Heads-up tourneys and variance

[TNixon]

After doing a bit of digging for information about the variance in heads-up sit-n-gos vs normal 9 or 10 player sit-n-gos, I found an article (through a link on the forums here) titled Basic Bankroll Management, which gives basic guidelines for bankroll requirements for various types of play.

The relevant quote about heads-up-tourneys is:


[ QUOTE ]
Heads up sit and go tournaments are a variance monster all to themselves. Your win rate will have a massive effect on your variance in heads up play, and a player who is only winning 55% of his matches will have huge swings, while a really solid heads up player with a 70% win rate can get away with using numbers about twice the size of the regular SNG numbers in the chart above. The 55% player probably can't have a big enough bankroll no matter what he does; the variance is just too high.

[/ QUOTE ]

I'm having a little bit of trouble understanding why this would be true.

I can certainly see why it would be true that variance would increase heads-up for cash compared to a full ring game, with the reasoning being nearly identical to the reasoning for variance being higher at pot-limit omaha compared to no-limit holdem: because hands are often so close statistically, there are more marginal situations where it is correct to make big bets or calls, while being a slight favorite or underdog.

Similarly, when short-handed, it's less likely that your opponents have premium hands, so you end up putting in more money in marginal situations.

However, when playing a freezeout tournament, isn't your win/cash rate basically the *only* thing that determines variance?

The reasoning given for needing more buyins for a multitable tournament is that with more people, you will reach significant cash less often, so the reccomendation is for a higher bankroll to ride out the dry spells. The "average" player (whatever that really means) has an 11% chance to win a 9 person sit-n-go, but only a 1% chance to win a 100 person tourney. It seems quite obvious that there will be bigger gaps between bigger wins in the MTT situation.

Of course, the fact that you don't have to win to cash in a STT surely makes a difference, with the top 30% getting 2x their buy-in, the top 20% getting 3x, and the top 10% getting 5x, but in a heads-up tournament, the average player has a 50% chance to win. Shouldn't that fact alone *reduce* the variance of a heads-up sit-n-go in comparison to a full single-table tournament, making the dry spells that much shorter?

I have no clue whatsoever know how to calculate it, but I have a hard time imagining that a 20 buyin downswing is more likely to happen on a 50% average over other situations with smaller percentage chances to win bigger pots.

Before this weekend, I would also have said with certainty that my personal experience backed up my gut instinct about heads-up tourneys having less severe downswings, especially at low buyins, because I was having far smaller downswings playing $2-$10 heads-up than I was playing $5 STTs, after working $20 up to almost $200 on heads-up tourneys, playing with about a tenth of my bankroll or less in a single tourney. (if I had $120, for example, I'd play 1 $10 and a couple $5s).

Of course, we're talking about a super-small sample size, and I dropped back down to $20 again on a very long string of losses, but I did clearly identify some very bad tendencies on my part, especially in the $10 tourneys, so at this point, I really have no clue if the downswing was the result of a lot of poor play, combined with a little bit of bad luck, or if that's the normal expected variance playing heads-up?

My gut says that once I stop being stupid in the places I'm being stupid, heads-up tourneys are actually the safest way to slowly increase a bankroll.

Now, I realize completely that even if I'm right about that, $20 isn't really enough to start with (I actually started with $100, but dropped almost instantly to $20 on what was mainly a bad streak of luck on the $5 STTs), and my chances of going broke before getting a stable bankroll are pretty good anyway, but I wanted to get other people's thoughts on the matter.

If it makes a difference, I appear to be significantly better than a 55% winner at the $2 HU tourneys on full-tilt (closer to 70%), and before this weekends big losing streak, which, as I mentioned, appeared to be more me being a donkey and less normal variance, looked to be about a 60%-ish winner on the 5s and 10s.

[Kharlog]

I've read that HUSNGs have lower variance than STTs somewhere in this forum. I started this month and have played 93 HUSNGs winning about 64% of them. That's extremely small sample size and I've been winning almost every day so far (played every day). The losing days haven't been too bad as well. Could be just a good run though...

