Should Virat Kohli have enforced the follow-on in the Galle Test?

While we don't know the game's result yet, not enforcing the follow-on is a decision that is taken to reduce the opposition's chances of making a comeback in the match and often shuts them out of the game. In recent years, numbers suggest that not enforcing the follow-on has worked better for teams.

Really? I thought batting once again minimises the chances of winning a Test?

Not at all. Since the famous Kolkata Test in 2001, just 19 of the 141 matches in which teams were eligible to enforce the follow-on have ended in a draw. Of these 19 draws, only five came in matches when the side leading chose to bat again. On the other hand, 14 of the 80 matches when follow-on was enforced ended in a draw. Teams that have not enforced the follow-on have won close to 92% of the matches, which is nearly 10 percentage points more than when they have.

So, is this another "VVS effect" explainer piece?

Far from it. The last question explains why.

Has there been a shift in the follow-on strategy from teams since the 2001 Kolkata Test?

Yes, in the 20-year period preceding that Test (1981-2001), there were 77 occasions when a team had the opportunity of enforcing the follow-on. They sent the opposition in to bat again on 70 of those occasions. That's almost 91%.

This percentage has come down to 56.70 after the Kolkata Test. With captains looking to rest their bowlers and scoring rates going up, and leaving enough time for the opposition to be bowled out, teams are opting to bat again. In other words, the limitations of time, which used to dictate tactics in Test cricket in the past, apply to a lesser extent in recent years.

India led Sri Lanka by 309 runs, after being on the field for just 78.3 overs. Was that margin not safe enough to enforce the follow-on?

It definitely used to be, before VVS Laxman and Rahul Dravid did their thing. Since then, 300 seems to be the mark when captains are more likely to enforce follow-on. Of the 83 times when teams had a first-innings lead under 300 runs, they enforced the follow-on in 34 occasions (40.96%).

This percentage almost doubles when teams have a lead of over 300 runs. Of the 58 instances when teams led by 300 or more runs, they ended up enforcing the follow-on in 46 Tests. Interestingly, Australia - who were at the receiving end in the Kolkata Test - are the only team to have not enforced the follow-on in spite of having a lead of over 400 runs since then. They had a lead of 445 runs over England in the first Ashes Test at the Gabba in 2006-07, but chose to bat again.

Follow-ons enforced in Tests, 11-Mar-1981 to present
Period Follow-on applicable Follow-on enforced %age
 11-Mar-81 to 11-Mar-01  77  70  90.90
 Since 12-Mar-01  141  80  56.73

So, are you saying captains were less cautious before the 2001 Test?

Numbers say so. Earlier, captains enforced follow-ons despite their team's lead being less than 300. In the same 20-year period before the Kolkata Test, teams enforced follow-ons 50 out of 57 times when they led by a margin of less than 300 runs. That's 87.72%, compared to just 40.96% now.

In fact, there is no significant change in percentages even if the lead was fewer than 250. Before Kolkata, teams enforced follow-on in 31 out of 36 matches when they led by less than 250 runs. Post that, teams have asked opposition to bat again only 19 times off the 54 instances when they've had the opportunity.

Does the workload of bowlers have any bearing on whether the follow-on is enforced or not even when the lead is big?

Yes. Of the 58 instances when teams have led by over 300 runs since 2001, they have enforced follow-on in 19 of the 21 instances when their bowlers have had to bowl 60 or fewer overs. When the bowlers have bowled more than 60 overs to get the opposition out, this percentage has gone down. Of course, the 60-over number is indicative at best. Also, whether teams choose to bat again or not also depends on factors such as time left in the match and wear and tear of the pitch among other factors.

Bowler workloads determining follow-on enforcement
Overs bowled Matches Follow-on enforced Percentage
 up to 60 overs  21  19  90.48
 More than 60 overs  37  27  72.97

Correlation does not imply causation. Does this pre-and-post-Kolkata theory pass that test?

No, of course we are not saying that one Test match is the sole reason for a paradigm shift. Steve Waugh, who was the man who enforced that follow-on, continued to do so every further time he got the opportunity, and his side won each time. Kolkata 2001 is a black-swan event - sides following on have won only three times in Test cricket history - and has been among the many factors influencing captains' conservative approach since.

There are a number of other reasons for this drastic shift. For starters, the volume of cricket being played has increased dramatically in the period since, and captains have grown ever more conscious about managing their bowlers' workloads. Schedules have grown ever more congested, with back-to-back Test matches becoming the norm worldwide.

Finally, batting out time, which used to be a common feature in Tests, happens far less these days, as the five draws out of the 61 unused follow-on opportunities attests to.

Shiva Jayaraman is a senior stats analyst at ESPNcricinfo; Srinath Sripath is a sub-editor at ESPNcricinfo