, but if I miss any, I hope some of my errors give control of the Senate to the Democrats. It's hard to pay attention to anything but Trump this year, but I'll take this moment to remind people that the Republican Party has very few places where they part company with Trump on policy and they have decided they would rather take hostages than enact legislation through the democratic process. Hillary is not going to have an easy gig, but a Democratic Senate would make it a little easier.

Go out and vote and if you are on the left side of the political spectrum, vote down ticket (!) and hope for the best in the Senate. It will take a evil miracle for Trump to come back from his current position.

Sunday, November 6, 2016

Clinton vs Trump.
6 November 2016,
2 days before the election

I usually wait on Sundays for the morning talk shows to be over, since it's common that at least one will roll out some new polls. I'm breaking with that rule today because my next update is the final post late Monday night. This is the dress rehearsal.

I could never have started this hobby without the Internet making it so easy to find polling data. I always wanted information I could rely on, especially because it's been a cliche for generations that there will be reports on the last weekend that state "the race is tightening", even in the big blowouts like Reagan-Mondale and Bush-Dukakis.

This weekend, it appears to be true. I'll save the in-depth stuff for Monday night and just give the snapshot info in this post.

Here is the list of states and districts, from most likely to go Trump to most likely to go Clinton.  RED indicates a state moving from one category to another in Trump's direction, while BLUE means the state moved in Clinton's direction.
Solid Trump (more than 95% Confidence of Victory [CoV]): 
Total: 145

Leaning Trump (between 50% and 95% CoV for Trump): SC NE-2 AZ GA UT OH AK
Total: 64

Toss-up: NV IA
Total: 12
Leaning Clinton (between 50% and 95% for Clinton): 
Total: 99

Solid Clinton (more than 95% CoV for Clinton): 
Total: 215

The Current Count


The only red moves that really matter are Ohio back in the Trump column and Nevada and Iowa now rated toss-ups.

We are at a time in the contest when valuable information about early voting is available. My algorithm does not take it into account. The late polls say Nevada got closer and moved in Trump's favor. The lines for early voting in Nevada at a predominantly Hispanic polling place and the numbers from Clark County in general look to favor Clinton and the Democrats. This is interesting and useful data, but I don't factor it in. If the late polls today and tomorrow still rate Nevada and Iowa toss-ups, I will squeeze my algorithm a little harder.

Here are the odds my algorithm produces for victory for each candidate if the election were being held today. In a very real sense, it is being held today, given the number of states with early voting, available in all but ten states, and the increasing popularity of absentee ballots, available everywhere, though some states make people fill out a reason.

Trump's chances are no longer parts per million, but his paths to victory are extremely narrow and time is running out fast. If instead of using my algorithm I used my gut feeling about the states he should not lose and let my algorithm look at the states he needs to steal from Clinton, his odds climb up to about 4.0%.

Even giving Trump advantages I would deny to Clinton, it's not very close.

My last update will be Monday night after 9:00 pm Pacific time, which is midnight on the East Coast. At that time, I'll make my final predictions in the 34 Senate races, The 50 states, the District of Columbia, and the congressional districts in Maine and Nebraska that make those states proportional.

Hold on. The end really is in sight.

Saturday, November 5, 2016

The Senate race
5 November 2016

My schedule for the weekend is Senate today and the Presidential races on Sunday, but I'll also have a final prediction on Monday night. My feeling is that there are only three races where changing a seat from one party to another looks like a lock now, and there are five races that are not completely decided. Here's the chances for how many senators will be in the Republican caucus this January.

Three races where the Republican seat is likely to switch.

Confidence of Victory for Democrat: over 99.99%
Polls in the last three weeks: 3
Polls showing a lead for the Democrat: 3

Even before Mark Kirk went Full Metal Asshole in the debate (first mistake) and then apologized (for a 21st Century Republican, his second mistake), the polls showed him way behind. My system counts this as a lock. Eventually, my system will get one of these wrong, but it hasn't happened yet. Tammy Duckworth will be in the Senate in January if 1+1 = 2 still is true this Tuesday.