[ChicagoRy]

Losing 16 buyins at the 5 dollar level or more buyins at the 2 dollar level (unless you continued playing 10s during your downswing) is probably more of a result of a few leaks in your play and not just variance.

To be sure you should post some big pots that you lost (say, pots over 1k) and the corresponding action previously before some of these big pots (where needed) and let the forum determine if you have some tendencies that might need fixing.

[TNixon]

[ QUOTE ]
Losing 16 buyins at the 5 dollar level or more buyins at the 2 dollar level (unless you continued playing 10s during your downswing) is probably more of a result of a few leaks in your play and not just variance.

[/ QUOTE ]
Oh, I know for a fact there was a big poor play factor in the big downswing. There was a little bit of bad luck, too, but mostly just bad play. I clearly have some tendencies that need fixing, but they're not particularly difficult to identify. I just didn't bother until it was way too late. Most of them were big all-in calls where I had no business whatsoever calling. And I did continue playing the 10s much longer than I should have before stepping down as well. [img]/images/graemlins/smile.gif[/img]

I'm just trying to resolve my (admittedly limited) experience and gut instinct with an article claiming HUSNGs have a much higher variance than normal ones.

The question I *really* want answered is what's the cheapest, lowest-variance way to build up a bankroll, outside of grinding away at limit cash tables, which I simply don't have the patience for. So far, (minus a day's worth of complete stupidity), heads-up tourneys seem the best of all the things I've tried, but the one article I was able to find on the issue claims otherwise, and a forum search here is actually what pointed me to that article in the first place.

[img]/images/graemlins/smile.gif[/img]

[ChicagoRy]

Link please.

[omgwtfnoway]

[ QUOTE ]
However, when playing a freezeout tournament, isn't your win/cash rate basically the *only* thing that determines variance?]
this is absolutely correct

consider this analogy for thinking about hu sngs and 6handed or 9handed sngs:
for hu sngs there are two outcomes, you win or lose. flipping a coin is a decent but not great analogy (i'm assuming that you have a decent edge)
for sngs or mtts with more players roll d&d style die with 50+ sides.
the point is that you're simply going to need a lot more trials with the die before you even get all the possible outcomes much less all the outcomes in their expected proportions. hence, there is far less variance in husngs than in other sngs and tournies.

chicagory - if i remember correctly that's from an article written by pokerfox for the pocketfives forum. if you look up their article archive and search for his first articles you'll find it pretty easily. i think it was the first article he wrote for p5.

[ChicagoRy]

Yea, I definitely think that article is misguided. He says that for a 70% winrate you need 90 buyins. Even more (I think like 120) if you play professionally. Lol at that?

Oh and that's for HUSNGs.

[TNixon]

Yeah, it was that article by fox for pocketfives. That was pretty much the only thing useful I turned up searching for terms including heads up, tournament, and variance.

Thanks much for the verification. Claiming the need for that many buyins for a 70% winrate, in what seems like the tournament mode with the least variance of all sounded *very* off to me as well.

I just wanted to make sure there weren't any big gaping holes in my logic.

Anybody know how to run the odds of going broke calculations for a certain number of buyins vs winrate? I'm pretty sure I'm playing fast and loose with just 10-15 buyins, but I'm curious to know just how bad I'm really being.

[img]/images/graemlins/smile.gif[/img]

[ChicagoRy]

I'd recommend at least 25 buyins for fast loose play, maybe closer to 35-40 for turbos, unless you are a huge winner.

I move up at 20-25, move down if I fall below 30 for previous level or feel my play is not what it should be, so I basically take 5-8 buyin downswing-move down shots.

10-15 is a little too small IMO, your risk of ruin is too great.

[jay_shark]

Tnixon , I had a post a while back and I gave out all the necessary calculations of a players RoR given his win rate and the number of buy ins he has . You may want to search for it but if I find it I'll let you know .