Does late registering lower your variance?

It helps your hourly rate for lots of reasons, but are the swings of coming in at a shallow stack worth it?

I have enjoyed writing about variance reduction strategies in poker, which so far has included dropping down in stakes vs playing fewer tables, payout structure, as well as increasing volume and reducing rake. Alongside Dara O'Kearney, we have also previously looked at the variance differences between large and small-field MTTs.  

Today I wanted to see if I could use the variance calculator at PrimeDope to see if we could map out the impact of late registration on poker variance. We have written a lot about late registration before and broadly speaking you get an instant ICM boost on your buy-in when you do it (except in PKOs). Often you can get a 10 or 20% ICM boost by late regging and because you spend less time in each tournament this also means your hourly improves. 

However, you come in at a shallower stack depth meaning you have less of an edge. It can also be mentally quite taxing, because you often have to stack off soon after you register. You can certainly have periods where you seem to just register, go all-in, and bust out without cashing, again and again. 

To stimulate this, we have a lot of assumptions to make, so I asked Dara O'Kearney for some pointers. We took a $10 MTT grinder with a 20% ROI in 100-runner tournaments. We said they would play 1,000 tournaments in a year and they registered at the start, or near enough. This is a typical sample of tournaments for them:

In their best sample, they won almost $6,000, in the worst-case scenario they lost just over $1,000. On average they made $2,051 over 1,000 tournaments. They would lose money 3.5% of the time and their required bankroll was $1,349. 

To simulate late registering first of all we had to account for the reduced edge of playing shallower. We made it 14% instead of 20%. Then we agreed that late registering would typically provide a 20% ICM boost. A 20% ICM boost on a 14% ROI gives us a 16.8% ROI. Finally, we get to add more volume. Dara has previously estimated that max late regging has meant he can play twice as many tournaments, but to be conservative let's assume our grinder can only play 50% more MTTs. 

So we are comparing a $10 MTT grinder with a 20% ROI playing 1,000 MTTs a year, to a $10 MTT grinder with a 16.8% ROI, playing 1,500 MTTs a year. 

This is a typical sample of results:

Our best-case scenario is much better, we win almost $8,000. Our worst-case scenario is much worse, we lose just under $2,000. On average we win $2,252. Our chance of loss is 2.9% which is actually lower than in the first example. Our required bankroll is $1,394, which is slightly higher than in the first example. 

Having done several of these articles, I was actually expecting max late regging to be more of a slam dunk way to reduce variance. However, as somebody who max late regs every non-PKO I play, I am not surprised that is not quite the case. The upside of late registration is the ROI and volume boost, the downside is that you do have some rough periods where you can't even mincash for days on end.  

Ultimately the increased volume and ICM boost will reduce your variance over a big enough sample, but prepare for some nasty outlier periods because of that reduction in edge. 

Has late regging helped your game? Let us know in the comments: