All three parties claimed to have won the Michoacán election, with the surprise winner Fausto Vallejo Figueroa on top in the PREP, or Programa de Resultados Electorales Preliminares. As its name implies, these are not the official results and have no legal value - they are based on a quick count of the actas or tally sheets, which it appears were scanned electronically just as they came in.
For the official count, the complete electoral packets must be sent in to the electoral districts of the state, and the numbers of the tally sheets checked. They may also be opened and actually recounted, if the data on the tally sheet indicate, well, that something just ain't right - that the numbers don't add up with votes cast and reported remaining ballots, etc.
What is interesting is the ballot boxes reported in the PREP - access, in Excel form, to all the boxes is given here. Of the 6075 boxes reported for governor, the more one skims down the page (they are ordered according to when first reported in to the PREP), there are many hundreds of instances of 1) Tally sheet ineligible, 2) Sum of votes superior to the number of people voting, 3) Numbers do not match up, 4) Envelope empty, and 5) Envelope not handed over.
See for yourself here. Skimming the sheet, at least 500 boxes were not counted as part of the PREP. Though we don't know the absolute size of this vote, in terms of percentage of ballot boxes, that is quite a bit (8 percent based on purely eyeball measure) - and if we assume the number of votes in each box is roughly on average with the ones that were counted, (2-300 it seems, again just skimming), we are talking 100,000-150,000 votes not counted.
They could, of course, follow a complete normal distribution, in that they are spread out evenly among the candidates, but I will certainly withhold judgment on who won in Michoacán Sunday until both the state electoral institute and electoral tribunal have ratified the election.