Confidence of Victory for Democrat: 85%
Polls in the last three weeks: 9
Polls showing a lead for the Democrat: 9

The median poll says this shouldn't be a blowout, but it would be a real shock if this is wrong. Russ Feingold should be back.

Confidence of Victory for Democrat: 75%
Polls in the last week: 8
Polls showing a lead for the Democrat: 7

Pennsylvania has no early voting, so I change the freshness dating to one week. Knowing the universal popularity of absentee ballots, which are allowed in every state though some ask for an excuse ("I'm going to be on one of the moons of Saturn that week"), this is a part of the algorithm I might change. Even if I did, I do limit my data set to the most recent ten polls, and in that set the Democrat leads in eight, with one tie for the GOP and one lead. Katie McGinty should prevail over Pat Toomey in this all-Irish American donnybrook. For me, this is kind of a sweet reminder of old school East Coast politics.

Five races that are still not completely determined.

Confidence of Victory for Democrat: 50%
Polls in the last three weeks: 5
Polls showing a lead for the Democrat:2
Polls showing a tie: 1
Polls showing a lead for the Republican:2

Any new poll in the next 60 hours or so could change my current view of the race. Right now, with a tie in three weeks, I'd look at the latest three polls and give the nod to the Republicans. One new poll that doesn't show a tie between now an Monday night will tip the scale, which now favors the GOP.

New Hampshire
Confidence of Victory for Democrat: 50.73%
Polls in the last week: 8
Polls showing a lead for the Democrat:4
Polls showing a lead for the Republican: 4

As the designer of the algorithm in which I put my faith, this race shakes that faith to the core. Maggie Hassan's lead here is one due to her two point lead in one poll looks a little better than Kelly Ayotte's two point lead in another. Any new poll will turn the algorithm around. I'm only use eight polls, all from the last seven days because New Hampshire has no early voting. If I used more, it would be advantage Ayotte.

I really don't know what will happen here and right now, this is the race that decides the control of the Senate.

No pressure.


Confidence of Victory for the Republican: 59%
Polls in the last three weeks: 8
Polls showing a lead for the Democrat: 2
Polls showing a tie: 2
Polls showing a lead for the Republican: 4

It will take several polls to get my algorithm to favor upstart Jason Kander over incumbent Roy Blount. I consider this a hold for the GOP unless new data goes crazy in favor of the challenger in the next few days.

North Carolina  

Confidence of Victory for the Republican: 59%
Polls in the last three weeks:11
Polls showing a lead for the Democrat:3
Polls showing a tie: 2
Polls showing a lead for the Republican: 6

My best guess is a held seat for the Republicans. Only a pro-Democratic flurry between now and Monday night will sway me from this opinion.


Confidence of Victory for the Democrat: 66%
Polls in the last three weeks: 10
Polls showing a lead for the Democrat: 6
Polls showing a tie: 1
Polls showing a lead for the Republican:3

This is the only seat where the Democrats are playing defense in a serious way, hoping Cathy Cortez Masto can defend Harry Reid's vacant seat. The lines for early voting in Latino precincts are a data point worth noting, but they aren't in my algorithm. I'm assuming this prediction will not change between now and Monday night.

The dress rehearsal for the final prediction.

Democrats hold 10 of 10 seats.
Republicans lose 4 of 24 seats.
Senate numbers in January: 50 Dem and allies, 50 Republican

I don't do any work towards the prediction of the House, but my guess is the closeness of the Senate and the race for the White House is likely not enough for the Democrats to re-take the House from the GOP. Unless the Democrats can convince their base that elections happen every two years instead of every four, this will mean obstructionism will likely continue in Washington for the foreseeable future.

In some ways I hope I'm wrong, and I put no confidence in my guess about the House, but that's what the numbers say right now.

Tomorrow: the dress rehearsal of the final prediction for Clinton-Trump. (Spoiler alert: Clinton.)

Sunday, October 30, 2016

Clinton vs Trump.
30 October 2016,
9 days before the election

This post is going to be MUCH longer than usual.  If you just want the numbers, you can skip down to the graphs. If you want to see me take out The Big Ugly Stick on Nate Silver, Bill Gates, Survey Monkey and James Comey, read the whole essay.

Let's take a look at the states where the odds are really close, where the favorite is has a 60% chance or less of winning.

Alaska: 3 EV: Trump 60%, Clinton 40%
Alaska a battleground? I am as amazed to type this as you are to read it. Five polls make up this very surprising result. Two polls, Survey Monkey and C-VOTER International have this race as a sane person would expect, a complete whomping of the Democrat by the Republican, any Republican. But two polls, Google Consumer Survey and Lake Research, say Trump has a slender one point lead and both candidate are under 40%. And then there's Craciun Research, which I pronounce "Crazy One Research", that gives Clinton a comfortable four point lead 47% to 43%. This means the median shows a one point lead, which usually converts to about 60%-40% in Confidence of Victory.

I can only hope a more established polling company will take a crack at this race next week. Suffolk polled Alaska in June and gave Trump a nine point lead. I hope they get hired to go back, whatever the result.

Utah: 6 EV: Trump 52%, McMullin 30%, Clinton 18%
There was some talk early on that Utah might become a battleground because the Mormons disapprove of Trump's lifestyle and he, of course, hearing himself insulted, had to insult the state and the religion right back. Until polls in mid September, there was no sign of any trouble, but Trump, still in first place, was polling at 40% or less. In early October, polls started showing where the missing votes were going, supporting Evan McMullin, a Utah born former CIA operative who started running in August and is on the ballot in only 11 states.

For me, it was kinda fun to break out the algorithm for a really close three way race and the idea of a "favorite son" candidate is pleasantly quaint. That said, there have only been five polls since October 18, the first time McMullin looked competitive, and Trump leads in four of five.  Because of some messy math, I have to use averages instead of medians in my calculations and it is very unlikely the polls will make McMullin the favorite in the next nine days.

Ohio: 18 EV, Clinton 50%, Trump 50%
Ohio is Trump's proudest achievement is the race so far, and it is still a coin toss. In the eleven most recent polls, four of them - done by Emerson, Quinnipiac, Ipsos and Suffolk - have the race at dead even. Four give the advantage to Clinton and three say Trump is ahead.

My system may well rate this a toss-up a week from today. Stay tuned.

Arizona: 11 EV, Clinton 51%, Trump 49%
Polling companies are taking the idea that Arizona is a battleground state very seriously. Ten polls by ten different companies in the past three weeks and Clinton leads in five of them and Trump leads in five. (It's a little odd in a race this close to have no flat-footed ties.) This means the race is this close because it's the average of a Trump victory with a Clinton victory. I can't remember the last state where this was the case.

As older polls leave due to my freshness dating rules, good polls for Trump could swing this around.

Iowa: 6 EV, Clinton 59%, Trump 41%
Clinton's advantage is based on only six polls taken in the last three weeks, a sparse amount of polling for a swing state. She leads in four, one calls the race a tie and Trump is only given a lead by Survey Monkey, a very busy but erratic polling company that favors Trump almost exclusively in the true swing states.

Maine District 2: 1 EV, Clinton 59%, Trump 41%
There's sparse and then there's drought conditions. There have only been three polls in the past three weeks released dealing with Maine's two districts. District 1 is heavily Democratic, but District 2 is closer. This morning, University of New Hampshire released a poll they finished on October 25. One poll says it's tie, the median says it's a one point lead and the last gives Clinton a comfortable advantage.

I would love to see more polling here and in Nebraska's three districts as well, only the one including Omaha has any chance of being competitive.

