The Punjab Assembly elections have become a nightmare for pollsters, largely courtesy the rise of the Aam Aadmi Party (AAP).
Two recent opinion polls - ABP-CSDS and India Today-Axis - came up with completely opposite predictions.
While the India Today Axis Survey gave AAP a 29% vote share, the ABP-CSDS survey gave it 21%, a large difference of 8%. The differential in the SAD-BJP vote share in the two surveys is even greater.
Be it the emergence of the Telugu Desam Party in the 1983 Assembly election in Andhra Pradesh or the AAP in Delhi exactly 30 years later, predicting a debutant party's performance has always been difficult for political observers.
Data from previous elections becomes redundant and calculations based on swing "away from" and "towards" parties, get thrown out of the window.
Often the rise of a new party is the product of a broader shift in politics that also brings about a collapse in an existing party, as was the case in the 2013 and 2015 Delhi elections.
For Punjab after AAP's advent, the only meaningful starting point would be the 2014 Lok Sabha elections. The 2014 results would give us some indication as to how well the AAP can or cannot do.
These are the number of segments each of the parties led in.
Lok Insaf Party: 2
Lok Insaf Party led by the Bains brothers of Ludhiana has now aligned with AAP. This will give it three additional seats, two where LIP was leading and one more in which the combined vote share of the two parties exceeded that of the Congress.
Now, even if AAP-LIP alliance retains all the 36 seats it led in, it will still need 23 more seats to win Punjab.
AAP stood second in 10 segments and a respectable third in 22 others. In all, AAP's vote share was above 20% in 68 seats. AAP will need to win at least 80% of these seats to win a majority in Punjab.
The problem for the party arises in the remaining 49 seats, where its vote share was below 20%. In 22 seats, it was less than even 10%.
Perhaps the only thing the two surveys agreed upon was the Congress vote share, with both putting it in the 30-35% range. This is similar to the party's performance in the 2014 Lok Sabha elections, in which it won 33% votes. This seems to be the only certainty in the present election - that the Congress is not going to fall below the 30-35% range. It could well go above that.
Given that the Congress seems to be holding its ground, only a complete collapse in the SAD-BJP vote share can bring about an AAP wave.
AAP's popularity in the 2014 elections was concentrated heavily in the Malwa region - Punjab's southern heartland that has traditionally dominated politics in the state. 34 out of the 36 segments AAP and LIP led in, were in this region, which has 69 Assembly seats in all. AAP swept districts like Sangrur, Faridkot and Fatehgarh Sahib and it can expect to maintain, if not expand its hold.
In alliance with the Bains brothers', it is likely to put up a strong performance in the 14 seats of Ludhiana district as well. It might not be able to repeat its performance in Patiala district, where the Congress has staged a revival under Captain Amarinder Singh.
Significantly, AAP won 13 out of 18 reserved constituencies in the Malwa region.
In the Majha region - the North Western border districts of Gurdaspur, Pathankot, Amritsar and Tarn Taran - the party performed poorly getting just around 12% of the votes and coming third in all the 25 constituencies. It's vote share exceeded 20% only in four seats.
Even in a Lok Sabha seat like Khadoor Sahib which cuts across two regions, AAP did better in the Doaba region areas (Kapurthala district) as opposed to the segments in the Tarn Taran district (Majha region).
Therefore, AAP can do well in Majha only if the SAD-BJP gets completely decimated in what has otherwise been its stronghold.
Dissensions have made matters worse for AAP in Majha. It expelled its former convenor Sucha Singh Chhotepur, who is contesting from Gurdaspur. Even the party unit in Tarn Taran faced over 100 resignations in September last year.
It's present convenor Gurpreet Singh Warraich is also from the same region and is contesting from Batala, a seat where AAP would fancy its chances.
The real battleground for AAP is the Doaba region - which includes the districts of Kapurthala, Jalandhar, Hoshiarpur and Nawanshahr.
This region is likely to witness a direct fight between AAP and Congress. In the 2014 elections, AAP led in two segments in this region, both in Nawanshahr district (now known as Shaheed Bhagat Singh Nagar). The party came second in 4 segments and a respectable third (20-30% vote share) in as many as 14 segments. These 18 seats put together are the real zone of opportunity for AAP.
Dalits form close to 45% of the population in this region and they aren't likely to vote as a monolith. In 2014, Congress led in 5 out the 8 reserved seats in this region, AAP led in 1. The party is pulling out all stops to woo the various Dalit communities here. AAP would need major shift in Dalit votes to turn the tables on the Congress in this region.
The question is, has AAP done enough in the past two and a half years to expand its vote share to such a significant extent? Mind you, AAP has performed this feat in Delhi.
The Delhi experience
The pattern of AAP's expansion in Delhi provides an interesting precedent for Punjab. In the 2013 elections in which AAP sprung a surprise by winning 29 seats, it's victory was concentrated in the core areas of the national capital.
It performed poorly in two zones - the semi-rural and rural areas of Outer Delhi, where the BJP did well and the 6 Muslim dominated constituencies, where the Congress retained its hold. By the 2014 Lok Sabha polls, AAP successfully overthrew the Congress in all the Muslim pockets. It went on an expansion drive across the city and by 2015, it swept Outer Delhi, even seats which BJP had won by huge margins in 2013.
For instance, in Kirari in North West Delhi, the AAP's vote share increased by a whopping 44% between 2013 and 2015.
Has AAP managed to expand as effectively in Punjab between 2014 and 2017? The crowds in its rallies even outside Malwa do indicate that the party organisation has definitely strengthened. But is it enough to double its vote share in the Majha region, or in Malwa districts like Bathinda and Fazilka where it had performed poorly in 2014?
The Congress is unlikely to fall below 30% in its vote share, much like the BJP in Delhi between 2013 and 2015. So AAP's victory would almost entirely depend on how badly SAD-BJP does. There is a palpable anger towards the ruling coalition, but it might benefit the Congress as well, not just AAP, especially in areas where AAP is organisationally weak.
But then knowing AAP, the election won't be about arithmetic but hawa. It's just a question of how strongly the wind is blowing.