French Presidential Election SMI has lowest Forecasting Error
Emmanuel Macron beat Marine Le Pen with 66.1% versus 33.9% of the popular vote, respectively. Social Media Influence (SMI) was the most accurate known forecasting method with a forecasting error of 2.7% in comparison to an average forecasting error of 4.1% for polls (see table). SMI was closer to the actual result than any known published poll.
Table 1: 2017 French Presidential Election, Comparison between Final Results, Social Media Influence (SMI) Forecast, and Polls from the week before the second round of voting
| Macron | Le Pen | Forecasting Error | |
| Final Result |
66.1% |
33.9% |
|
| SMI |
63.4% |
36.6% |
2.7% |
| Average of Polls |
62.0% |
38.0% |
4.1% |
|
Ipsos |
63.0% | 37.0% | 3.1% |
|
Harris |
62.0% | 38.0% |
4.1% |
|
Ifop-Fiducial |
63.0% | 37.0% |
3.1% |
|
Odoxa |
62.0% | 38.0% |
4.1% |
|
Elabe |
62.0% | 38.0% |
4.1% |
|
OpinionWay |
62.0% | 38.0% |
4.1% |
| BVA | 60.0% | 40.0% |
6.1% |
Source: ZettaCap, Ipsos, Harris, Ifop, Odoxa, Elabe, OpinionWay, BVA, Wikipedia
Not only was SMI the most accurate forecasting method, but it also appears to have been the first to forecast a victory for Macron, having made that forecast in early December, when most polls, betting markets and analysis had him in a distant third / fourth place.
SMI similarly outperformed other forecasting methods in the 2016 US Presidential Election when it predicted that Trump would win the election with 306 electoral votes, the only known method to correctly forecast the electoral college.