I have about a week to fish my wish.

Nevada: 6 EV, Clinton 59%, Trump 41%
Nevada is being polled like crazy. Ten polls by ten different companies in the past ten days make up the list I'm working with. Clinton leads in six, three make it a flat-footed tie and Trump leads in one, yet again from Survey Monkey, the most Trump-friendly pollster in the battleground states.

Having mentioned all of these, even if Trump sweeps these races, he has a long way to go. Let's ignore the 2nd District's one electoral vote, which doesn't really help Trump win.  His chance to sweep the vital contests about 1.3%. He also desperately needs Florida and North Carolina, and multiplying those odds together, he gets way below a percent at 0.0487%, better stated as 487 parts per million.

And then he needs to win one more state where his chances are less than 10%, like Wisconsin, Pennsylvania, New Hampshire or Michigan.

There are other paths to 269 electoral votes or more. I assume the Republicans are going to keep the House and in a contested race, they will elect the guy their voters really want, not pull some 1824 style shenanigans. But even adding all those up, his chances to win an election held today round to about one hundredth of a percent.

And make no mistake, the election IS being held as you are reading this. Nate Silver's numbers right now put Trump at 21.4%. I assumed his numbers would start to agree with others as the election got closer and his margin of error shrunk, but no. He is peddling horse dung and the public still eat it up. I haven't seen such an undeserved reputation for quality since Microsoft produced its truly awful 1980s software that turned a dozen or so nerds into billionaires.

I programmed for a living in the 1980s. Don't get me started on Bill Gates.

And then there's James Comey. I was on Twitter when the news broke Friday, and I got that numb "oh, crap" feeling I got when Hillary fainted on September 11. Then the news clarified. Comey didn't have a warrant to look at these e-mails, they were likely from Anthony Weiner on his former wife's computer. They were a complete waste of time and Comey comes out looking worse than anybody else in this horrible episode, having more egg on his face than even Weiner or the eager beavers in the House.

I tend to stay with the numbers and report on what they say about the race, but the polling that will come out may not give a clear view of the impact. My best instincts say the impact is minimal. People supporting Clinton will find the news outlets that correctly call this a nothing burger and people supporting Trump already think Clinton should be executed by a combination of hanging and shooting only after being given a slow acting, painful poison, so this will not change their view.

This won't change a lot of minds. Trump doesn't need a bombshell at this point, he needs an asteroid hitting the planet. This is not the asteroid he was looking for.

And now the numbers and the pretty pictures, prettier if you are a Clinton supporter.

Here is the list of states and districts, from most likely to go Trump to most likely to go Clinton.  RED indicates a state moving from one category to another in Trump's direction, while BLUE means the state moved in Clinton's direction.
Solid Trump (more than 95% Confidence of Victory [CoV]): 
Total: 145

Leaning Trump (between 50% and 95% CoV for Trump): MO GA AK UT
Total: 35

Toss-up: OH
Total: 18
Leaning Clinton (between 50% and 95% for Clinton): 
Total: 188

Solid Clinton (more than 95% CoV for Clinton): 
Total: 222

The Current Count


Utah moved from Toss-up to leaning Trump and given that it's a three way race, having more than 50% means a lot. States where Trump is a prohibitive lead got a little stronger for him, notable Texas and Indiana, and a few states where Clinton was a 95% favorite have changed to her being about a 90% favorite. Eventually, some state where one candidate is a 90% favorite is going to switch sides, but it hasn't happened yet in roughly 200 contests I followed in 2004, 2008 and 2012 general elections. (Primary races are a completely different kettle of fish.) So far, the states I've gotten wrong have been where the favorite was no better than 70%. That's why I listed the seven races at the top of the post. With just over a week to go, that's where the action is. If Trump wins all of those, he still isn't president.

And now the overall odds if the election were held today and the fifteen races my algorithm thinks are worth watching.

