As we enter the final week of the election, everyone wants to know how big the House GOP wave will be. No one has a precise answer yet, but if past performance is any guide, the gold-standard psephologists when it comes to political fortune-telling are Charlie Cook and Stu Rothenberg — and yesterday, both of Cook and Rothenberg showed a massive shift to the GOP. If we use the accuracy of their 2006 and 2008 predications as an indicator, we will be able get a good idea of exactly how large the wave will be from their final House-race ratings, to be released on Monday.

Both men use similar categories: Some races are “Likely” (Cook) or “Favored” (Rothenberg) to go for one party or another; slightly tougher races to call are labeled as “Leaning” toward a party; those hardest to call are rated as toss-ups, which Rothenberg breaks down into three categories (pure, tilting Democratic, and tilting Republican).

For 2006 and 2008, both men have 100 percent records in their “Likely” or “Favored” categories for the winning party (in both years, the Democrats). Their records in the races they rate as “Leaning” toward the winning party are nearly as good: combined, only two seats in two years. Thus, any Democrat whose opponent is said to be “likely” or “favored” to win come Monday will almost surely go down to defeat.

What about the toss-ups? They break only slightly toward the winning party. In 2008, Cook rated 35 races as toss-ups; Democrats won 19 of them, or 54 percent. In 2006, he rated 39 races as toss-ups; Democrats won 22 of them, or 56 percent.

Rothenberg’s ratings are more complicated, but obtain nearly similar results. In 2008, he said 14 races were pure toss-ups; Democrats won seven of them. Democrats won nine of the 13 races labeled as tilting Democratic, and Republicans won all of toss-ups tilting their way. All together, Democrats won exactly half of the 32 toss-ups. In 2006, Democrats won ten of the 19 pure toss-ups, all of the races that were tilting Democratic, and one of the ten tilting Republican. Combine all the toss-ups and Democrats won 21 of 40, or 52.5 percent.

While the “wave” party picks up most of its seats from these categories, it can also win seats categorized as likely or leaning in the losing party’s direction. In 2008, both men thought VA-5 (Perriello) would remain Republican, but it flipped to the Democrats. In 2006, the bigger wave year, Rothenberg missed two of the eleven seats he said favored or leaned Republican (NY-19 and TX-23). In Cook’s case, Democrats picked up four of the 25 seats he rated as likely or leaning GOP.

Furthermore, Republicans can pretty much write off any seat not on one of those men’s ratings. Only once in the last two elections (2006) was one of the two men (Rothenberg) totally blindsided by a Democratic pick-up (David Loebsack’s defeat of Jim Leach).

So, what do all these numbers tell us about the coming House election? Based on their most recent House ratings, yesterday’s, Republicans should gain about 57 seats. But those ratings, even though they moved the needle in the GOP’s favor from just a week ago, could once again shift in the Republicans’ favor by Monday.

While most of their ratings changes occurred yesterday, both men have a pattern of shifting races toward the “waving” party even in the final days. In 2008, even after moving 16 GOP seats in the Democrats’ favor between October 22 and October 29, Rothenberg moved another eight GOP seats toward the Democrats between the October 29 and November 2. In a year as fluid and historic as this one, I wouldn’t be surprised to see even more last-minute shifts than in 2008.

The mutual-fund industry never tires of telling us that “past performance does not guarantee future results.” So too with election predictions. But if the past is any guide at all, Cook and Rothenberg’s data will tell us within two or three seats the magnitude of the GOP’s victory a day before a single vote is counted.

— Henry Olsen is vice president of the American Enterprise Institute and director of their National Research Initiative.