Here are the 15 states considered this week's by the algorithm.

Trump's non-battleground count of electors: 128

The battleground states ranked by pqn:
(Clinton %, Trump %, electors)
1. FL: 76%, 24%, 29
. OH: 50%, 50%, 18

3. AZ: 51%, 49%, 11
4. GA: 21%, 79%, 16
5. NC: 85%, 15%, 15
6: UT: 48%, 52%, 6 (48% is Clinton + McMullin)
7. IA: 59%, 41%, 6
8. NV: 59%, 41%, 6
9, PA: 93%, 7%, 20
10. MI: 95%, 5%, 16 
11. WI: 92%, 8%, 10
12. AK: 40%, 60%, 3
13. MO: 7%, 93%, 10 
14. TX: 1%, 99%, 38

15. SC: 5%, 95%, 9

Current probability of victory if the election were held today:
Clinton 99.87%
Trump    0.13%

I hate celebrating early, but there isn't time for Trump to turn this mess around. Breaking 200 electoral votes will be a moral victory for him, but 269? Like I said earlier, he needs an asteroid to hit, not a little skipping stone like James Comey's career suicide last Friday.

Back on Saturday with the last weekend, starting with the still close battle for control of the Senate. Frickin' finally.

Saturday, October 29, 2016

The Senate races
29 October 2016

This is the next to last weekend before the election. Polling companies are paying close attention to most of the close races and the general trend is favoring the Democrats. Here is the graph of probabilities for how many seats the Republicans will hold in the Senate that convenes in January.

The most likely outcome is still a 50-50 split, with control of the Senate going to party that controls the White House. It is extremely likely that will mean Tim Kaine will be breaking any ties, as will be made apparent in tomorrow's post about the presidential race. But for the first time since any serious polling has been done, the second most likely outcome is for the Republicans to be in the actual minority.

Here's the lay of the land. There are 34 Senate contests, and 29 of them aren't close. In those races, it looks like the GOP are going to lose 3 of their 54 seats, bringing it down to 51-49 with five seats still closely contested.

The Republicans are on defense in four of the five closely contested races. If they win in Nevada, they could gain a seat. In every other close race, they can either stay at the current level or lose a seat. That means the optimal split for the GOP is 52-48 and the optimal for the Democrats is 47-53, barring major upsets. Here's the news on the five close races.

Nevada: Advantage Democrats, 55% to 45%
This race was advantage Republicans until this month. Now it is being polled by a lot of companies and of the last ten polls, five favor the Democrat Catherine Cortez Masto, four show a lead for Republican Joe Heck and one says it's a tie. It's about as small a lead as my system can create, but the momentum is in her favor.

North Carolina: Advantage Republicans, 66% to 34%
Of the "close" races, this is the least close. My system uses the median poll, and the ten polls currently being counted have seven in favor of incumbent Richard Burr. It would take a serious flurry of polls favoring his opponent Deborah Ross in the next week for this to change sides.

Missouri: Advantage Republicans. 55% to 45%
This race is unusual among the close five in that not a lot of companies have decided to do polling in Missouri. Trump's lead looks solid, but Roy Blount is in a contest, if we can trust the four polls taken this month. This is the only close race in a non-battlefield state. Go figure.

New Hampshire: Toss-up, 50% to 50%
Kelly Ayotte and Maggie Hassan are in a serious tussle. The last ten polls have four favoring Hassan, four giving Hassan the lead and two ties. Whoever polls better in the last week will get the advantage in my system.

Pennsylvania: Toss-up, 50% to 50%
Control of the Senate may very well come down to our all-Irish donnybrook in The Keystone State. In the last ten polls, Democrat Kate McGinty has the lead in four, there are four ties, and incumbent GOP Senator Pat Toomey leads in two. Again, the last week of polls will make a big difference, but the wind is at McGinty's back.  

Back tomorrow with more news about the long national nightmare that is Clinton vs Trump.